Federal Regulators Reportedly Drafting "Assisted Arithmetic Transparency Act" Requiring Diners to Disclose Calculator-Assisted Gratuity

Washington, D.C. — Federal regulators have reportedly proposed a new rule requiring diners to disclose when a calculator was used to determine gratuity, formalizing what officials describe as a long-overdue extension of computational transparency principles into the domain of restaurant arithmetic.

The policy, drafted jointly by the Bureau of Consumer Disclosure, the Department of Treasury’s Office of Honest Numbers, and a Department of Education advisory body created earlier this year specifically to study the cognitive consequences of widespread calculator adoption, mirrors recent AI disclosure requirements that have been working their way through the federal regulatory apparatus. The new framework, titled the Assisted Arithmetic Transparency Act, treats calculator-assisted gratuity computation as a category of machine-assisted economic decision-making and accordingly subjects it to the disclosure obligations that have accumulated, in recent years, around any human activity demonstrably enhanced by automation.

At a press briefing held in a Treasury Department auditorium notable for the absence of any visible calculators on the lectern, an Assistant Undersecretary for Consumer Disclosure characterized the Act as the natural conclusion of regulatory principles that have governed digital labor since the introduction of mandatory AI authorship footnoting. “The country has decided, as a matter of policy, that work performed with computational assistance must be marked as such,” the official said. “We see no principled basis on which to exempt tip calculation from that decision.”

Industry response was immediate and largely procedural. Within ninety-six hours of the proposed rule’s publication in the Federal Register, the National Restaurant Association had submitted a request for clarification regarding whether the disclosure obligation falls on the diner, the restaurant, the calculator manufacturer, or some combination of the three. The Bureau responded that the question was, in its current formulation, “productively ambiguous.”

Regulatory Lineage

The Assisted Arithmetic Transparency Act does not arrive in a vacuum. It is the latest entry in a sequence of federal disclosure rules whose common feature is the application of AI-era authorship principles to domains of routine cognition previously considered too modest for regulatory attention. Bureau communications describe this sequence as the “Authentic Cognition Initiative,” though the phrase appears only in internal materials and has not been formally adopted by any of the agencies participating in the work.

The first explicit precedent, in regulatory terms, is the 2024 Calculator Manufacturer Transparency Protocol, which obligated educational institutions and credentialing bodies to document the use of calculators in any academic or professional work product. The protocol, which generated significant initial controversy and modest subsequent compliance, established the structural template the present rule extends: a disclosure obligation triggered by the use of an electronic device, a tiered classification of the assistance involved, and a self-attestation enforcement regime backed by the threat of episodic federal audit.

The second precedent, finalized in late 2025, was the Department of Justice’s autocorrect disclosure guidance, which extended the same principle to written communication, requiring Americans to footnote the predictive-text and spellcheck interventions that had, by that point, become invisible accompaniments to ordinary digital correspondence. The autocorrect guidance produced what Bureau researchers describe as “the foundational dataset” for the present rule: a body of consumer-response data demonstrating that disclosure obligations of this general kind produce, in initial pilots, high rates of confusion followed by gradual procedural normalization.

The third precedent, less commonly cited but more conceptually proximate, is a 2025 Bureau working paper titled Toward a Unified Framework for the Disclosure of Machine-Assisted Routine Cognition, which proposed that any human activity demonstrably enhanced by computational assistance — from grocery shopping to driving directions to the splitting of restaurant bills among friends — was, in principle, eligible for disclosure regulation once an appropriate administrative surface could be constructed for it. The working paper attracted relatively little public attention at the time of its release. The Assisted Arithmetic Transparency Act represents, in the view of several legal scholars, the first concrete implementation of the framework the working paper anticipated.

The Rule, