Public Comments on Proposed Federal Rules
Here's a look at public comments on proposed Federal Register rules
More Recent Headlines
Featured Stories
Box Elder Urges Clarification of Census Residency Rules to Protect Military Community Counts
Carter Struck
WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site.
420 Villa Drive, Box Elder SD 57719 605.923.1404 ph. 605.923.4264 fax www.boxeldersd.gov The City of Box Elder is an Equal Opportunity Provider October 14, 2026
U.S. Census Bureau Designation: ADDEMO/FRN 4600 Silver Hill Road Washington, DC 20746
RE: Comments - Proposed 2030 Census Residence Criteria (91 FR 57524)
To the U.S. Census Bureau:
On behalf of the City of Box Elder, we appreciate the opportunity to comment on ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. 420 Villa Drive, Box Elder SD 57719 605.923.1404 ph. 605.923.4264 fax www.boxeldersd.gov The City of Box Elder is an Equal Opportunity Provider October 14, 2026 U.S. Census Bureau Designation: ADDEMO/FRN 4600 Silver Hill Road Washington, DC 20746 RE: Comments - Proposed 2030 Census Residence Criteria (91 FR 57524) To the U.S. Census Bureau: On behalf of the City of Box Elder, we appreciate the opportunity to comment onthe proposed residence criteria for the upcoming 2030 Decennial Census.
Box Elder, South Dakota is a thriving community that is proud to host Ellsworth Air Force Base. Our community is one of South Dakota's fastest growing cities with over 14,800 residents, or a nearly 12% increase since the 2020 census. It is estimated that our population will exceed 16,000 by 2029. These numbers do not include any personnel at Ellsworth Air Force Base, which is more than 4,000 airmen/women or 8,000 when you include dependents. This number is expected to balloon to more than 10,000 in the coming years because of the new B-21 mission.
The City of Box Elder has identified the following potential implications of the proposed changes to the 2030 Decennial Census for our military families and our community:
1. Risk of Counting Military Personnel in the Wrong Locations The proposed rule establishes a new definition of "usual residence". Nonpartisan demographers and data policy experts warn that changing the traditional standards could cause military personnel to be counted in places that do not reflect where they actually live. Because service members frequently move, deploy, or split time between home stations and temporary training environments, a rigid new rule could easily mislocate them.
2. Discrepancies in Deployments and Overseas Counting For military personnel serving abroad, the 2030 draft criteria (specifically under proposed 15 CFR Sec. 60.5) explicitly distinguish between federally affiliated personnel and ordinary citizens:
Temporarily Deployed: Service members temporarily deployed overseas but officially stationed or assigned at a domestic base will be counted at their usual U.S. residence (their stateside home station). Permanently Assigned Overseas: Personnel permanently stationed overseas and their dependents will continue to have separate counting provisions channeled primarily for congressional apportionment.
Because these residency rules are being codified into federal law for the first time-rather than remaining flexible internal agency policies-any errors in how transitory military families are tracked will permanently distort local data for a decade.
3. Impact on Local Base Funding and Community Infrastructure The proposed rule restricts the 2030 Census apportionment count strictly to U.S. citizens and lawful permanent residents (green card holders). Non-citizens in other temporary legal categories will be excluded.
For communities surrounding military installations (such as Box Elder), this could result in a net population undercount. Local towns hosting bases heavily rely on accurate census tallies to secure federal and state funding for shared civilian-military infrastructure, including:
Public schools and defense-impacted school aid (Impact Aid). Local highway and transportation grants. Regional healthcare facilities and emergency services.
4. Loss of Demographic Insights on Race and Ethnicity The new regulations propose to completely eliminate all questions regarding race, ethnicity, and sexual orientation from the decennial census questionnaire.
While the Census Bureau may still collect some of this data via separate samplings like the American Community Survey, the main 2030 count will lack it. For military leadership, researchers, and local policymakers, this restriction strips away a primary data tool used to monitor regional healthcare disparities, evaluate housing equity, and track the demographic realities of diverse military families.
Recommendations:
Clarify in Sec. 60.2 that temporary military assignments and deployments do not independently establish a new "usual" residence. The Census Bureau should distinguish temporary military-directed absences from permanent changes of residence while establishing clear procedures for counting overseas deployed personnel at their established stateside residence. Establish (Sec. 60.2, 60.6) military-specific procedures to evaluate and reconcile permanent duty station information, Permanent Change of Station (PCS), and Temporary Duty (TDY) orders, Department of War personnel records, home-of-record, legal residence, tax records, and actual physical residence. Tax domicile or home-of-record information should not independently determine census residence. In Sec. 60.3, explicitly clarify that military dependents are to be counted at their own usual physical residence, independent of the temporary assignment or deployment location of the service member. Conduct a military-specific impact analysis as to whether the proposed methodology materially changes the geographic distribution of military personnel and their dependents compared with currently established census residence criteria.
The City of Box Elder is an essential partner in supporting the military readiness of Ellsworth Air Force Base. The ability to plan for transportation, housing, emergency response, and military family services depends on accurate population data. That said, we believe that military deployments, training, and temporary assignments should not inadvertently result in service members and their dependents being counted outside the communities where they maintain an established residence.
The City of Box Elder respectfully requests that the U.S. Census Bureau clarify the proposed criteria, establish military-specific residency procedures, while evaluating potential impacts on defense communities before implementing changes to the 2030 Decennial Census.
Sincerely,
Hank Prim
City Administrator
Cc: Senator John Thune
Senator Mike Rounds
Congressman Dusty Johnson
South Dakota Governor Larry Rhoden
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-71938
420 Villa Drive, Box Elder SD 57719 605.923.1404 ph. 605.923.4264 fax www.boxeldersd.gov The City of Box Elder is an Equal Opportunity Provider October 14, 2026
U.S. Census Bureau Designation: ADDEMO/FRN 4600 Silver Hill Road Washington, DC 20746
RE: Comments - Proposed 2030 Census Residence Criteria (91 FR 57524)
To the U.S. Census Bureau:
On behalf of the City of Box Elder, we appreciate the opportunity to comment on ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. 420 Villa Drive, Box Elder SD 57719 605.923.1404 ph. 605.923.4264 fax www.boxeldersd.gov The City of Box Elder is an Equal Opportunity Provider October 14, 2026 U.S. Census Bureau Designation: ADDEMO/FRN 4600 Silver Hill Road Washington, DC 20746 RE: Comments - Proposed 2030 Census Residence Criteria (91 FR 57524) To the U.S. Census Bureau: On behalf of the City of Box Elder, we appreciate the opportunity to comment onthe proposed residence criteria for the upcoming 2030 Decennial Census.
Box Elder, South Dakota is a thriving community that is proud to host Ellsworth Air Force Base. Our community is one of South Dakota's fastest growing cities with over 14,800 residents, or a nearly 12% increase since the 2020 census. It is estimated that our population will exceed 16,000 by 2029. These numbers do not include any personnel at Ellsworth Air Force Base, which is more than 4,000 airmen/women or 8,000 when you include dependents. This number is expected to balloon to more than 10,000 in the coming years because of the new B-21 mission.
The City of Box Elder has identified the following potential implications of the proposed changes to the 2030 Decennial Census for our military families and our community:
1. Risk of Counting Military Personnel in the Wrong Locations The proposed rule establishes a new definition of "usual residence". Nonpartisan demographers and data policy experts warn that changing the traditional standards could cause military personnel to be counted in places that do not reflect where they actually live. Because service members frequently move, deploy, or split time between home stations and temporary training environments, a rigid new rule could easily mislocate them.
2. Discrepancies in Deployments and Overseas Counting For military personnel serving abroad, the 2030 draft criteria (specifically under proposed 15 CFR Sec. 60.5) explicitly distinguish between federally affiliated personnel and ordinary citizens:
Temporarily Deployed: Service members temporarily deployed overseas but officially stationed or assigned at a domestic base will be counted at their usual U.S. residence (their stateside home station). Permanently Assigned Overseas: Personnel permanently stationed overseas and their dependents will continue to have separate counting provisions channeled primarily for congressional apportionment.
Because these residency rules are being codified into federal law for the first time-rather than remaining flexible internal agency policies-any errors in how transitory military families are tracked will permanently distort local data for a decade.
3. Impact on Local Base Funding and Community Infrastructure The proposed rule restricts the 2030 Census apportionment count strictly to U.S. citizens and lawful permanent residents (green card holders). Non-citizens in other temporary legal categories will be excluded.
For communities surrounding military installations (such as Box Elder), this could result in a net population undercount. Local towns hosting bases heavily rely on accurate census tallies to secure federal and state funding for shared civilian-military infrastructure, including:
Public schools and defense-impacted school aid (Impact Aid). Local highway and transportation grants. Regional healthcare facilities and emergency services.
4. Loss of Demographic Insights on Race and Ethnicity The new regulations propose to completely eliminate all questions regarding race, ethnicity, and sexual orientation from the decennial census questionnaire.
While the Census Bureau may still collect some of this data via separate samplings like the American Community Survey, the main 2030 count will lack it. For military leadership, researchers, and local policymakers, this restriction strips away a primary data tool used to monitor regional healthcare disparities, evaluate housing equity, and track the demographic realities of diverse military families.
Recommendations:
Clarify in Sec. 60.2 that temporary military assignments and deployments do not independently establish a new "usual" residence. The Census Bureau should distinguish temporary military-directed absences from permanent changes of residence while establishing clear procedures for counting overseas deployed personnel at their established stateside residence. Establish (Sec. 60.2, 60.6) military-specific procedures to evaluate and reconcile permanent duty station information, Permanent Change of Station (PCS), and Temporary Duty (TDY) orders, Department of War personnel records, home-of-record, legal residence, tax records, and actual physical residence. Tax domicile or home-of-record information should not independently determine census residence. In Sec. 60.3, explicitly clarify that military dependents are to be counted at their own usual physical residence, independent of the temporary assignment or deployment location of the service member. Conduct a military-specific impact analysis as to whether the proposed methodology materially changes the geographic distribution of military personnel and their dependents compared with currently established census residence criteria.
The City of Box Elder is an essential partner in supporting the military readiness of Ellsworth Air Force Base. The ability to plan for transportation, housing, emergency response, and military family services depends on accurate population data. That said, we believe that military deployments, training, and temporary assignments should not inadvertently result in service members and their dependents being counted outside the communities where they maintain an established residence.
The City of Box Elder respectfully requests that the U.S. Census Bureau clarify the proposed criteria, establish military-specific residency procedures, while evaluating potential impacts on defense communities before implementing changes to the 2030 Decennial Census.
Sincerely,
Hank Prim
City Administrator
Cc: Senator John Thune
Senator Mike Rounds
Congressman Dusty Johnson
South Dakota Governor Larry Rhoden
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-71938
Black American Cultural Council Ethnic Classification Proposal Seeks Recognition of Descendants of U S Chattel Slavery as a Distinct Lineage-Based Group
Carter Struck
WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site.
The Black American Cultural Council "A Lineage Based Organization Dedicated To Preserving Black American History, Identity & Culture" Date:
TO: Office of Management and Budget
Office of the Chief Statistician of the United States
Executive Office of the President
Washington, D.C.
Dear Director and Office of the Chief Statistician:
On behalf of the Black American Cultural Council (BACC), we respectfully submit this formal ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. The Black American Cultural Council "A Lineage Based Organization Dedicated To Preserving Black American History, Identity & Culture" Date: TO: Office of Management and Budget Office of the Chief Statistician of the United States Executive Office of the President Washington, D.C. Dear Director and Office of the Chief Statistician: On behalf of the Black American Cultural Council (BACC), we respectfully submit this formalrequest for the recognition and distinct ethnic classification of Black Americans descended from United States chattel slavery, including descendants of pre-Emancipation free communities of color, as a historically established ethnic population native in its ethnogenesis, cultural development, and collective identity to the United States of America, situated on the North American continent. We contend that the existing federal classification of "Black or African American," as administered under Office of Management and Budget Statistical Policy Directive No. 15 (SPD 15), insufficiently distinguishes the specific historical, genealogical, cultural, and socioeconomic experiences of Black Americans whose collective identity developed within the territorial boundaries of the United States.
I. HISTORICAL AND LEGAL BASIS FOR ETHNIC RECOGNITION
Black Americans constitute a distinct ethnic population whose collective identity was formed through centuries of hereditary enslavement, the establishment of free communities of color, emancipation, Reconstruction, Black Codes, Jim Crow segregation, racial terrorism, disenfranchisement, and subsequent systems of institutional discrimination. The historical foundation for this delineation is reflected in the Thirteenth, Fourteenth, and Fifteenth Amendments to the United States Constitution; the Civil Rights Act of 1866; the Civil Rights Act of 1875; and the Civil Rights Act of 1964. These legislative and constitutional developments addressed conditions arising from the uniquely American institution of slavery and its enduring consequences.
Accordingly, Black American ethnicity must be understood through its specific historical formation within the United States, rather than exclusively through broad continental ancestry, racial appearance, or generalized classifications encompassing populations with materially different historical origins.
II. NECESSITY OF LINEAGE-BASED ETHNIC DELINEATION
We respectfully assert that the designation Black American, when employed as a distinct ethnic classification, should apply specifically to individuals who can establish ancestral continuity with persons enslaved within the United States, qualifying pre-Emancipation free communities of color, and their descendants. This proposed ethnic designation would not encompass immigrant populations or their descendants solely by virtue of racial identification, citizenship, skin complexion, or residence within the United States, absent the requisite historical lineage. The purpose of this distinction is not to invalidate the identities or civil rights of other populations, but to ensure that a historically identifiable American ethnic group is not statistically conflated with populations whose immigration histories, ancestral experiences, and relationships to United States chattel slavery are fundamentally different. Failure to recognize these distinctions risks obscuring disparities, compromising historical accuracy, and weakening the precision of federal research concerning intergenerational disadvantage, discriminatory practices, and potential lineage-specific remedies.
III. REQUEST FOR FEDERAL ADMINISTRATIVE ACTION
Pursuant to OMB's authority to establish federal statistical standards under SPD 15, we respectfully request that the Office:
1. Establish "Black American--Descendant of U.S. Chattel Slavery" as a distinct, lineage-specific ethnic classification, separate from immigrant-origin ethnic populations within the broader Black racial category, to distinguish Black Americans as an ethnic group with ZERO immigrant history or background.
2. Recognize the United States as the geographic place of ethnic formation for this population, distinguishing ethnic origin from remote ancestral geography.
3. Develop standardized lineage-identification criteria incorporating historical census records, genealogical documentation, pre-Emancipation free communities of color,synonymous nomenclature (Negro, Colored, Mulatto, Black) and other reliable evidence of ancestral continuity.
4. Implement disaggregated federal statistical reporting to preserve the distinction between descendants of U.S. chattel slavery and populations of immigrant origin, while maintaining existing civil rights protections and statistical comparability.
5. Implement a policy of protection for Black American designation to limit anyone who this designation does not apply to, liable for falsifying identity.
IV. CONCLUSION
The recognition of Black Americans as a distinct, historically constituted American ethnic group is a matter of historical accuracy, administrative precision, and institutional accountability. A classification that acknowledges race without adequately recognizing ethnicity, historical lineage, and the specific legacy of United States chattel slavery cannot fully represent the population whose collective experiences gave rise to some of this nation's most consequential constitutional and civil rights reforms. We therefore respectfully request formal consideration of this proposed classification, an administrative review of its feasibility under SPD 15, and a written response identifying the appropriate procedural avenues for further consideration. Black American history is American history, and Black American ethnic identity warrants recognition within the nation in which it was formed.
Respectfully submitted,
David Walker
10/31/2026
davidsharpewalker@gmail.com
Sign Our Petition:
https://c.org/xPgQKQRbpH
"Preserving Heritage. Protecting Identity. Upholding Historical Accountability"
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-69489
The Black American Cultural Council "A Lineage Based Organization Dedicated To Preserving Black American History, Identity & Culture" Date:
TO: Office of Management and Budget
Office of the Chief Statistician of the United States
Executive Office of the President
Washington, D.C.
Dear Director and Office of the Chief Statistician:
On behalf of the Black American Cultural Council (BACC), we respectfully submit this formal ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. The Black American Cultural Council "A Lineage Based Organization Dedicated To Preserving Black American History, Identity & Culture" Date: TO: Office of Management and Budget Office of the Chief Statistician of the United States Executive Office of the President Washington, D.C. Dear Director and Office of the Chief Statistician: On behalf of the Black American Cultural Council (BACC), we respectfully submit this formalrequest for the recognition and distinct ethnic classification of Black Americans descended from United States chattel slavery, including descendants of pre-Emancipation free communities of color, as a historically established ethnic population native in its ethnogenesis, cultural development, and collective identity to the United States of America, situated on the North American continent. We contend that the existing federal classification of "Black or African American," as administered under Office of Management and Budget Statistical Policy Directive No. 15 (SPD 15), insufficiently distinguishes the specific historical, genealogical, cultural, and socioeconomic experiences of Black Americans whose collective identity developed within the territorial boundaries of the United States.
I. HISTORICAL AND LEGAL BASIS FOR ETHNIC RECOGNITION
Black Americans constitute a distinct ethnic population whose collective identity was formed through centuries of hereditary enslavement, the establishment of free communities of color, emancipation, Reconstruction, Black Codes, Jim Crow segregation, racial terrorism, disenfranchisement, and subsequent systems of institutional discrimination. The historical foundation for this delineation is reflected in the Thirteenth, Fourteenth, and Fifteenth Amendments to the United States Constitution; the Civil Rights Act of 1866; the Civil Rights Act of 1875; and the Civil Rights Act of 1964. These legislative and constitutional developments addressed conditions arising from the uniquely American institution of slavery and its enduring consequences.
Accordingly, Black American ethnicity must be understood through its specific historical formation within the United States, rather than exclusively through broad continental ancestry, racial appearance, or generalized classifications encompassing populations with materially different historical origins.
II. NECESSITY OF LINEAGE-BASED ETHNIC DELINEATION
We respectfully assert that the designation Black American, when employed as a distinct ethnic classification, should apply specifically to individuals who can establish ancestral continuity with persons enslaved within the United States, qualifying pre-Emancipation free communities of color, and their descendants. This proposed ethnic designation would not encompass immigrant populations or their descendants solely by virtue of racial identification, citizenship, skin complexion, or residence within the United States, absent the requisite historical lineage. The purpose of this distinction is not to invalidate the identities or civil rights of other populations, but to ensure that a historically identifiable American ethnic group is not statistically conflated with populations whose immigration histories, ancestral experiences, and relationships to United States chattel slavery are fundamentally different. Failure to recognize these distinctions risks obscuring disparities, compromising historical accuracy, and weakening the precision of federal research concerning intergenerational disadvantage, discriminatory practices, and potential lineage-specific remedies.
III. REQUEST FOR FEDERAL ADMINISTRATIVE ACTION
Pursuant to OMB's authority to establish federal statistical standards under SPD 15, we respectfully request that the Office:
1. Establish "Black American--Descendant of U.S. Chattel Slavery" as a distinct, lineage-specific ethnic classification, separate from immigrant-origin ethnic populations within the broader Black racial category, to distinguish Black Americans as an ethnic group with ZERO immigrant history or background.
2. Recognize the United States as the geographic place of ethnic formation for this population, distinguishing ethnic origin from remote ancestral geography.
3. Develop standardized lineage-identification criteria incorporating historical census records, genealogical documentation, pre-Emancipation free communities of color,synonymous nomenclature (Negro, Colored, Mulatto, Black) and other reliable evidence of ancestral continuity.
4. Implement disaggregated federal statistical reporting to preserve the distinction between descendants of U.S. chattel slavery and populations of immigrant origin, while maintaining existing civil rights protections and statistical comparability.
5. Implement a policy of protection for Black American designation to limit anyone who this designation does not apply to, liable for falsifying identity.
IV. CONCLUSION
The recognition of Black Americans as a distinct, historically constituted American ethnic group is a matter of historical accuracy, administrative precision, and institutional accountability. A classification that acknowledges race without adequately recognizing ethnicity, historical lineage, and the specific legacy of United States chattel slavery cannot fully represent the population whose collective experiences gave rise to some of this nation's most consequential constitutional and civil rights reforms. We therefore respectfully request formal consideration of this proposed classification, an administrative review of its feasibility under SPD 15, and a written response identifying the appropriate procedural avenues for further consideration. Black American history is American history, and Black American ethnic identity warrants recognition within the nation in which it was formed.
Respectfully submitted,
David Walker
10/31/2026
davidsharpewalker@gmail.com
Sign Our Petition:
https://c.org/xPgQKQRbpH
"Preserving Heritage. Protecting Identity. Upholding Historical Accountability"
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-69489
Babbage Proposal Submitted to CFTC for Standardized Compute Service Disclosures and Verifiable Pricing
Carter Struck
WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
1
Defining the compute service behind the price
BABBAGE LTD response on compute derivatives
To: Commodity Futures Trading Commission
Re: Request for Comment on the Listing of Compute Derivatives Contracts
RIN 3038-AF77 | FR Doc. 2026-17163 | Revised 18 October 2026
Submitted by BABBAGE LTD | Company number 14360828
Position and recommendations
Babbage supports developing ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY 1 Defining the compute service behind the price BABBAGE LTD response on compute derivatives To: Commodity Futures Trading Commission Re: Request for Comment on the Listing of Compute Derivatives Contracts RIN 3038-AF77 | FR Doc. 2026-17163 | Revised 18 October 2026 Submitted by BABBAGE LTD | Company number 14360828 Position and recommendations Babbage supports developingcompute derivatives and reference data in parallel. A useful reference price must identify what the buyer can purchase, establish credible evidence of its price and relate that price to the exposure being hedged. For infrastructure rentals, the product is an instance or coordinated group of instances, with defined resources, access rights, service obligations and duration. The processor or accelerator model is only one attribute.
These distinctions have practical consequences. A lower quote may buy a smaller resource allocation, different connectivity or weaker service-level commitments for availability, incident response and restoration. A falling average may reflect a change in the mix of services. Even a completed transaction may be unsuitable for a particular index, and an advertised price does not establish that capacity was available to purchase.
Babbage proposes to help procurement teams compare offers, explain price differences and assess renewals consistently. A clearer account of the resources, sustained performance and service-level commitments being purchased would improve supplier selection and negotiation. We therefore recommend the following safeguards for compute reference prices.
Define the purchased service.
Publish a common minimum set of disclosures covering resource rights, the surrounding system, connected scale, sustained performance, operational assurance and commercial terms.
Establish comparability over time.
Publish service-class eligibility criteria and tolerances supported by evidence. Explain how changes in configuration, allocation, service-level commitments and sample composition are handled. Distinguish services only where the difference matters to the intended contract.
Make prices verifiable.
Retain traceable evidence and independently verify material inputs. Publish proportionate information about methodology, coverage, limitations, assurance and controls over investor and affiliate conflicts. Identify public listings, executable offers and completed transactions separately.
This response focuses on infrastructure rentals and addresses Q1(e), Q2(a), Q2(g), Q3(c) and related questions on verification and capacity [1]. The proposed framework applies to CPU-only instances and systems using GPUs or other accelerators. Inference services require their own units and eligibility rules. Our common specification, provider mappings and qualification framework remain proposals.
Interest of the commenter.
Babbage publishes a daily B200 accelerator price index based on provider-published rates and intends to seek its use as a settlement reference. We have a direct commercial interest in this proceeding and are a prospective competitor of other index administrators.
Our series does not currently apply the availability verification proposed here. We recommend the same safeguards for Babbage as for other administrators. We claim neither a settlement-ready index nor audited market-wide coverage, and provide no estimates of transaction coverage or producer concentration.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
2
Q1(e). Define economically comparable services
An instance-hour should specify resources and commercial rights. Hardware can be stored; unused access time expires. Distinguish reserved time, elapsed access and metered usage, and whether the buyer receives a whole device, a partition, shared access or a connected cluster.
Published prices illustrate the problem.
Babbage's archived H100 tier display groups providers, not service quality [2]. On 3 October 2026, AWS listed p5.48xlarge (eight H100s, Linux on-demand, US East/N. Virginia) at $55.04 per instance-hour, or $6.88 per GPU-hour [3]. Lambda listed its eight-H100 SXM instance at $3.99 per GPU-hour [4]. These rates require comparison of host resources, connectivity, availability and incident-response commitments, region and purchase terms. They establish neither executable availability nor arbitrage. Competition, discounts and commercial strategy may also explain the gap.
Specify the allocation.
Virtualization alone does not establish sharing: exclusive devices, hardware partitions and scheduled shared access confer different rights. NVIDIA documents time-slicing in bare-metal and virtualized deployments and warns that requesting more time-sliced resources does not guarantee proportional compute access [5].
Memory capacity alone is insufficient. NVIDIA's B200 Multi-Instance GPU (MIG) profiles 1g.45gb and 2g.45gb have equal nominal memory but one-seventh and two-sevenths of the streaming multiprocessors, respectively [6]. Record the exact profile and any further sharing. Equal fractions of memory, compute resources and scheduled time are not necessarily interchangeable.
Describe the complete system and its connections.
Specify CPU architecture and model, physical-core or vCPU allocation, simultaneous multithreading (SMT) and sharing conditions, host memory capacity and bandwidth, storage, software and network conditions. Where accelerators are present, specify their allocation and connections to the host and one another. Identical GPU counts can span different communication domains. Instances, rack allocations and multi-rack clusters need explicit boundaries and simultaneous availability. Fabric names leave topology, oversubscription and usable bandwidth unresolved.
Measure sustained performance.
Cooling and power policy affect performance under stated conditions. NVIDIA exposes clock-slowdown statistics for thermal and power constraints [7]. Test representative runs after thermal steady state, with relevant contention, repeated allocations and disclosed variation. Published high-performance computing (HPC) measurements show workload-dependent runtime variability within one GPU model; Annex B gives a scoped example [8][9]. Cooling methods or virtualization technology alone should not determine service grades.
Assess reliability, availability and service commitments.
Short tests cannot establish long-term reliability or rare-failure rates. Assess reliability, availability and serviceability (RAS) from operating records. Specify contractual incident-response and restoration obligations, coverage hours, replacement capacity, and service-level agreement (SLA) availability targets, exclusions and remedies. Distinguish response from restoration commitments, and provider declarations from contractual promises and observed outcomes. Service credits may leave lost work and recovery costs uncompensated.
Qualification should specify workload, output-quality and latency requirements, numerical settings and tested software. For agentic workloads, include relevant CPU-based tool execution, orchestration and interactions with inference services. Include communication, queueing and recovery where relevant. Total cost per successfully completed workload supports comparison, but cannot supply a universal conversion between architectures or replace settlement-price evidence.
Group services by economically material differences
Classes should reflect differences affecting the intended exposure. Across architectures, apply the same material resource, software, precision, quality and service requirements for the specified use. Memory or interruption rights may preclude substitution despite similar throughput on one workload. Disclose each conversion or performance adjustment's calibration, scope and uncertainty.
Classes may be too narrow to yield enough independent prices. Disclose this limitation and narrow or defer settlement use, rather than pool incompatible services. Keep classes open to competing implementations and administrators. Annexes A and B propose specification and qualification procedures.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
3
Q2(a). Establish verifiable price evidence
Defining the service must be followed by verifying its price. Each settlement observation needs a service class and supporting evidence: a trade can concern an ineligible service, while an eligible service may have only an advertised price. These recommendations also address Q1(c) and Q2(b).
The LIBOR lesson for price verification
LIBOR shows why publication alone is insufficient. Panel banks assessed borrowing costs rather than simply reporting trades. The CFTC documented submissions influenced by derivatives positions and reputational concerns, although individual submissions and the fixing were public. A credible market connection and controls over contributor incentives remain essential [10].
Competition can discipline prices while leaving index inputs vulnerable to selective reporting, omitted discounts or unavailable offers. These are design risks, not allegations about any provider. Publicly verifiable terms and prices help buyers compare offers, challenge discrepancies and negotiate, while helping entrants demonstrate value. Each observation should identify its evidence category.
Evidence category
What it establishes and what must be checked
Completed transactions
Agreed service, quantity, period and net consideration. Reconcile contracts, invoices, credits and usage; seek counterparty corroboration where appropriate.
Firm executable offers
Accessible terms and supply, subject to quantity, duration and acceptance rights. Verify the service, price, quantity, delivery window and customer access conditions within the observation window; distinguish offers from completed commercial transactions.
Indicative listings or estimates
Context about advertised or assessed prices. Identify assumptions and limits. Unverified listings should not enter an executable-offer settlement reference; any assessed-price fallback requires separately disclosed rules, evidence and safeguards.
Independently verified confidential trades in the intended market may provide stronger evidence than public listings. Mandatory invoice publication could weaken representativeness. A public payment record alone establishes neither compute unit price nor delivered quantity, quality or counterparty independence. Whatever the recording system, link net consideration to the eligible service, quantity, terms and delivery evidence.
Verify advertised availability
Advertised rates should enter an executable-offer settlement reference only with evidence that the specified customer class could buy the service and quantity at that price during the observation window. Evidence may come from a bounded provisioning test conducted independently of the provider being tested, a binding reservation, or an independently verifiable commercial execution.
A successful test verifies only the quantity and conditions tested. It does not establish total supply, justify quantity weights or demonstrate representative coverage. Keep unverified listings available for information. Annex C details testing, weighting and record controls.
Public disclosure should cover service definitions, sources and collection, calculation and fallback rules, coverage, concentration, corrections and assurance findings. Aggregate or delay sensitive records. Public scrutiny and independent confidential verification serve different purposes. Annex B details calculation and governance disclosures.
Independent review requires records from which decisions can be reproduced, checks for related parties and duplicates, and oversight of conflicts. Administrators should hold enforceable audit and supervisory production rights for contributed non-public records and exclude unverifiable material inputs. Annex C provides the detailed controls.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
4
Q2(g). Preserve the meaning of the reference price
Existing methodologies provide a foundation. Ornn describes GPU-specific transaction indices and verification of hardware, software, host resources, connectivity and reliability. Silicon Data describes normalization for hardware, rental terms, geography and platform differences [11][12].
ICE Futures U.S.'s 29 September 2026 notice identifies planned OCPI H100 (HPR) and B200 (BKL) futures based on Ornn's indices. The notice states that listing is pending completion of the CFTC comment period and that no listing date or timeline had been set [13]. On 21 September 2026, the CFTC Division of Market Oversight extended the initial review period for NYMEX Submission Nos. 26-370 and 26-370S, concerning proposed Silicon Data H100 and B200 Rental Index Futures, until the end of 9 November 2026, citing this Request [14].
Babbage proposes complementary, transparent service specifications and workload qualification to establish which offers are comparable for a stated use and how reference prices relate to buyers' costs. Genuine price differences within qualified classes should remain visible. Eligibility alone does not establish equivalent workload outcomes. The proposed specification would require pilot validation.
We have not audited either provider's records or determined whether either index satisfies the criteria for a particular contract. Babbage's series has not been validated against these settlement safeguards.
Assess each candidate's service and market, eligible activity, concentration, verification access, calculations, adjustments, governance and disruption procedures. For indices claiming similar exposures, compare service boundaries, terms, observation windows, sources and weights to explain material divergence. Valid indices need not produce identical prices, but hedgers should understand why they differ.
Worked example of a change in the service mix
Consider two services sold under the same GPU label. Class A is exclusive; Class B is shared and interruptible. In this hypothetical example, both prices stay unchanged while the mix of billed hours shifts. The denominator counts billed service-hours across different services, not equivalent GPU capacity.
Class
Price per billed hour
Both periods
Period 1
Billed hours
Period 2
Billed hours
A: exclusive
$4.00
80
20
B: shared / interruptible
$2.00
20
80
Illustration only. Period 1: (80 x $4 + 20 x $2) / 100 = $3.60. Period 2: (20 x $4 + 80 x $2) / 100 = $2.40. The blended average falls 33.3%; each class price changes by 0%. The blended average correctly describes expenditure per billed hour for each period's mix. It does not show that an unchanged service became cheaper. Keeping the first period's 80/20 weights would leave the basket at $3.60. Fixed weights illustrate the distinction, but do not by themselves establish a suitable settlement index or resolve obsolescence and liquidity problems.
Explain how the index handles change
Each observation therefore needs a versioned service profile. Specify constituent entry and exit, materiality thresholds, requalification, reclassification and adjustments supported by evidence. Distinguish a changed contract from temporary underperformance of unchanged obligations. An outage should not disappear from the record merely because the service is reclassified.
Report eligible activity, freshness and concentration by independent economic group within each class. Test source withdrawal and service-mix changes. Publish time-limited fallback or delay rules, cessation procedures and contractual settlement remedies.
Where adequacy is unproven, undertake targeted data collection and pilot validation, and defer settlement use until adequacy is demonstrated. A universal standard or elimination of all basis risk is unnecessary. Informational series can support procurement with disclosed limits; settlement also requires market depth, governance and resilience. Undocumented discretion or prices inferred from the derivative itself should not fill evidence gaps.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
5
Q3(c). Match the contract to the buyer's exposure
Even a sound reference price can track a different exposure from the buyer's. Proponents should demonstrate how the index relates to the intended service, location, purchase mode, period and quantity. Spot access, reserved capacity rights and spending commitments may follow different price patterns.
Contract terms should make that relationship explicit. They should identify the service entitlement, pricing unit, connected allocation where relevant, geography, time window, included charges and methodology version. They should explain how corrections, source failure, hardware transitions and cessation are handled. A successor GPU or different contractual availability, response or restoration commitments should not silently replace the original exposure.
The buyer also needs to understand what remains uncovered. A price hedge does not provision capacity, restore jobs or reimburse all recovery costs. Switching providers may involve egress charges. Software portability, migration time and data-location restrictions also matter. RAS and SLA disclosures should identify covered resources, restoration obligations, exclusions and remedies.
Disclosures should distinguish costs included in the reference from those outside the hedge. They should also explain exposure to quantity and timing changes, margin calls and differences between procurement prices and the index.
Worked example of an incomplete hedge
Suppose a buyer needs one million instance-hours of a specified service. Its procurement price rises by $1.00 per instance-hour, while the index rises by $0.10 per index unit. A hypothetical linear hedge paying that index change on one million units produces $100,000 against a $1 million cost increase, leaving $900,000 exposed.
This illustration assumes entry at the initial index level and settlement at the ending level. It excludes fees, financing, taxes and margin timing. It is not a measured hedge result.
Under these assumptions, the relationship is:
change in net cost = (purchased quantity x procurement-price change) (hedge quantity x index change)
Similar GPU labels give no assurance that those price changes will track each other. We recommend testing matched changes over relevant procurement periods, including stress and out-of-sample observations. Correlation between trending price levels is insufficient. Actual hedge effectiveness also depends on the derivative's terms, maturity, liquidity and the hedge implemented.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
6
Measure the capacity buyers can access
For Q1(b), Q1(g) and Q2(h), the relevant supply is the entitlement buyers can access in the specified window. Installed accelerator counts are insufficient. Distinguish owners, providers, resellers and independent economic groups. Multiple listings, endpoints, payment addresses or virtual allocations may refer to the same supplier or capacity. Count enforceable rights consistently with supporting resources and scheduling windows.
Connected allocations need compute, memory and networks together at the required time, with power, cooling and contractual access. Committed capacity is freely available only where its rights permit resale or release. We provide no market-wide quantity estimate. Annex B also addresses surveillance of influence and feedback between futures, reservations and eligible availability.
Supply estimates should distinguish announced or reserved resources from capacity that is installed, energized and accepted for service. Each price observation should identify quantity, delivery window and access rights, with evidence appropriate to that observation. An advertisement alone does not establish immediately purchasable supply.
Physical delivery requires lawful rights sufficient for that delivery, including any necessary assignment or resale. Cash settlement does not inherently require transfer of the resources, but its reference prices should represent the defined service and conditions of access.
Power data can corroborate capacity claims if independently supported. Use energized power attributable to the relevant systems, distinguish it from planned connections and allow for other loads. Estimates should use complete-system power under the qualified operating conditions. Convert facility power to IT power only once, using an appropriate overhead assumption. Neither permits nor an engineering estimate establishes available inventory, customer access or delivery readiness.
How Babbage can contribute
Babbage's proposed common specification, linking the service purchased to its economic comparability and price evidence, would require development and pilot validation. We would discuss the fields and qualification principles with competing providers and administrators. Adoption should depend on evidence and usefulness, with room for different technical implementations.
We offer Annexes A-C as starting points for technical working groups convened by the Commission, designated contract markets (DCMs) or administrators. We would be glad to provide staff, on request, with further detail on service specification, workload qualification and verification tests, subject to the confidential-treatment procedures noted below. We would welcome a meeting with the Division of Market Oversight or the CFTC Innovation Task Force.
Before submitting commercially sensitive information, we would contact the staff named in the Request and follow the confidential-treatment procedures in 17 CFR 145.9 [1][15]. The verification framework requires validation before any claim of settlement readiness.
Requested Commission action
We ask the Commission to encourage common minimum service disclosures, qualification supported by evidence, and verifiable prices. Advertised rates used in executable-offer settlement references should be verified as available for purchase. Administrators should hold audit and production rights for contributed non-public records, publish proportionate disclosures, and explain each input category's contribution and the basis for any quantity weights. Listing proponents should demonstrate suitability for the particular contract.
These requirements should support competition and remain proportionate to the proposed use. They should require the evidence needed for credible settlement without prescribing a proprietary methodology, mandating a cooling or virtualization technology, or treating every configuration difference as a separate market.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
7
Annex A. Minimum instance and price evidence schema
This proposed schema supports qualification and reporting. Detailed evidence should be retained for authorized verification, with service definitions, eligibility criteria and appropriate summaries made public. Full publication of underlying records is unnecessary. A field should define a separate class only when it matters to the intended exposure.
Field group
Information to retain
Identity and version
Provider and economic group; physical origin; SKU and version; region; observation and contract identifiers; effective dates.
Resource entitlement
Whole device or exact partition; exclusive or shared access; minimum scheduling share, interval and burst rules; isolation and interruption rights.
Complete system
CPU architecture and model; physical-core or vCPU allocation; SMT and sharing conditions; relevant NUMA placement; host memory capacity and bandwidth; accelerator type/count, usable memory and connections where present; storage; software and network entitlements.
Connected scale
Instance, server, rack and cluster boundaries; communication domains; topology; bandwidth and contention; minimum allocation available simultaneously.
Sustained capability
Power limits; relevant thermal conditions; qualification date; workload, numerical settings and quality/latency requirements; tested software and, where applicable, model versions; test duration and scale; contention; throughput, latency and variation.
RAS and SLA commitments
Fault, interruption and availability records; contractual incident-response and restoration obligations; coverage hours; replacement capacity and time to resume useful work; SLA availability targets, exclusions and remedies.
Commercial terms
Purchase mode; quantity; term; cancellation, transfer and commitment rights; included/excluded charges, including egress; software portability, migration time and data-location restrictions.
Delivery window and readiness
Offered delivery window; readiness status and supporting evidence as of the observation date; distinguish currently operational capacity from capacity committed for future delivery.
Price and evidence
Currency, billing unit and net consideration; discounts and credits; commercial transaction, availability-verified offer, panel assessment or indicative status; source and verification date; test outcome, verified increment and separately supported available quantity where applicable.
Verification and changes
Supporting record references; configuration and policy versions; material firmware, scheduling or power-limit changes; reassessment decision and effective date; reviewer and assurance scope; related-party checks; exclusions, corrections and reclassification history.
Evidence should match the claim
A contract establishes promised rights. Telemetry and operating records establish observed delivery. A workload test establishes performance within its stated scope, while commercial records establish consideration. No one form of evidence establishes all four. Retain each source's provenance and uncertainty, and protect customer workload contents and unrelated commercial information.
Reconcile aggregate supply with underlying capacity and time windows. Match each price observation to the service profile, terms and configuration effective at agreement or delivery, as the methodology requires. Preserve earlier published values and explain historical corrections.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
8
Annex B. Qualification and benchmark controls
Define the test before interpreting its results.
Specify the intended service and commercial exposure, minimum entitlements and proposed service class. Set the workloads, connected scale, software, quality criteria, latency or throughput targets, contention conditions and tolerances in advance.
Measure sustained operation and document limitations.
Repeat tests across relevant allocations or periods, and use operating histories for reliability claims. Record missing evidence and investigate differences without assuming a universal conversion between CPU core-hours, GPU-hours or other service units.
Scoped research example.
On Longhorn's V100 cluster, Sinha et al. report 14% runtime dispersion for single-GPU ResNet-50, 22% for four-GPU ResNet-50 and 1% for PageRank. Their metric is the boxplot whisker range divided by the median, excluding outliers; it is not an average slowdown [8]. These results and subsequent variability-aware scheduling research support workload-specific qualification, not a variability estimate for B200 cloud offerings [9].
Slynko et al. report a 37.82% normalized range in device-to-device memory-copy bandwidth for H200 SXM GPUs across their sampled devices and providers, calculated as (maximum minimum) divided by the mean [16]. This is a microbenchmark result within the study's sample, not an application-performance or B200 variability estimate. It provides further support for measuring the delivered service rather than assuming equivalence from the GPU model alone.
Review classes and test resilience.
Requalify after material changes. Test source withdrawal, related-party concentration, changes in the service mix and missing data. Compare relevant procurement-price changes where records exist, and separately assess settlement and hedge behavior for the proposed derivative.
Make inputs and calculations auditable
Data provenance.
Publish source categories, collection channels, sampling frequency, observation windows and coverage limitations. For each published value, explain how completed commercial transactions, availability-verified offers, panel assessments and unverified posted or administered rates contributed to the calculation. State the measurement basis, age of observations and any fallback use. Identify verification purchases separately and exclude them from weights based on ordinary commercial transaction volume. Retain source identifiers, timestamps, revisions, exclusions and related-party checks. Authorized independent reviewers should have access to confidential records.
Calculation and changes.
Publish the aggregation formula, weights and caps, unit conversions, quality adjustments, outlier treatment, minimum-data thresholds and fallback order. An aggregation method alone cannot establish service comparability, representative inputs or contributor independence; each needs supporting evidence. Identify permitted discretion and record reasons for overrides. Provide worked calculations, and publish material changes, reasons and effective dates through a defined notice or consultation process.
Protect benchmark decisions from conflicts
Ownership and separation of responsibilities.
Disclose material ownership, control and commercial relationships with trading firms, contributors and benchmark users. Use documented responsibilities, recusals and independent oversight to separate data selection, calculation, corrections and methodology decisions from conflicted sales, trading and investor interests. Restrict access to confidential inputs and unpublished values, and prevent preferential disclosure to investors or affiliates.
Accountability.
Keep decision logs, provide independent review and complaints procedures, and assess whether controls work in practice. IOSCO's benchmark principles provide a reference for governance and accountability [17]. These safeguards are proposals. We assert neither current compliance nor a particular ownership model.
Observe connected influence and feedback
Map economic groups across capacity supply, benchmark contributions and reservations. Where records are available, compare the timing of futures-price changes, reservation activity, committed capacity and eligible spot availability. Reservations may respond faster than deployment, power and construction. Such feedback may support investment or tighten near-term availability, but overlapping roles or movements occurring together do not establish causation or manipulation.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
9
Annex C. Executable offer verification
This proposed framework sets out the evidence needed to verify executable offers. The method should be validated for its target service and customer class before settlement use. Verification cannot establish availability for every buyer or delivery period.
1 Define the offer.
Record the service class, price and mandatory charges, purchasable increment, duration, delivery window, location and customer eligibility conditions. The account used for verification should represent the customer class measured by the benchmark. Check the hardware variant and purchase mode against provisioning records and applicable terms, including PCIe versus SXM and on-demand versus interruptible access.
2 Verify executable access.
The administrator, or a verification agent, should conduct the provisioning test independently of the provider being tested. A successful test, with defined limits, can verify a purchasable increment and its applicable rate within the observation window. Binding reservations or independently verifiable commercial executions may provide alternative evidence. Respect minimum commitments, release or cancellation terms, provider access rules and operational limits.
3 Limit quantity claims and weights.
Record the tested quantity and conditions at that time. Keep any larger claimed executable quantity separate, with its supporting evidence. Quantity weights require consistent service and duration, verified available quantity, checks for duplicate or shared inventory, and contributor caps. A minimum-increment test does not automatically determine index weight. Identify verification purchases as tests and exclude them from weights based on ordinary commercial transaction volume.
4 Record failures and limitations.
Keep unverified listings in a separate information tier. Distinguish confirmed unavailability from account restrictions, interface errors and inconclusive tests. Retain timestamps, endpoints, terms, rate confirmations, underlying provisioning requests and raw responses, outcomes, verified increments and supporting records for independent and supervisory review. Protect credentials and unrelated customer information while preserving the evidence needed to reproduce the determination.
5 Publish rules and control tests.
Publish admissibility criteria, observation windows, weighting rules, data-sufficiency thresholds and fallback procedures before use. Specific test times may be randomized and withheld publicly to reduce preferential treatment, but should remain available for audit. Disclose the evidence category used by any fallback; do not describe an indicative assessment as a verified executable offer.
Independent access and controls.
Reviewers and supervisors should reproduce determinations from retained records covering eligible observations, exclusions, credits and reversals. Check selective reporting, common control, wash activity and duplicates. Assess meaningful eligible volume alongside counterparty diversity; counts alone are insufficient. Separate commercial incentives from eligibility and corrections, with independent oversight of investor and affiliate conflicts. Scale audit intensity to contributor influence; exclude unverifiable material inputs from settlement.
Contributed record access.
For contributed non-public records supporting settlement, the administrator should hold enforceable rights to audit, retain and produce the necessary records to the listing DCM and the Commission. Exclude material inputs if that access cannot be demonstrated. Confirm that records can lawfully be provided in the relevant jurisdictions. Retain administrator-observed listings and verification records separately. Those observations do not require a contributor agreement, but cannot substitute for inaccessible contributed records.
Assess economic incentives.
The listing DCM should assess the net economic cost of influencing settlement under the proposed formula, observation window, weights and contributor caps. Compare plausible costs with potential gains across relevant derivative positions, including affiliated and cross-market exposures where material. Consider changes to offers, executed activity and withholding of eligible capacity or inputs. A successful small provisioning test does not establish that moving settlement would be uneconomic.
Verification creates evidence against which an offer can be checked. It does not eliminate manipulation or establish independent commercial demand or representative market coverage. Cost and feasibility depend on access, interfaces and minimum commitments. Babbage's series does not currently apply this framework.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
10
Supporting sources and submitting organisation
[1] CFTC, Request for Comment on the Listing of Compute Derivatives Contracts, 21 August 2026.
Question numbering and consultation context.
[2] Babbage Index, "H100 Price by Tier," frozen chart data, 3 October 2026.
Data 14:10:09 UTC; archive 16:02:14 UTC; methodology 0.56.
Archived source metadata.
Provider groupings, not verified service classes.
[3] AWS, EC2 price list, US East (N. Virginia), version 25 September 2026.
p5.48xlarge, Linux OnDemand, SKU 3D4V8UAYEMB38GU2: $55.04 per eight-H100 instance-hour.
[4] Lambda, on-demand pricing, archived 3 October 2026.
Eight-H100 SXM 80 GB instance: $3.99 per GPU-hour; self-service, subject to availability; taxes extra.
[5] NVIDIA, Time-Slicing GPUs in Kubernetes.
Allocation and time-sharing behavior.
[6] NVIDIA, Supported MIG Profiles.
B200 profiles and resource fractions.
[7] NVIDIA, DCGM runtime statistics.
Clock-slowdown reporting for power and thermal constraints.
[8] Sinha et al., Not All GPUs Are Created Equal: Characterizing Variability in Large-Scale, Accelerator-Rich Systems, SC '22. Sub Sec. III, V-A and V-D of arXiv:2208.11035v3. Longhorn workload-specific runtime dispersion and its definition.
[9] Jain et al., PAL: A Variability-Aware Policy for Scheduling ML Workloads in GPU Clusters, SC '24. Sub Sec. II-A and III-A of arXiv:2408.11919v2. Workload dependence and variability-aware scheduling; not independent replication of every prior result.
[10] CFTC, Barclays order, 27 June 2012.
LIBOR submissions, public dissemination and documented incentives.
[11] Ornn, OCPI Methodology.
Administrator's description of its transaction-based indices; not independently certified here.
[12] Silicon Data, GPU Rental Price Indices.
Administrator's published approach to normalization and basis adjustment; not independently certified here.
[13] ICE Futures U.S., New Compute Index Reference Data to Be Made Available on Monday, October 5, 2026, exchange notice, 29 September 2026, p. 1. Planned OCPI H100 and B200 futures; listing pending completion of the CFTC comment period, with no listing date or timeline set in that notice.
[14] CFTC Division of Market Oversight, Extension of Initial Review Period of NYMEX Submission No. 26-370 and Submission No. 26-370S for an Additional 45-Days, letter dated 21 September 2026, p. 1. Initial review period extended until the end of 9 November 2026.
[15] CFTC, 17 CFR 145.9, Petition for confidential treatment of information submitted to the Commission. Confidential-treatment requests do not guarantee protection from disclosure.
[16] Slynko et al., Did You Win the GPU Cloud Lottery? Benchmarking from TFLOPS to Tokens/$, GPGPU '26, 22 March 2026. Sub Sec. 2.1 and 3.1, Table 2. H200 SXM device-to-device memory-copy bandwidth: 37.82% normalized range within the study sample. DOI: 10.1145/3818671.3818674.
[17] Financial Stability Board, Principles for Financial Benchmarks, 17 July 2013. Overview of IOSCO's governance, methodology and accountability principles.
References [1], [5]-[12], [15] and [17] were rechecked on 10 October 2026. References [13], [14] and [16] were consulted on 18 October 2026. Historical H100 observations [2]-[4] retain their 3 October 2026 evidence date. The named rates are published offers, not verified purchases. The service-mix and hedge examples are hypothetical. No proprietary transaction records, completed verification pilot or independent audit findings are asserted on Babbage's behalf.
Submitting organisation and contact
BABBAGE LTD's work concerns compute instance specifications, service comparability and price indices. Babbage's commercial interest is stated on the opening page.
BABBAGE LTD | Company number 14360828
Registered office and correspondence: 128 City Road, London, United Kingdom, EC1V 2NX
Website: babbageindex.com | Email: methodology@babbageindex.com
*
Original text of letter here: https://www.regulations.gov/comment/CFTC-2026-1850-0046
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
1
Defining the compute service behind the price
BABBAGE LTD response on compute derivatives
To: Commodity Futures Trading Commission
Re: Request for Comment on the Listing of Compute Derivatives Contracts
RIN 3038-AF77 | FR Doc. 2026-17163 | Revised 18 October 2026
Submitted by BABBAGE LTD | Company number 14360828
Position and recommendations
Babbage supports developing ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY 1 Defining the compute service behind the price BABBAGE LTD response on compute derivatives To: Commodity Futures Trading Commission Re: Request for Comment on the Listing of Compute Derivatives Contracts RIN 3038-AF77 | FR Doc. 2026-17163 | Revised 18 October 2026 Submitted by BABBAGE LTD | Company number 14360828 Position and recommendations Babbage supports developingcompute derivatives and reference data in parallel. A useful reference price must identify what the buyer can purchase, establish credible evidence of its price and relate that price to the exposure being hedged. For infrastructure rentals, the product is an instance or coordinated group of instances, with defined resources, access rights, service obligations and duration. The processor or accelerator model is only one attribute.
These distinctions have practical consequences. A lower quote may buy a smaller resource allocation, different connectivity or weaker service-level commitments for availability, incident response and restoration. A falling average may reflect a change in the mix of services. Even a completed transaction may be unsuitable for a particular index, and an advertised price does not establish that capacity was available to purchase.
Babbage proposes to help procurement teams compare offers, explain price differences and assess renewals consistently. A clearer account of the resources, sustained performance and service-level commitments being purchased would improve supplier selection and negotiation. We therefore recommend the following safeguards for compute reference prices.
Define the purchased service.
Publish a common minimum set of disclosures covering resource rights, the surrounding system, connected scale, sustained performance, operational assurance and commercial terms.
Establish comparability over time.
Publish service-class eligibility criteria and tolerances supported by evidence. Explain how changes in configuration, allocation, service-level commitments and sample composition are handled. Distinguish services only where the difference matters to the intended contract.
Make prices verifiable.
Retain traceable evidence and independently verify material inputs. Publish proportionate information about methodology, coverage, limitations, assurance and controls over investor and affiliate conflicts. Identify public listings, executable offers and completed transactions separately.
This response focuses on infrastructure rentals and addresses Q1(e), Q2(a), Q2(g), Q3(c) and related questions on verification and capacity [1]. The proposed framework applies to CPU-only instances and systems using GPUs or other accelerators. Inference services require their own units and eligibility rules. Our common specification, provider mappings and qualification framework remain proposals.
Interest of the commenter.
Babbage publishes a daily B200 accelerator price index based on provider-published rates and intends to seek its use as a settlement reference. We have a direct commercial interest in this proceeding and are a prospective competitor of other index administrators.
Our series does not currently apply the availability verification proposed here. We recommend the same safeguards for Babbage as for other administrators. We claim neither a settlement-ready index nor audited market-wide coverage, and provide no estimates of transaction coverage or producer concentration.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
2
Q1(e). Define economically comparable services
An instance-hour should specify resources and commercial rights. Hardware can be stored; unused access time expires. Distinguish reserved time, elapsed access and metered usage, and whether the buyer receives a whole device, a partition, shared access or a connected cluster.
Published prices illustrate the problem.
Babbage's archived H100 tier display groups providers, not service quality [2]. On 3 October 2026, AWS listed p5.48xlarge (eight H100s, Linux on-demand, US East/N. Virginia) at $55.04 per instance-hour, or $6.88 per GPU-hour [3]. Lambda listed its eight-H100 SXM instance at $3.99 per GPU-hour [4]. These rates require comparison of host resources, connectivity, availability and incident-response commitments, region and purchase terms. They establish neither executable availability nor arbitrage. Competition, discounts and commercial strategy may also explain the gap.
Specify the allocation.
Virtualization alone does not establish sharing: exclusive devices, hardware partitions and scheduled shared access confer different rights. NVIDIA documents time-slicing in bare-metal and virtualized deployments and warns that requesting more time-sliced resources does not guarantee proportional compute access [5].
Memory capacity alone is insufficient. NVIDIA's B200 Multi-Instance GPU (MIG) profiles 1g.45gb and 2g.45gb have equal nominal memory but one-seventh and two-sevenths of the streaming multiprocessors, respectively [6]. Record the exact profile and any further sharing. Equal fractions of memory, compute resources and scheduled time are not necessarily interchangeable.
Describe the complete system and its connections.
Specify CPU architecture and model, physical-core or vCPU allocation, simultaneous multithreading (SMT) and sharing conditions, host memory capacity and bandwidth, storage, software and network conditions. Where accelerators are present, specify their allocation and connections to the host and one another. Identical GPU counts can span different communication domains. Instances, rack allocations and multi-rack clusters need explicit boundaries and simultaneous availability. Fabric names leave topology, oversubscription and usable bandwidth unresolved.
Measure sustained performance.
Cooling and power policy affect performance under stated conditions. NVIDIA exposes clock-slowdown statistics for thermal and power constraints [7]. Test representative runs after thermal steady state, with relevant contention, repeated allocations and disclosed variation. Published high-performance computing (HPC) measurements show workload-dependent runtime variability within one GPU model; Annex B gives a scoped example [8][9]. Cooling methods or virtualization technology alone should not determine service grades.
Assess reliability, availability and service commitments.
Short tests cannot establish long-term reliability or rare-failure rates. Assess reliability, availability and serviceability (RAS) from operating records. Specify contractual incident-response and restoration obligations, coverage hours, replacement capacity, and service-level agreement (SLA) availability targets, exclusions and remedies. Distinguish response from restoration commitments, and provider declarations from contractual promises and observed outcomes. Service credits may leave lost work and recovery costs uncompensated.
Qualification should specify workload, output-quality and latency requirements, numerical settings and tested software. For agentic workloads, include relevant CPU-based tool execution, orchestration and interactions with inference services. Include communication, queueing and recovery where relevant. Total cost per successfully completed workload supports comparison, but cannot supply a universal conversion between architectures or replace settlement-price evidence.
Group services by economically material differences
Classes should reflect differences affecting the intended exposure. Across architectures, apply the same material resource, software, precision, quality and service requirements for the specified use. Memory or interruption rights may preclude substitution despite similar throughput on one workload. Disclose each conversion or performance adjustment's calibration, scope and uncertainty.
Classes may be too narrow to yield enough independent prices. Disclose this limitation and narrow or defer settlement use, rather than pool incompatible services. Keep classes open to competing implementations and administrators. Annexes A and B propose specification and qualification procedures.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
3
Q2(a). Establish verifiable price evidence
Defining the service must be followed by verifying its price. Each settlement observation needs a service class and supporting evidence: a trade can concern an ineligible service, while an eligible service may have only an advertised price. These recommendations also address Q1(c) and Q2(b).
The LIBOR lesson for price verification
LIBOR shows why publication alone is insufficient. Panel banks assessed borrowing costs rather than simply reporting trades. The CFTC documented submissions influenced by derivatives positions and reputational concerns, although individual submissions and the fixing were public. A credible market connection and controls over contributor incentives remain essential [10].
Competition can discipline prices while leaving index inputs vulnerable to selective reporting, omitted discounts or unavailable offers. These are design risks, not allegations about any provider. Publicly verifiable terms and prices help buyers compare offers, challenge discrepancies and negotiate, while helping entrants demonstrate value. Each observation should identify its evidence category.
Evidence category
What it establishes and what must be checked
Completed transactions
Agreed service, quantity, period and net consideration. Reconcile contracts, invoices, credits and usage; seek counterparty corroboration where appropriate.
Firm executable offers
Accessible terms and supply, subject to quantity, duration and acceptance rights. Verify the service, price, quantity, delivery window and customer access conditions within the observation window; distinguish offers from completed commercial transactions.
Indicative listings or estimates
Context about advertised or assessed prices. Identify assumptions and limits. Unverified listings should not enter an executable-offer settlement reference; any assessed-price fallback requires separately disclosed rules, evidence and safeguards.
Independently verified confidential trades in the intended market may provide stronger evidence than public listings. Mandatory invoice publication could weaken representativeness. A public payment record alone establishes neither compute unit price nor delivered quantity, quality or counterparty independence. Whatever the recording system, link net consideration to the eligible service, quantity, terms and delivery evidence.
Verify advertised availability
Advertised rates should enter an executable-offer settlement reference only with evidence that the specified customer class could buy the service and quantity at that price during the observation window. Evidence may come from a bounded provisioning test conducted independently of the provider being tested, a binding reservation, or an independently verifiable commercial execution.
A successful test verifies only the quantity and conditions tested. It does not establish total supply, justify quantity weights or demonstrate representative coverage. Keep unverified listings available for information. Annex C details testing, weighting and record controls.
Public disclosure should cover service definitions, sources and collection, calculation and fallback rules, coverage, concentration, corrections and assurance findings. Aggregate or delay sensitive records. Public scrutiny and independent confidential verification serve different purposes. Annex B details calculation and governance disclosures.
Independent review requires records from which decisions can be reproduced, checks for related parties and duplicates, and oversight of conflicts. Administrators should hold enforceable audit and supervisory production rights for contributed non-public records and exclude unverifiable material inputs. Annex C provides the detailed controls.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
4
Q2(g). Preserve the meaning of the reference price
Existing methodologies provide a foundation. Ornn describes GPU-specific transaction indices and verification of hardware, software, host resources, connectivity and reliability. Silicon Data describes normalization for hardware, rental terms, geography and platform differences [11][12].
ICE Futures U.S.'s 29 September 2026 notice identifies planned OCPI H100 (HPR) and B200 (BKL) futures based on Ornn's indices. The notice states that listing is pending completion of the CFTC comment period and that no listing date or timeline had been set [13]. On 21 September 2026, the CFTC Division of Market Oversight extended the initial review period for NYMEX Submission Nos. 26-370 and 26-370S, concerning proposed Silicon Data H100 and B200 Rental Index Futures, until the end of 9 November 2026, citing this Request [14].
Babbage proposes complementary, transparent service specifications and workload qualification to establish which offers are comparable for a stated use and how reference prices relate to buyers' costs. Genuine price differences within qualified classes should remain visible. Eligibility alone does not establish equivalent workload outcomes. The proposed specification would require pilot validation.
We have not audited either provider's records or determined whether either index satisfies the criteria for a particular contract. Babbage's series has not been validated against these settlement safeguards.
Assess each candidate's service and market, eligible activity, concentration, verification access, calculations, adjustments, governance and disruption procedures. For indices claiming similar exposures, compare service boundaries, terms, observation windows, sources and weights to explain material divergence. Valid indices need not produce identical prices, but hedgers should understand why they differ.
Worked example of a change in the service mix
Consider two services sold under the same GPU label. Class A is exclusive; Class B is shared and interruptible. In this hypothetical example, both prices stay unchanged while the mix of billed hours shifts. The denominator counts billed service-hours across different services, not equivalent GPU capacity.
Class
Price per billed hour
Both periods
Period 1
Billed hours
Period 2
Billed hours
A: exclusive
$4.00
80
20
B: shared / interruptible
$2.00
20
80
Illustration only. Period 1: (80 x $4 + 20 x $2) / 100 = $3.60. Period 2: (20 x $4 + 80 x $2) / 100 = $2.40. The blended average falls 33.3%; each class price changes by 0%. The blended average correctly describes expenditure per billed hour for each period's mix. It does not show that an unchanged service became cheaper. Keeping the first period's 80/20 weights would leave the basket at $3.60. Fixed weights illustrate the distinction, but do not by themselves establish a suitable settlement index or resolve obsolescence and liquidity problems.
Explain how the index handles change
Each observation therefore needs a versioned service profile. Specify constituent entry and exit, materiality thresholds, requalification, reclassification and adjustments supported by evidence. Distinguish a changed contract from temporary underperformance of unchanged obligations. An outage should not disappear from the record merely because the service is reclassified.
Report eligible activity, freshness and concentration by independent economic group within each class. Test source withdrawal and service-mix changes. Publish time-limited fallback or delay rules, cessation procedures and contractual settlement remedies.
Where adequacy is unproven, undertake targeted data collection and pilot validation, and defer settlement use until adequacy is demonstrated. A universal standard or elimination of all basis risk is unnecessary. Informational series can support procurement with disclosed limits; settlement also requires market depth, governance and resilience. Undocumented discretion or prices inferred from the derivative itself should not fill evidence gaps.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
5
Q3(c). Match the contract to the buyer's exposure
Even a sound reference price can track a different exposure from the buyer's. Proponents should demonstrate how the index relates to the intended service, location, purchase mode, period and quantity. Spot access, reserved capacity rights and spending commitments may follow different price patterns.
Contract terms should make that relationship explicit. They should identify the service entitlement, pricing unit, connected allocation where relevant, geography, time window, included charges and methodology version. They should explain how corrections, source failure, hardware transitions and cessation are handled. A successor GPU or different contractual availability, response or restoration commitments should not silently replace the original exposure.
The buyer also needs to understand what remains uncovered. A price hedge does not provision capacity, restore jobs or reimburse all recovery costs. Switching providers may involve egress charges. Software portability, migration time and data-location restrictions also matter. RAS and SLA disclosures should identify covered resources, restoration obligations, exclusions and remedies.
Disclosures should distinguish costs included in the reference from those outside the hedge. They should also explain exposure to quantity and timing changes, margin calls and differences between procurement prices and the index.
Worked example of an incomplete hedge
Suppose a buyer needs one million instance-hours of a specified service. Its procurement price rises by $1.00 per instance-hour, while the index rises by $0.10 per index unit. A hypothetical linear hedge paying that index change on one million units produces $100,000 against a $1 million cost increase, leaving $900,000 exposed.
This illustration assumes entry at the initial index level and settlement at the ending level. It excludes fees, financing, taxes and margin timing. It is not a measured hedge result.
Under these assumptions, the relationship is:
change in net cost = (purchased quantity x procurement-price change) (hedge quantity x index change)
Similar GPU labels give no assurance that those price changes will track each other. We recommend testing matched changes over relevant procurement periods, including stress and out-of-sample observations. Correlation between trending price levels is insufficient. Actual hedge effectiveness also depends on the derivative's terms, maturity, liquidity and the hedge implemented.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
6
Measure the capacity buyers can access
For Q1(b), Q1(g) and Q2(h), the relevant supply is the entitlement buyers can access in the specified window. Installed accelerator counts are insufficient. Distinguish owners, providers, resellers and independent economic groups. Multiple listings, endpoints, payment addresses or virtual allocations may refer to the same supplier or capacity. Count enforceable rights consistently with supporting resources and scheduling windows.
Connected allocations need compute, memory and networks together at the required time, with power, cooling and contractual access. Committed capacity is freely available only where its rights permit resale or release. We provide no market-wide quantity estimate. Annex B also addresses surveillance of influence and feedback between futures, reservations and eligible availability.
Supply estimates should distinguish announced or reserved resources from capacity that is installed, energized and accepted for service. Each price observation should identify quantity, delivery window and access rights, with evidence appropriate to that observation. An advertisement alone does not establish immediately purchasable supply.
Physical delivery requires lawful rights sufficient for that delivery, including any necessary assignment or resale. Cash settlement does not inherently require transfer of the resources, but its reference prices should represent the defined service and conditions of access.
Power data can corroborate capacity claims if independently supported. Use energized power attributable to the relevant systems, distinguish it from planned connections and allow for other loads. Estimates should use complete-system power under the qualified operating conditions. Convert facility power to IT power only once, using an appropriate overhead assumption. Neither permits nor an engineering estimate establishes available inventory, customer access or delivery readiness.
How Babbage can contribute
Babbage's proposed common specification, linking the service purchased to its economic comparability and price evidence, would require development and pilot validation. We would discuss the fields and qualification principles with competing providers and administrators. Adoption should depend on evidence and usefulness, with room for different technical implementations.
We offer Annexes A-C as starting points for technical working groups convened by the Commission, designated contract markets (DCMs) or administrators. We would be glad to provide staff, on request, with further detail on service specification, workload qualification and verification tests, subject to the confidential-treatment procedures noted below. We would welcome a meeting with the Division of Market Oversight or the CFTC Innovation Task Force.
Before submitting commercially sensitive information, we would contact the staff named in the Request and follow the confidential-treatment procedures in 17 CFR 145.9 [1][15]. The verification framework requires validation before any claim of settlement readiness.
Requested Commission action
We ask the Commission to encourage common minimum service disclosures, qualification supported by evidence, and verifiable prices. Advertised rates used in executable-offer settlement references should be verified as available for purchase. Administrators should hold audit and production rights for contributed non-public records, publish proportionate disclosures, and explain each input category's contribution and the basis for any quantity weights. Listing proponents should demonstrate suitability for the particular contract.
These requirements should support competition and remain proportionate to the proposed use. They should require the evidence needed for credible settlement without prescribing a proprietary methodology, mandating a cooling or virtualization technology, or treating every configuration difference as a separate market.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
7
Annex A. Minimum instance and price evidence schema
This proposed schema supports qualification and reporting. Detailed evidence should be retained for authorized verification, with service definitions, eligibility criteria and appropriate summaries made public. Full publication of underlying records is unnecessary. A field should define a separate class only when it matters to the intended exposure.
Field group
Information to retain
Identity and version
Provider and economic group; physical origin; SKU and version; region; observation and contract identifiers; effective dates.
Resource entitlement
Whole device or exact partition; exclusive or shared access; minimum scheduling share, interval and burst rules; isolation and interruption rights.
Complete system
CPU architecture and model; physical-core or vCPU allocation; SMT and sharing conditions; relevant NUMA placement; host memory capacity and bandwidth; accelerator type/count, usable memory and connections where present; storage; software and network entitlements.
Connected scale
Instance, server, rack and cluster boundaries; communication domains; topology; bandwidth and contention; minimum allocation available simultaneously.
Sustained capability
Power limits; relevant thermal conditions; qualification date; workload, numerical settings and quality/latency requirements; tested software and, where applicable, model versions; test duration and scale; contention; throughput, latency and variation.
RAS and SLA commitments
Fault, interruption and availability records; contractual incident-response and restoration obligations; coverage hours; replacement capacity and time to resume useful work; SLA availability targets, exclusions and remedies.
Commercial terms
Purchase mode; quantity; term; cancellation, transfer and commitment rights; included/excluded charges, including egress; software portability, migration time and data-location restrictions.
Delivery window and readiness
Offered delivery window; readiness status and supporting evidence as of the observation date; distinguish currently operational capacity from capacity committed for future delivery.
Price and evidence
Currency, billing unit and net consideration; discounts and credits; commercial transaction, availability-verified offer, panel assessment or indicative status; source and verification date; test outcome, verified increment and separately supported available quantity where applicable.
Verification and changes
Supporting record references; configuration and policy versions; material firmware, scheduling or power-limit changes; reassessment decision and effective date; reviewer and assurance scope; related-party checks; exclusions, corrections and reclassification history.
Evidence should match the claim
A contract establishes promised rights. Telemetry and operating records establish observed delivery. A workload test establishes performance within its stated scope, while commercial records establish consideration. No one form of evidence establishes all four. Retain each source's provenance and uncertainty, and protect customer workload contents and unrelated commercial information.
Reconcile aggregate supply with underlying capacity and time windows. Match each price observation to the service profile, terms and configuration effective at agreement or delivery, as the methodology requires. Preserve earlier published values and explain historical corrections.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
8
Annex B. Qualification and benchmark controls
Define the test before interpreting its results.
Specify the intended service and commercial exposure, minimum entitlements and proposed service class. Set the workloads, connected scale, software, quality criteria, latency or throughput targets, contention conditions and tolerances in advance.
Measure sustained operation and document limitations.
Repeat tests across relevant allocations or periods, and use operating histories for reliability claims. Record missing evidence and investigate differences without assuming a universal conversion between CPU core-hours, GPU-hours or other service units.
Scoped research example.
On Longhorn's V100 cluster, Sinha et al. report 14% runtime dispersion for single-GPU ResNet-50, 22% for four-GPU ResNet-50 and 1% for PageRank. Their metric is the boxplot whisker range divided by the median, excluding outliers; it is not an average slowdown [8]. These results and subsequent variability-aware scheduling research support workload-specific qualification, not a variability estimate for B200 cloud offerings [9].
Slynko et al. report a 37.82% normalized range in device-to-device memory-copy bandwidth for H200 SXM GPUs across their sampled devices and providers, calculated as (maximum minimum) divided by the mean [16]. This is a microbenchmark result within the study's sample, not an application-performance or B200 variability estimate. It provides further support for measuring the delivered service rather than assuming equivalence from the GPU model alone.
Review classes and test resilience.
Requalify after material changes. Test source withdrawal, related-party concentration, changes in the service mix and missing data. Compare relevant procurement-price changes where records exist, and separately assess settlement and hedge behavior for the proposed derivative.
Make inputs and calculations auditable
Data provenance.
Publish source categories, collection channels, sampling frequency, observation windows and coverage limitations. For each published value, explain how completed commercial transactions, availability-verified offers, panel assessments and unverified posted or administered rates contributed to the calculation. State the measurement basis, age of observations and any fallback use. Identify verification purchases separately and exclude them from weights based on ordinary commercial transaction volume. Retain source identifiers, timestamps, revisions, exclusions and related-party checks. Authorized independent reviewers should have access to confidential records.
Calculation and changes.
Publish the aggregation formula, weights and caps, unit conversions, quality adjustments, outlier treatment, minimum-data thresholds and fallback order. An aggregation method alone cannot establish service comparability, representative inputs or contributor independence; each needs supporting evidence. Identify permitted discretion and record reasons for overrides. Provide worked calculations, and publish material changes, reasons and effective dates through a defined notice or consultation process.
Protect benchmark decisions from conflicts
Ownership and separation of responsibilities.
Disclose material ownership, control and commercial relationships with trading firms, contributors and benchmark users. Use documented responsibilities, recusals and independent oversight to separate data selection, calculation, corrections and methodology decisions from conflicted sales, trading and investor interests. Restrict access to confidential inputs and unpublished values, and prevent preferential disclosure to investors or affiliates.
Accountability.
Keep decision logs, provide independent review and complaints procedures, and assess whether controls work in practice. IOSCO's benchmark principles provide a reference for governance and accountability [17]. These safeguards are proposals. We assert neither current compliance nor a particular ownership model.
Observe connected influence and feedback
Map economic groups across capacity supply, benchmark contributions and reservations. Where records are available, compare the timing of futures-price changes, reservation activity, committed capacity and eligible spot availability. Reservations may respond faster than deployment, power and construction. Such feedback may support investment or tighten near-term availability, but overlapping roles or movements occurring together do not establish causation or manipulation.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
9
Annex C. Executable offer verification
This proposed framework sets out the evidence needed to verify executable offers. The method should be validated for its target service and customer class before settlement use. Verification cannot establish availability for every buyer or delivery period.
1 Define the offer.
Record the service class, price and mandatory charges, purchasable increment, duration, delivery window, location and customer eligibility conditions. The account used for verification should represent the customer class measured by the benchmark. Check the hardware variant and purchase mode against provisioning records and applicable terms, including PCIe versus SXM and on-demand versus interruptible access.
2 Verify executable access.
The administrator, or a verification agent, should conduct the provisioning test independently of the provider being tested. A successful test, with defined limits, can verify a purchasable increment and its applicable rate within the observation window. Binding reservations or independently verifiable commercial executions may provide alternative evidence. Respect minimum commitments, release or cancellation terms, provider access rules and operational limits.
3 Limit quantity claims and weights.
Record the tested quantity and conditions at that time. Keep any larger claimed executable quantity separate, with its supporting evidence. Quantity weights require consistent service and duration, verified available quantity, checks for duplicate or shared inventory, and contributor caps. A minimum-increment test does not automatically determine index weight. Identify verification purchases as tests and exclude them from weights based on ordinary commercial transaction volume.
4 Record failures and limitations.
Keep unverified listings in a separate information tier. Distinguish confirmed unavailability from account restrictions, interface errors and inconclusive tests. Retain timestamps, endpoints, terms, rate confirmations, underlying provisioning requests and raw responses, outcomes, verified increments and supporting records for independent and supervisory review. Protect credentials and unrelated customer information while preserving the evidence needed to reproduce the determination.
5 Publish rules and control tests.
Publish admissibility criteria, observation windows, weighting rules, data-sufficiency thresholds and fallback procedures before use. Specific test times may be randomized and withheld publicly to reduce preferential treatment, but should remain available for audit. Disclose the evidence category used by any fallback; do not describe an indicative assessment as a verified executable offer.
Independent access and controls.
Reviewers and supervisors should reproduce determinations from retained records covering eligible observations, exclusions, credits and reversals. Check selective reporting, common control, wash activity and duplicates. Assess meaningful eligible volume alongside counterparty diversity; counts alone are insufficient. Separate commercial incentives from eligibility and corrections, with independent oversight of investor and affiliate conflicts. Scale audit intensity to contributor influence; exclude unverifiable material inputs from settlement.
Contributed record access.
For contributed non-public records supporting settlement, the administrator should hold enforceable rights to audit, retain and produce the necessary records to the listing DCM and the Commission. Exclude material inputs if that access cannot be demonstrated. Confirm that records can lawfully be provided in the relevant jurisdictions. Retain administrator-observed listings and verification records separately. Those observations do not require a contributor agreement, but cannot substitute for inaccessible contributed records.
Assess economic incentives.
The listing DCM should assess the net economic cost of influencing settlement under the proposed formula, observation window, weights and contributor caps. Compare plausible costs with potential gains across relevant derivative positions, including affiliated and cross-market exposures where material. Consider changes to offers, executed activity and withholding of eligible capacity or inputs. A successful small provisioning test does not establish that moving settlement would be uneconomic.
Verification creates evidence against which an offer can be checked. It does not eliminate manipulation or establish independent commercial demand or representative market coverage. Cost and feasibility depend on access, interfaces and minimum commitments. Babbage's series does not currently apply this framework.
BABBAGE | COMPUTE PRICING AND SERVICE COMPARABILITY
10
Supporting sources and submitting organisation
[1] CFTC, Request for Comment on the Listing of Compute Derivatives Contracts, 21 August 2026.
Question numbering and consultation context.
[2] Babbage Index, "H100 Price by Tier," frozen chart data, 3 October 2026.
Data 14:10:09 UTC; archive 16:02:14 UTC; methodology 0.56.
Archived source metadata.
Provider groupings, not verified service classes.
[3] AWS, EC2 price list, US East (N. Virginia), version 25 September 2026.
p5.48xlarge, Linux OnDemand, SKU 3D4V8UAYEMB38GU2: $55.04 per eight-H100 instance-hour.
[4] Lambda, on-demand pricing, archived 3 October 2026.
Eight-H100 SXM 80 GB instance: $3.99 per GPU-hour; self-service, subject to availability; taxes extra.
[5] NVIDIA, Time-Slicing GPUs in Kubernetes.
Allocation and time-sharing behavior.
[6] NVIDIA, Supported MIG Profiles.
B200 profiles and resource fractions.
[7] NVIDIA, DCGM runtime statistics.
Clock-slowdown reporting for power and thermal constraints.
[8] Sinha et al., Not All GPUs Are Created Equal: Characterizing Variability in Large-Scale, Accelerator-Rich Systems, SC '22. Sub Sec. III, V-A and V-D of arXiv:2208.11035v3. Longhorn workload-specific runtime dispersion and its definition.
[9] Jain et al., PAL: A Variability-Aware Policy for Scheduling ML Workloads in GPU Clusters, SC '24. Sub Sec. II-A and III-A of arXiv:2408.11919v2. Workload dependence and variability-aware scheduling; not independent replication of every prior result.
[10] CFTC, Barclays order, 27 June 2012.
LIBOR submissions, public dissemination and documented incentives.
[11] Ornn, OCPI Methodology.
Administrator's description of its transaction-based indices; not independently certified here.
[12] Silicon Data, GPU Rental Price Indices.
Administrator's published approach to normalization and basis adjustment; not independently certified here.
[13] ICE Futures U.S., New Compute Index Reference Data to Be Made Available on Monday, October 5, 2026, exchange notice, 29 September 2026, p. 1. Planned OCPI H100 and B200 futures; listing pending completion of the CFTC comment period, with no listing date or timeline set in that notice.
[14] CFTC Division of Market Oversight, Extension of Initial Review Period of NYMEX Submission No. 26-370 and Submission No. 26-370S for an Additional 45-Days, letter dated 21 September 2026, p. 1. Initial review period extended until the end of 9 November 2026.
[15] CFTC, 17 CFR 145.9, Petition for confidential treatment of information submitted to the Commission. Confidential-treatment requests do not guarantee protection from disclosure.
[16] Slynko et al., Did You Win the GPU Cloud Lottery? Benchmarking from TFLOPS to Tokens/$, GPGPU '26, 22 March 2026. Sub Sec. 2.1 and 3.1, Table 2. H200 SXM device-to-device memory-copy bandwidth: 37.82% normalized range within the study sample. DOI: 10.1145/3818671.3818674.
[17] Financial Stability Board, Principles for Financial Benchmarks, 17 July 2013. Overview of IOSCO's governance, methodology and accountability principles.
References [1], [5]-[12], [15] and [17] were rechecked on 10 October 2026. References [13], [14] and [16] were consulted on 18 October 2026. Historical H100 observations [2]-[4] retain their 3 October 2026 evidence date. The named rates are published offers, not verified purchases. The service-mix and hedge examples are hypothetical. No proprietary transaction records, completed verification pilot or independent audit findings are asserted on Babbage's behalf.
Submitting organisation and contact
BABBAGE LTD's work concerns compute instance specifications, service comparability and price indices. Babbage's commercial interest is stated on the opening page.
BABBAGE LTD | Company number 14360828
Registered office and correspondence: 128 City Road, London, United Kingdom, EC1V 2NX
Website: babbageindex.com | Email: methodology@babbageindex.com
*
Original text of letter here: https://www.regulations.gov/comment/CFTC-2026-1850-0046
Alaska Center Education Fund Opposes Proposed Census Changes Over Fair Representation and Data Integrity
Carter Struck
WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site.
TO: Daniel Sweeney, Deputy General Counsel for Economic, Statistical, and Regulatory Affairs, Office of the General Counsel, U.S. Department of Commerce FROM: The Alaska Center Education Fund
DATE: September 30, 2026
RE: Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (FR Doc. 2026-18481, RIN 0607-AA75)
Dear Mr. Sweeney,
The Alaska Center ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. TO: Daniel Sweeney, Deputy General Counsel for Economic, Statistical, and Regulatory Affairs, Office of the General Counsel, U.S. Department of Commerce FROM: The Alaska Center Education Fund DATE: September 30, 2026 RE: Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (FR Doc. 2026-18481, RIN 0607-AA75) Dear Mr. Sweeney, The Alaska CenterEducation Fund is a nonprofit organization working to educate and engage Alaskans of all ages to support clean air and water, build healthy and resilient communities, and become empowered stewards. On behalf of our members and supporters, we submit the following comments.
We oppose the proposed rule's changes to the residence criteria and demographic items for the 2030 Census. The proposed residence criteria and changes to demographic questions would undermine the accuracy, integrity, and usefulness of census data. Our concerns follow:
1. Excluding noncitizens without lawful permanent residence from census apportionment data is unconstitutional and would deprive our communities of their fair share of resources and representation. The Constitution, statutes, and settled law clearly provide that all persons residing in the United States must be counted for the apportionment of Congressional seats. Research has also shown that adding a question on citizenship to the decennial census form would have a chilling effect on census participation. If the census does not count all residents, our communities will lose resources to support their schools, hospitals, transportation infrastructure, emergency services, and more. This would impact many Alaskans who rely on resources like SNAP (nearly 9% of Alaskans, including thousands of children, elders and veterans) and Medicaid (more than one-third of Alaskans used Medicaid services as of September 2024).
2. https://health.alaska.gov/en/office-of-the-commissioner/transformation-of-care/ https://usafacts.org/answers/how-many-people-receive-snap-benefits-in-the-us-every-month/state/alaska/
2. Eliminating census race and ethnicity data collection is unprecedented, violates statutes, and severely undermines the utility of census data for a range of critical purposes. Since the first census in 1790, every single decennial census has collected race and ethnicity data in some form. Census race and ethnicity data underpin many other critical federal datasets and serve as an irreplaceable benchmark for nearly every population-based survey and study in the U.S., ensuring that our research is reliable and leads to better outcomes for all of us. Race and ethnicity data are vital to protecting public health and safety because they allow health officials to identify and address differences in health risk factors. They are necessary to enforce civil rights laws and ensure equal opportunity for all Americans, helping to ensure fairness and strengthen the U.S. economy. Race and ethnicity data also give businesses market insights to grow and compete and help local governments and nonprofits target services more efficiently.
Additionally, a 30-day comment period is inadequate for the public to review and provide input on this rule, given the scope of the proposed changes and the effect they would have on census data. The Department of Commerce and the Census Bureau must immediately publish a Federal Register notice extending the comment period by at least 60 days to provide a meaningful opportunity for public input.
We look forward to engaging with you further on this process. Please feel free to contact us with any questions.
Sincerely,
Piper Haugan
Civic Engagement Program Manager, The Alaska Center Education Fund
piper@akcenter.org
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-39902
TO: Daniel Sweeney, Deputy General Counsel for Economic, Statistical, and Regulatory Affairs, Office of the General Counsel, U.S. Department of Commerce FROM: The Alaska Center Education Fund
DATE: September 30, 2026
RE: Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (FR Doc. 2026-18481, RIN 0607-AA75)
Dear Mr. Sweeney,
The Alaska Center ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. TO: Daniel Sweeney, Deputy General Counsel for Economic, Statistical, and Regulatory Affairs, Office of the General Counsel, U.S. Department of Commerce FROM: The Alaska Center Education Fund DATE: September 30, 2026 RE: Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (FR Doc. 2026-18481, RIN 0607-AA75) Dear Mr. Sweeney, The Alaska CenterEducation Fund is a nonprofit organization working to educate and engage Alaskans of all ages to support clean air and water, build healthy and resilient communities, and become empowered stewards. On behalf of our members and supporters, we submit the following comments.
We oppose the proposed rule's changes to the residence criteria and demographic items for the 2030 Census. The proposed residence criteria and changes to demographic questions would undermine the accuracy, integrity, and usefulness of census data. Our concerns follow:
1. Excluding noncitizens without lawful permanent residence from census apportionment data is unconstitutional and would deprive our communities of their fair share of resources and representation. The Constitution, statutes, and settled law clearly provide that all persons residing in the United States must be counted for the apportionment of Congressional seats. Research has also shown that adding a question on citizenship to the decennial census form would have a chilling effect on census participation. If the census does not count all residents, our communities will lose resources to support their schools, hospitals, transportation infrastructure, emergency services, and more. This would impact many Alaskans who rely on resources like SNAP (nearly 9% of Alaskans, including thousands of children, elders and veterans) and Medicaid (more than one-third of Alaskans used Medicaid services as of September 2024).
2. https://health.alaska.gov/en/office-of-the-commissioner/transformation-of-care/ https://usafacts.org/answers/how-many-people-receive-snap-benefits-in-the-us-every-month/state/alaska/
2. Eliminating census race and ethnicity data collection is unprecedented, violates statutes, and severely undermines the utility of census data for a range of critical purposes. Since the first census in 1790, every single decennial census has collected race and ethnicity data in some form. Census race and ethnicity data underpin many other critical federal datasets and serve as an irreplaceable benchmark for nearly every population-based survey and study in the U.S., ensuring that our research is reliable and leads to better outcomes for all of us. Race and ethnicity data are vital to protecting public health and safety because they allow health officials to identify and address differences in health risk factors. They are necessary to enforce civil rights laws and ensure equal opportunity for all Americans, helping to ensure fairness and strengthen the U.S. economy. Race and ethnicity data also give businesses market insights to grow and compete and help local governments and nonprofits target services more efficiently.
Additionally, a 30-day comment period is inadequate for the public to review and provide input on this rule, given the scope of the proposed changes and the effect they would have on census data. The Department of Commerce and the Census Bureau must immediately publish a Federal Register notice extending the comment period by at least 60 days to provide a meaningful opportunity for public input.
We look forward to engaging with you further on this process. Please feel free to contact us with any questions.
Sincerely,
Piper Haugan
Civic Engagement Program Manager, The Alaska Center Education Fund
piper@akcenter.org
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-39902
AARP Urges IRS to Simplify Retirement Account Rollovers to Prevent Savings Leakage
Carter Struck
WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site.
October 20, 2026
Internal Revenue Service
Attn: CC:PA:01:PR (Notice 2026-49)
Room 5503
P.O. Box 7604
Ben Franklin Station
Washington, D.C. 20044
Re: Notice 2026-49, Guidance on Section 324 of the SECURE 2.0 Act with Respect to Rollovers
Dear Mr. Burns:
AARP, which advocates for the 125 million Americans age 50 and older, appreciates the opportunity to comment on the Internal Revenue Service's ("IRS") proposed ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. October 20, 2026 Internal Revenue Service Attn: CC:PA:01:PR (Notice 2026-49) Room 5503 P.O. Box 7604 Ben Franklin Station Washington, D.C. 20044 Re: Notice 2026-49, Guidance on Section 324 of the SECURE 2.0 Act with Respect to Rollovers Dear Mr. Burns: AARP, which advocates for the 125 million Americans age 50 and older, appreciates the opportunity to comment on the Internal Revenue Service's ("IRS") proposedNotice 2026-49, which proposes sample forms and standardized procedures to implement Section 324 of the SECURE 2.0 Act of 2022. AARP has long supported policies that make retirement savings more accessible, portable, and secure, including many of the reforms in SECURE 2.0. We strongly support the IRS' work to simplify how workers move savings between employer-sponsored plans and individual retirement accounts (IRAs). Congress laid the groundwork in SECURE 2.0, and implementation will determine whether workers and retirees actually see the benefit. Done well, this guidance will cut needless paperwork for savers and help Americans preserve and consolidate the savings they have built over a working lifetime.
Simplifying Retirement Savings Portability is Critical to Preventing Savings Leakage
Americans increasingly accumulate retirement savings through multiple employers over the course of their careers and changing jobs also means confronting complicated decisions about what to do with their workplace retirement accounts. While these workers generally have the option to roll over their retirement account to another eligible retirement plan or IRA, the process often requires coordinating with multiple plan administrators, completing sometimes complex and unfamiliar paperwork, and navigating difficult administrative and tax requirements. This is unnecessarily complicated. Additionally, in some circumstances, participants receive paper checks that must be deposited or forwarded to another retirement plan within applicable deadlines to avoid adverse tax consequences, and this occurs even when the worker provides information about the plan or IRA to which they wish their account to be transferred.
Guidance on Section 324 of the SECURE 2.0 Act with Respect to Rollovers, Notice 2026-49 (Aug. 12, 2026). SECURE 2.0 Act of 2022, Pub. L. 117-328, Division T Sec. 324 (Dec. 29, 2022). Bureau of Labor Statistics, "Number of Jobs, Labor Market Experience, Marital Status, and Health for Those Born 1957-1965" Aug. 26, 2025, https://www.bls.gov/news.release/pdf/nlsoy.pdf (finding that "individuals bornin the latter years of the baby boom (1957-64) held an average of 12.9 jobs from 18 to 58, according to the U.S. Bureau of Labor Statistics (BLS)"). See generally 26 U.S.C. Sec. 402(c) (requiring transfers to be made within 60 days unless an individual petitions for a hardship waiver and if the requirements of this section are not met, the distribution will be includible in gross income and may be subject to early withdrawal penalties).
These burdens can discourage workers from consolidating their retirement accounts, increasing the likelihood of losing track of those retirement assets or in a worst-case scenario, withdrawing their savings. These problems can be particularly consequential for older workers, who may have accumulated retirement savings over decades and across several employers and have fewer working years left to replace savings lost through premature distributions. Additionally, such premature withdrawals can trigger taxes and early withdrawal penalties, thereby permanently reducing the resources available to support these workers in retirement. Even relatively modest balances can make a meaningful difference in retirement, particularly for households with limited financial resources.
Research consistently demonstrates the importance of addressing these challenges. In a recent paper published by the Pension Research Council, AARP Public Policy Institute Senior Policy Advisor Manita Rao examined the consequences of losing employer-sponsored retirement plan coverage following a job change. Using 24 years of household data, Rao found that workers who lose employer retirement plan coverage are significantly more likely to withdraw retirement savings and generally do not replace interrupted workplace savings with contributions to an IRA. Among workers with defined contribution retirement account balances, losing employer coverage increased the probability of cashing out retirement savings by 7.1 percentage points, approximately three times the pre-change cash-out rate. Notably, among workers who already held an IRA before changing jobs, the increase in cash-outs was even greater, at 14.4 percentage points. These findings suggest that simply having access to another tax-preferred retirement account is insufficient to ensure that workers preserve their savings, thus emphasizing the need to make rollovers easier for workers.
The long-term consequences of interrupted retirement savings can be substantial. For example, the paper includes a hypothetical demonstrating that a worker displaced at age 35 who experiences a seven-month employment interruption, lower wages, and four years without employer retirement plan coverage may save approximately 19 percent less in retirement savings by age 65 than a comparable worker without an interruption. If that worker also cashes out retirement savings upon separation from employment, the shortfall increases by 34 percent.
These findings reinforce the importance of preserving retirement savings when workers change jobs. While standardized rollover procedures cannot alone address all the challenges of losing employer retirement plan coverage, reducing the administrative burden of transferring a worker's savings is an important part of a broader strategy to prevent retirement savings leakage.
Support for Standardized Rollover Forms and Procedures
We appreciate the IRS's efforts to develop standardized forms and procedures to facilitate direct rollovers between eligible retirement savings accounts. In particular, we support efforts to standardize information requests, facilitate electronic transfers, improve communication between the distributing and receiving plans, and protect participants' personal information.
The proposed approach in Notice 2026-49 appropriately recognizes that retirement plan administrators and service providers are better positioned than individual participants to coordinate the technical and administrative aspects of transferring retirement assets. AARP believes the rollover burden on workers should be limited to deciding whether to roll over their savings and then requesting a rollover from their former employer's plan to their new employer's retirement plan. The balance of the rollover burden for such transfers should fall on the relevant service providers to coordinate the necessary steps for the transfer.
Behavioral economics has reshaped and improved the retirement savings ecosphere, so in that vein, we urge the IRS to focus on simplification and automation. When participants face burdens like contacting employers and completing multiple forms, among other things, these barriers reduce positive participant action, even if such actions would result in better outcomes for the worker. Accordingly, we urge the IRS to finalize the proposed standardized procedures and forms and consider additional guidance and safe harbors that would encourage widespread adoption by employers and retirement plan service providers.
Conclusion
For too long, workers have faced unnecessary hurdles when trying to preserve savings from a former employer. Notice 2026-49 is real progress toward a standardized process for moving retirement savings between plans. A streamlined approach would substantially reduce the burden placed on individual savers: the worker makes one request to the new plan, and the plans and their service providers handle the coordination. Savers carry far more of that burden under today's defined contribution system than workers did under traditional pension plans of the past. AARP appreciates the IRS's work on this guidance and welcomes the opportunity to discuss our comments further. If you have any questions, please feel free to contact me or have your staff contact Holly Biglow of AARP's Government Affairs team at hbiglow@aarp.org.
Sincerely,
Jennifer Jones
Vice President, Financial Security and Livable Communities
AARP
*
Original text of letter here: https://www.regulations.gov/comment/IRS-2026-0100-0013
October 20, 2026
Internal Revenue Service
Attn: CC:PA:01:PR (Notice 2026-49)
Room 5503
P.O. Box 7604
Ben Franklin Station
Washington, D.C. 20044
Re: Notice 2026-49, Guidance on Section 324 of the SECURE 2.0 Act with Respect to Rollovers
Dear Mr. Burns:
AARP, which advocates for the 125 million Americans age 50 and older, appreciates the opportunity to comment on the Internal Revenue Service's ("IRS") proposed ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. October 20, 2026 Internal Revenue Service Attn: CC:PA:01:PR (Notice 2026-49) Room 5503 P.O. Box 7604 Ben Franklin Station Washington, D.C. 20044 Re: Notice 2026-49, Guidance on Section 324 of the SECURE 2.0 Act with Respect to Rollovers Dear Mr. Burns: AARP, which advocates for the 125 million Americans age 50 and older, appreciates the opportunity to comment on the Internal Revenue Service's ("IRS") proposedNotice 2026-49, which proposes sample forms and standardized procedures to implement Section 324 of the SECURE 2.0 Act of 2022. AARP has long supported policies that make retirement savings more accessible, portable, and secure, including many of the reforms in SECURE 2.0. We strongly support the IRS' work to simplify how workers move savings between employer-sponsored plans and individual retirement accounts (IRAs). Congress laid the groundwork in SECURE 2.0, and implementation will determine whether workers and retirees actually see the benefit. Done well, this guidance will cut needless paperwork for savers and help Americans preserve and consolidate the savings they have built over a working lifetime.
Simplifying Retirement Savings Portability is Critical to Preventing Savings Leakage
Americans increasingly accumulate retirement savings through multiple employers over the course of their careers and changing jobs also means confronting complicated decisions about what to do with their workplace retirement accounts. While these workers generally have the option to roll over their retirement account to another eligible retirement plan or IRA, the process often requires coordinating with multiple plan administrators, completing sometimes complex and unfamiliar paperwork, and navigating difficult administrative and tax requirements. This is unnecessarily complicated. Additionally, in some circumstances, participants receive paper checks that must be deposited or forwarded to another retirement plan within applicable deadlines to avoid adverse tax consequences, and this occurs even when the worker provides information about the plan or IRA to which they wish their account to be transferred.
Guidance on Section 324 of the SECURE 2.0 Act with Respect to Rollovers, Notice 2026-49 (Aug. 12, 2026). SECURE 2.0 Act of 2022, Pub. L. 117-328, Division T Sec. 324 (Dec. 29, 2022). Bureau of Labor Statistics, "Number of Jobs, Labor Market Experience, Marital Status, and Health for Those Born 1957-1965" Aug. 26, 2025, https://www.bls.gov/news.release/pdf/nlsoy.pdf (finding that "individuals bornin the latter years of the baby boom (1957-64) held an average of 12.9 jobs from 18 to 58, according to the U.S. Bureau of Labor Statistics (BLS)"). See generally 26 U.S.C. Sec. 402(c) (requiring transfers to be made within 60 days unless an individual petitions for a hardship waiver and if the requirements of this section are not met, the distribution will be includible in gross income and may be subject to early withdrawal penalties).
These burdens can discourage workers from consolidating their retirement accounts, increasing the likelihood of losing track of those retirement assets or in a worst-case scenario, withdrawing their savings. These problems can be particularly consequential for older workers, who may have accumulated retirement savings over decades and across several employers and have fewer working years left to replace savings lost through premature distributions. Additionally, such premature withdrawals can trigger taxes and early withdrawal penalties, thereby permanently reducing the resources available to support these workers in retirement. Even relatively modest balances can make a meaningful difference in retirement, particularly for households with limited financial resources.
Research consistently demonstrates the importance of addressing these challenges. In a recent paper published by the Pension Research Council, AARP Public Policy Institute Senior Policy Advisor Manita Rao examined the consequences of losing employer-sponsored retirement plan coverage following a job change. Using 24 years of household data, Rao found that workers who lose employer retirement plan coverage are significantly more likely to withdraw retirement savings and generally do not replace interrupted workplace savings with contributions to an IRA. Among workers with defined contribution retirement account balances, losing employer coverage increased the probability of cashing out retirement savings by 7.1 percentage points, approximately three times the pre-change cash-out rate. Notably, among workers who already held an IRA before changing jobs, the increase in cash-outs was even greater, at 14.4 percentage points. These findings suggest that simply having access to another tax-preferred retirement account is insufficient to ensure that workers preserve their savings, thus emphasizing the need to make rollovers easier for workers.
The long-term consequences of interrupted retirement savings can be substantial. For example, the paper includes a hypothetical demonstrating that a worker displaced at age 35 who experiences a seven-month employment interruption, lower wages, and four years without employer retirement plan coverage may save approximately 19 percent less in retirement savings by age 65 than a comparable worker without an interruption. If that worker also cashes out retirement savings upon separation from employment, the shortfall increases by 34 percent.
These findings reinforce the importance of preserving retirement savings when workers change jobs. While standardized rollover procedures cannot alone address all the challenges of losing employer retirement plan coverage, reducing the administrative burden of transferring a worker's savings is an important part of a broader strategy to prevent retirement savings leakage.
Support for Standardized Rollover Forms and Procedures
We appreciate the IRS's efforts to develop standardized forms and procedures to facilitate direct rollovers between eligible retirement savings accounts. In particular, we support efforts to standardize information requests, facilitate electronic transfers, improve communication between the distributing and receiving plans, and protect participants' personal information.
The proposed approach in Notice 2026-49 appropriately recognizes that retirement plan administrators and service providers are better positioned than individual participants to coordinate the technical and administrative aspects of transferring retirement assets. AARP believes the rollover burden on workers should be limited to deciding whether to roll over their savings and then requesting a rollover from their former employer's plan to their new employer's retirement plan. The balance of the rollover burden for such transfers should fall on the relevant service providers to coordinate the necessary steps for the transfer.
Behavioral economics has reshaped and improved the retirement savings ecosphere, so in that vein, we urge the IRS to focus on simplification and automation. When participants face burdens like contacting employers and completing multiple forms, among other things, these barriers reduce positive participant action, even if such actions would result in better outcomes for the worker. Accordingly, we urge the IRS to finalize the proposed standardized procedures and forms and consider additional guidance and safe harbors that would encourage widespread adoption by employers and retirement plan service providers.
Conclusion
For too long, workers have faced unnecessary hurdles when trying to preserve savings from a former employer. Notice 2026-49 is real progress toward a standardized process for moving retirement savings between plans. A streamlined approach would substantially reduce the burden placed on individual savers: the worker makes one request to the new plan, and the plans and their service providers handle the coordination. Savers carry far more of that burden under today's defined contribution system than workers did under traditional pension plans of the past. AARP appreciates the IRS's work on this guidance and welcomes the opportunity to discuss our comments further. If you have any questions, please feel free to contact me or have your staff contact Holly Biglow of AARP's Government Affairs team at hbiglow@aarp.org.
Sincerely,
Jennifer Jones
Vice President, Financial Security and Livable Communities
AARP
*
Original text of letter here: https://www.regulations.gov/comment/IRS-2026-0100-0013
5 Arizona State University Professors Urge Census Bureau to Count All Residents and Retain Demographic Data in Proposed Census Changes
Carter Struck
WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site.
October 13, 2026
Census Bureau, Department of Commerce
4600 Silver Hill Road, Designation: ADDEMO/FRN
Washington, DC 20746
RE: Request for Public Comment on the Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (RIN 0607-AA75)
Dear Daniel Sweeney:
The undersigned researchers are grateful to the Census Bureau for the opportunity to review ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. October 13, 2026 Census Bureau, Department of Commerce 4600 Silver Hill Road, Designation: ADDEMO/FRN Washington, DC 20746 RE: Request for Public Comment on the Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (RIN 0607-AA75) Dear Daniel Sweeney: The undersigned researchers are grateful to the Census Bureau for the opportunity to reviewand comment on the proposed changes to the Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (RIN 0607-AA75). We appreciate the Census Bureau's solicitation of public input and offer these recommendations with research, policy, practice, and the experiences of children, families, and communities in mind.
As researchers concerned with children, families, and equitable opportunity, we rely on accurate population data to understand communities' needs, examine access to resources, and evaluate whether public policies are working. The Census helps us plan for the people who live in our communities. It helps inform where schools are needed, where healthcare capacity must grow, and how communities prepare transportation, housing, and emergency services. Census data also support population estimates, other research, and funding formulas. These decisions shape whether families can find care, whether children have access to adequately resourced schools, and whether communities have the infrastructure to respond when needs change.
An incomplete count does not make those needs disappear. It makes them harder to see and easier to overlook. Accurate data help us make decisions that reflect actual lives. Accuracy also matters when interpreting outcomes. If the population used to calculate a rate is incomplete, our understanding of access, participation, and need can be distorted. Researchers and public officials need reliable information about who lives in a community to assess whether services are reaching people and where additional investment is warranted.
The proposed rule would exclude foreign citizens who are neither U.S. citizens nor lawful permanent residents from the apportionment count. We oppose this exclusion. The Fourteenth Amendment directs apportionment based on the "whole number of persons in each State," rather than a count limited to citizens and lawful permanent residents (U.S. Const. amend. XIV, Sec. 2). Immigrant children and families live in our neighborhoods, attend our schools, use our roads, and contribute to community life. Their presence creates real planning responsibilities regardless of immigration classification. Excluding people from apportionment also denies states recognition of their full resident populations when congressional representation is allocated.
The Department must distinguish between accurate enumeration and decisions about eligibility for particular programs. Counting a person does not establish that person's eligibility for a benefit. It establishes that the person lives here.
Research also raises concerns about the effects of citizenship inquiries on participation. An analysis of the 2019 Census Test found that a citizenship question increased undercount among households with noncitizens through reduced response, omissions of household members, and incomplete internet responses. These risks can affect an entire household, including children who are U.S. citizens. The Bureau should protect participation and count all residents for apportionment, without regard to immigration status.
Race and ethnicity data help us identify what broad averages conceal. We oppose the proposed prohibition on race and ethnicity questions in questionnaires used to enumerate the population. Race data matter because racism has shaped where people can live, the resources available in their neighborhoods, and their access to opportunity. Collecting race data allows researchers and communities to examine whether those patterns persist and whether efforts to address them are working.
A citywide average can suggest adequate access to a service while concealing substantial differences across communities. A school system can report overall progress while some children continue to experience unequal opportunities. Race and ethnicity data help us ask more precise questions: Who is being reached? Who is missing? Where are resources concentrated? Which communities continue to face barriers?
These data do not explain every disparity or establish its cause on their own. They help identify patterns that require further investigation, including examination of policies, institutional practices, geography, and family experiences. Without them, it becomes harder to distinguish broad improvement from improvement that leaves some communities behind.
For children and families, that distinction matters. We need to know whether opportunities are expanding for Black children, Indigenous children, and other children of color, rather than relying on an average that obscures their experiences. The proposal leaves demographic questions available on the American Community Survey and other surveys not used for enumeration. That does not resolve the loss of race and ethnicity information from the decennial enumeration. A sample survey serves a different purpose from a population count, particularly when communities need information about small populations and local areas.
We urge the Bureau to retain race and ethnicity questions and implement the 2024 OMB Standards. Those revisions followed research, public engagement, and federal expert review aimed at improving how federal data represent the population.
The proposed rule would connect usual residence to lawful presence and tax records and introduce an enumeration period running from January 3 through April 1. We are concerned that these changes could produce confusion and misclassification. Tax records may not reflect a family's current living arrangement. A parent may have moved since filing a return. A child may be staying with relatives. A family may be living in a motel after losing housing. A young person may have moved between foster placements. The Census must be able to account for these circumstances. Documentation should help establish where people live without becoming a barrier to recognizing their actual residence.
Counting a family in the wrong place can leave the community where it lives with an inaccurate picture of its population. The Bureau should withdraw these changes and retain residence criteria grounded in actual living arrangements. Any future revisions should undergo rigorous testing and consultation with affected communities.
Accurate data help public institutions respond before unmet needs become crises. They also help communities assess what happened after a policy or investment was implemented. If a government promises to expand access to care, families need a way to assess who gained access and who remains underserved. If a district invests in new facilities, communities need information to examine whether those investments reflect population needs. Reliable data provide a foundation for those questions.
We also oppose a categorical prohibition on future sexual orientation questions. Decisions about demographic questions should follow research, testing, demonstrated data needs, privacy protections, and meaningful engagement with affected communities. A permanent prohibition could foreclose that process and could limit future understanding of families and households.
We urge the Bureau to address prison gerrymandering by developing an approach that counts incarcerated people in their home communities. The proposed rule continues counting people, including young people in juvenile facilities, at the facility. That approach can shift populations away from home communities and into places where institutions are located. The Bureau should examine these consequences for representation and develop clear, tested procedures for identifying home residence.
We appreciate the extension of the comment deadline from October 13 to November 2, 2026. However, the scope of these changes warrants a longer opportunity for review. We urge the Department to extend the deadline until at least December 10, 2026, so researchers, state, local, and Tribal governments, civil rights organizations, community organizations, and affected families can evaluate the proposal and provide informed input.
We urge the Department and Census Bureau to:
1. Count all people residing in the United States for apportionment, without regard to immigration status.
2. Retain race and ethnicity questions in the decennial enumeration and implement the 2024 OMB standards.
3. Withdraw the proposed changes connecting residence to lawful presence and tax records and establishing a January 3-April 1 enumeration period.
4. Remove the categorical prohibition on future sexual orientation questions and preserve a research-informed process for evaluating demographic data needs.
5. Develop and test procedures to count incarcerated people in their home communities.
6. Extend the comment period until at least December 10, 2026.
Children and families do not disappear when data fail to capture them. They remain here, needing schools, care, housing, transportation, and opportunities to thrive. The Census should help us see those needs clearly and respond to them.
Thank you for the opportunity to provide public comment.
Sincerely,
Shantel Meek, Ph.D., Professor of Practice, Arizona State University (affiliation provided for identification purposes only)
Tunette Powell, Ph.D., Assistant Research Professor, Arizona State University (affiliation provided for identification purposes only)
Conor P. Williams, Ph.D., Professor of Practice, Arizona State University (affiliation provided for identification purposes only)
Xigrid Soto-Boykin., Assistant Research Professor, Arizona State University (affiliation provided for identification purposes only)
Brittany Alexander, Ph.D., Assistant Research Professor, Arizona State University (affiliation provided for identification purposes only)
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-69240
October 13, 2026
Census Bureau, Department of Commerce
4600 Silver Hill Road, Designation: ADDEMO/FRN
Washington, DC 20746
RE: Request for Public Comment on the Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (RIN 0607-AA75)
Dear Daniel Sweeney:
The undersigned researchers are grateful to the Census Bureau for the opportunity to review ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. October 13, 2026 Census Bureau, Department of Commerce 4600 Silver Hill Road, Designation: ADDEMO/FRN Washington, DC 20746 RE: Request for Public Comment on the Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (RIN 0607-AA75) Dear Daniel Sweeney: The undersigned researchers are grateful to the Census Bureau for the opportunity to reviewand comment on the proposed changes to the Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions (RIN 0607-AA75). We appreciate the Census Bureau's solicitation of public input and offer these recommendations with research, policy, practice, and the experiences of children, families, and communities in mind.
As researchers concerned with children, families, and equitable opportunity, we rely on accurate population data to understand communities' needs, examine access to resources, and evaluate whether public policies are working. The Census helps us plan for the people who live in our communities. It helps inform where schools are needed, where healthcare capacity must grow, and how communities prepare transportation, housing, and emergency services. Census data also support population estimates, other research, and funding formulas. These decisions shape whether families can find care, whether children have access to adequately resourced schools, and whether communities have the infrastructure to respond when needs change.
An incomplete count does not make those needs disappear. It makes them harder to see and easier to overlook. Accurate data help us make decisions that reflect actual lives. Accuracy also matters when interpreting outcomes. If the population used to calculate a rate is incomplete, our understanding of access, participation, and need can be distorted. Researchers and public officials need reliable information about who lives in a community to assess whether services are reaching people and where additional investment is warranted.
The proposed rule would exclude foreign citizens who are neither U.S. citizens nor lawful permanent residents from the apportionment count. We oppose this exclusion. The Fourteenth Amendment directs apportionment based on the "whole number of persons in each State," rather than a count limited to citizens and lawful permanent residents (U.S. Const. amend. XIV, Sec. 2). Immigrant children and families live in our neighborhoods, attend our schools, use our roads, and contribute to community life. Their presence creates real planning responsibilities regardless of immigration classification. Excluding people from apportionment also denies states recognition of their full resident populations when congressional representation is allocated.
The Department must distinguish between accurate enumeration and decisions about eligibility for particular programs. Counting a person does not establish that person's eligibility for a benefit. It establishes that the person lives here.
Research also raises concerns about the effects of citizenship inquiries on participation. An analysis of the 2019 Census Test found that a citizenship question increased undercount among households with noncitizens through reduced response, omissions of household members, and incomplete internet responses. These risks can affect an entire household, including children who are U.S. citizens. The Bureau should protect participation and count all residents for apportionment, without regard to immigration status.
Race and ethnicity data help us identify what broad averages conceal. We oppose the proposed prohibition on race and ethnicity questions in questionnaires used to enumerate the population. Race data matter because racism has shaped where people can live, the resources available in their neighborhoods, and their access to opportunity. Collecting race data allows researchers and communities to examine whether those patterns persist and whether efforts to address them are working.
A citywide average can suggest adequate access to a service while concealing substantial differences across communities. A school system can report overall progress while some children continue to experience unequal opportunities. Race and ethnicity data help us ask more precise questions: Who is being reached? Who is missing? Where are resources concentrated? Which communities continue to face barriers?
These data do not explain every disparity or establish its cause on their own. They help identify patterns that require further investigation, including examination of policies, institutional practices, geography, and family experiences. Without them, it becomes harder to distinguish broad improvement from improvement that leaves some communities behind.
For children and families, that distinction matters. We need to know whether opportunities are expanding for Black children, Indigenous children, and other children of color, rather than relying on an average that obscures their experiences. The proposal leaves demographic questions available on the American Community Survey and other surveys not used for enumeration. That does not resolve the loss of race and ethnicity information from the decennial enumeration. A sample survey serves a different purpose from a population count, particularly when communities need information about small populations and local areas.
We urge the Bureau to retain race and ethnicity questions and implement the 2024 OMB Standards. Those revisions followed research, public engagement, and federal expert review aimed at improving how federal data represent the population.
The proposed rule would connect usual residence to lawful presence and tax records and introduce an enumeration period running from January 3 through April 1. We are concerned that these changes could produce confusion and misclassification. Tax records may not reflect a family's current living arrangement. A parent may have moved since filing a return. A child may be staying with relatives. A family may be living in a motel after losing housing. A young person may have moved between foster placements. The Census must be able to account for these circumstances. Documentation should help establish where people live without becoming a barrier to recognizing their actual residence.
Counting a family in the wrong place can leave the community where it lives with an inaccurate picture of its population. The Bureau should withdraw these changes and retain residence criteria grounded in actual living arrangements. Any future revisions should undergo rigorous testing and consultation with affected communities.
Accurate data help public institutions respond before unmet needs become crises. They also help communities assess what happened after a policy or investment was implemented. If a government promises to expand access to care, families need a way to assess who gained access and who remains underserved. If a district invests in new facilities, communities need information to examine whether those investments reflect population needs. Reliable data provide a foundation for those questions.
We also oppose a categorical prohibition on future sexual orientation questions. Decisions about demographic questions should follow research, testing, demonstrated data needs, privacy protections, and meaningful engagement with affected communities. A permanent prohibition could foreclose that process and could limit future understanding of families and households.
We urge the Bureau to address prison gerrymandering by developing an approach that counts incarcerated people in their home communities. The proposed rule continues counting people, including young people in juvenile facilities, at the facility. That approach can shift populations away from home communities and into places where institutions are located. The Bureau should examine these consequences for representation and develop clear, tested procedures for identifying home residence.
We appreciate the extension of the comment deadline from October 13 to November 2, 2026. However, the scope of these changes warrants a longer opportunity for review. We urge the Department to extend the deadline until at least December 10, 2026, so researchers, state, local, and Tribal governments, civil rights organizations, community organizations, and affected families can evaluate the proposal and provide informed input.
We urge the Department and Census Bureau to:
1. Count all people residing in the United States for apportionment, without regard to immigration status.
2. Retain race and ethnicity questions in the decennial enumeration and implement the 2024 OMB standards.
3. Withdraw the proposed changes connecting residence to lawful presence and tax records and establishing a January 3-April 1 enumeration period.
4. Remove the categorical prohibition on future sexual orientation questions and preserve a research-informed process for evaluating demographic data needs.
5. Develop and test procedures to count incarcerated people in their home communities.
6. Extend the comment period until at least December 10, 2026.
Children and families do not disappear when data fail to capture them. They remain here, needing schools, care, housing, transportation, and opportunities to thrive. The Census should help us see those needs clearly and respond to them.
Thank you for the opportunity to provide public comment.
Sincerely,
Shantel Meek, Ph.D., Professor of Practice, Arizona State University (affiliation provided for identification purposes only)
Tunette Powell, Ph.D., Assistant Research Professor, Arizona State University (affiliation provided for identification purposes only)
Conor P. Williams, Ph.D., Professor of Practice, Arizona State University (affiliation provided for identification purposes only)
Xigrid Soto-Boykin., Assistant Research Professor, Arizona State University (affiliation provided for identification purposes only)
Brittany Alexander, Ph.D., Assistant Research Professor, Arizona State University (affiliation provided for identification purposes only)
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-69240
12 Arizona Lawmakers Condemn Proposed Census Changes Limiting Data Collection on Race Ethnicity and Immigration Status
Carter Struck
WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site.
1700 WEST WASHINGTON
PHOENIX, ARIZONA 85007-2844
CAPITOL PHONE: (602) 926-5284
TOLL FREE:1-800-352-8404
Arizona State Senate
1700 West Washington
Phoenix, Arizona 85007
October 13, 2026
The President
The White House
1600 Pennsylvania Avenue, Northwest
Washington, DC 20500
Dear President Trump,
We are writing to express our extreme concern at this administration's proposed changes to the decennial census short-form ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. 1700 WEST WASHINGTON PHOENIX, ARIZONA 85007-2844 CAPITOL PHONE: (602) 926-5284 TOLL FREE:1-800-352-8404 Arizona State Senate 1700 West Washington Phoenix, Arizona 85007 October 13, 2026 The President The White House 1600 Pennsylvania Avenue, Northwest Washington, DC 20500 Dear President Trump, We are writing to express our extreme concern at this administration's proposed changes to the decennial census short-formquestionnaire. As you are aware, on September 10, 2026, the Census Bureau proposed changing the short-form census questionnaire to only include lawful permanent residents and U.S. Citizens in the census count and to prohibit the inclusion of questions about race, ethnicity or sexual orientation. We find these changes to be not only historically irregular and legally questionable, but short-sighted and likely to have detrimental impacts on rural communities, tribal communities and communities of color both in Arizona and nationwide.
To rationalize the first change, the proposed rule alleges that those who are not lawful permanent residents are not "true inhabitants due to their lack of a sufficient tie and allegiance to the United States." This statement ignores the lived experience and true reality of this nation and is in direct violation of the 14th Amendment's Apportionment Clause. The contrived legal rationale this proposed rule uses as a defense does not change, nor does it reflect the clear text of the Constitution. The Apportionment Clause clearly states that "Representatives shall be apportioned among the several States according to their respective numbers, counting the whole number of persons in each State." The Fourteenth Amendment differentiates between persons and citizens; therefore, if its drafters intended only to count citizens, and not lawful residents or any other subset of those persons in the United States, they could have stated so. They explicitly and intentionally did not. This change would affect the population data used to determine each state's share of the House of Representatives and the Electoral College, giving increased weight to states where you think you have an electoral advantage. This attempt to cut out immigrants to our country who may not be legal permanent residents or citizens is antithetical to the ideals this country was built on.
To rationalize the second change, the proposed rule cites: 1) response rates decreasing with more questions; 2) privacy concerns; and 3) "substantial doubts about the meaning, significance and impact of attempting to categorize persons by race and ethnicity." To the first point, we find this concern questionable, considering that in earlier paragraphs of the same proposed rule, the Census Bureau contemplates including a new question "about legal status on the short-form decennial census questionnaire." To the second point, while we would typically share grave concerns about how this administration has proven to handle personally identifiable information, we do not believe race, ethnicity or sexual orientation fall into this category. Neither is it compelling enough in this case to necessitate creating a less informative census. To the third point, we fail to understand why the United States government would want to learn less about the makeup and diversity of its own population. The proposed rule itself concedes that "demographic questions have been asked in every census since 1790". We find zero compelling reason why we should throw out 240 years of successful precedent. Despite this administration's efforts to hide America's proud 250-year multicultural history, refusing to even account for our nation's demographics has repercussions far beyond the culture clash headline you are hoping to achieve. It would risk federal data used for enforcing what is left of the Voting Rights Act, the Civil Rights Act and other anti-discrimination enforcement. Of course, this is likely not an unintended consequence, but the goal of this change.
Your administration's proposed changes to the census short-form questionnaire would deprive our nation and government at every level of critical data needed to make informed economic and health decisions for each person within its boundaries. Local governments use this data to decide where to build hospitals, schools, roads and other critical infrastructure. Businesses use this data to decide where to expand their operations. Researchers use this data to study trends, from disease and mortality rates to poverty and unemployment. Creating less informed governments, businesses, researchers and nonprofits will inevitably hurt every American. With trust in government approaching historic lows, there is no doubt these changes will do nothing but further depress census response rates. As one of the states with the highest percentage of indigenous and native communities, we are particularly concerned about the impact it will have on these communities, which could lose out on critical representation and obligations from the federal government. An accurate census is paramount in ensuring these communities receive the political representation and funding that they need.
We do not need to remind you of your first administration's attempt to add a citizenship question to the 2020 U.S. Census. We in Arizona saw firsthand the chilling effects and repercussions of those actions, particularly in communities of color. Arizona lost a congressional seat and billions of dollars in critical federal funding due to this undercount. Rather than creating a more accurate census count, these proposed rules would certainly create confusion and chill response rates again. Every state and every community across the political spectrum deserves an accurate accounting of their demographics. We urge you not to proceed with these changes to the census short-form questionnaire.
Sincerely,
Minority Leader Priya Sundareshan
District 18
Assistant Minority Leader Catherine Miranda
District 11
Minority Whip Rosanna Gabaldon
District 21
Caucus Chair Lela Alston
District 5
Senator Theresa Hatathlie
District 6
Senator Lauren Kuby
District 8
Senator Kiana Sears
District 9
Senator Mitzi Epstein
District 12
Senator Eva Diaz
District 22
Senator Brian Fernandez
District 23
Senator Analise Ortiz
District 24
Senator Flavio Bravo
District 26
cc// U.S. Secretary of Commerce, Howard Lutnick
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-68023
1700 WEST WASHINGTON
PHOENIX, ARIZONA 85007-2844
CAPITOL PHONE: (602) 926-5284
TOLL FREE:1-800-352-8404
Arizona State Senate
1700 West Washington
Phoenix, Arizona 85007
October 13, 2026
The President
The White House
1600 Pennsylvania Avenue, Northwest
Washington, DC 20500
Dear President Trump,
We are writing to express our extreme concern at this administration's proposed changes to the decennial census short-form ... Show Full Article WASHINGTON, Oct. 25 -- In response to a federal agency request for information, the following public comment letter was submitted electronically to the regulations.gov site. 1700 WEST WASHINGTON PHOENIX, ARIZONA 85007-2844 CAPITOL PHONE: (602) 926-5284 TOLL FREE:1-800-352-8404 Arizona State Senate 1700 West Washington Phoenix, Arizona 85007 October 13, 2026 The President The White House 1600 Pennsylvania Avenue, Northwest Washington, DC 20500 Dear President Trump, We are writing to express our extreme concern at this administration's proposed changes to the decennial census short-formquestionnaire. As you are aware, on September 10, 2026, the Census Bureau proposed changing the short-form census questionnaire to only include lawful permanent residents and U.S. Citizens in the census count and to prohibit the inclusion of questions about race, ethnicity or sexual orientation. We find these changes to be not only historically irregular and legally questionable, but short-sighted and likely to have detrimental impacts on rural communities, tribal communities and communities of color both in Arizona and nationwide.
To rationalize the first change, the proposed rule alleges that those who are not lawful permanent residents are not "true inhabitants due to their lack of a sufficient tie and allegiance to the United States." This statement ignores the lived experience and true reality of this nation and is in direct violation of the 14th Amendment's Apportionment Clause. The contrived legal rationale this proposed rule uses as a defense does not change, nor does it reflect the clear text of the Constitution. The Apportionment Clause clearly states that "Representatives shall be apportioned among the several States according to their respective numbers, counting the whole number of persons in each State." The Fourteenth Amendment differentiates between persons and citizens; therefore, if its drafters intended only to count citizens, and not lawful residents or any other subset of those persons in the United States, they could have stated so. They explicitly and intentionally did not. This change would affect the population data used to determine each state's share of the House of Representatives and the Electoral College, giving increased weight to states where you think you have an electoral advantage. This attempt to cut out immigrants to our country who may not be legal permanent residents or citizens is antithetical to the ideals this country was built on.
To rationalize the second change, the proposed rule cites: 1) response rates decreasing with more questions; 2) privacy concerns; and 3) "substantial doubts about the meaning, significance and impact of attempting to categorize persons by race and ethnicity." To the first point, we find this concern questionable, considering that in earlier paragraphs of the same proposed rule, the Census Bureau contemplates including a new question "about legal status on the short-form decennial census questionnaire." To the second point, while we would typically share grave concerns about how this administration has proven to handle personally identifiable information, we do not believe race, ethnicity or sexual orientation fall into this category. Neither is it compelling enough in this case to necessitate creating a less informative census. To the third point, we fail to understand why the United States government would want to learn less about the makeup and diversity of its own population. The proposed rule itself concedes that "demographic questions have been asked in every census since 1790". We find zero compelling reason why we should throw out 240 years of successful precedent. Despite this administration's efforts to hide America's proud 250-year multicultural history, refusing to even account for our nation's demographics has repercussions far beyond the culture clash headline you are hoping to achieve. It would risk federal data used for enforcing what is left of the Voting Rights Act, the Civil Rights Act and other anti-discrimination enforcement. Of course, this is likely not an unintended consequence, but the goal of this change.
Your administration's proposed changes to the census short-form questionnaire would deprive our nation and government at every level of critical data needed to make informed economic and health decisions for each person within its boundaries. Local governments use this data to decide where to build hospitals, schools, roads and other critical infrastructure. Businesses use this data to decide where to expand their operations. Researchers use this data to study trends, from disease and mortality rates to poverty and unemployment. Creating less informed governments, businesses, researchers and nonprofits will inevitably hurt every American. With trust in government approaching historic lows, there is no doubt these changes will do nothing but further depress census response rates. As one of the states with the highest percentage of indigenous and native communities, we are particularly concerned about the impact it will have on these communities, which could lose out on critical representation and obligations from the federal government. An accurate census is paramount in ensuring these communities receive the political representation and funding that they need.
We do not need to remind you of your first administration's attempt to add a citizenship question to the 2020 U.S. Census. We in Arizona saw firsthand the chilling effects and repercussions of those actions, particularly in communities of color. Arizona lost a congressional seat and billions of dollars in critical federal funding due to this undercount. Rather than creating a more accurate census count, these proposed rules would certainly create confusion and chill response rates again. Every state and every community across the political spectrum deserves an accurate accounting of their demographics. We urge you not to proceed with these changes to the census short-form questionnaire.
Sincerely,
Minority Leader Priya Sundareshan
District 18
Assistant Minority Leader Catherine Miranda
District 11
Minority Whip Rosanna Gabaldon
District 21
Caucus Chair Lela Alston
District 5
Senator Theresa Hatathlie
District 6
Senator Lauren Kuby
District 8
Senator Kiana Sears
District 9
Senator Mitzi Epstein
District 12
Senator Eva Diaz
District 22
Senator Brian Fernandez
District 23
Senator Analise Ortiz
District 24
Senator Flavio Bravo
District 26
cc// U.S. Secretary of Commerce, Howard Lutnick
*
Original text of letter here: https://www.regulations.gov/comment/USBC-2026-0628-68023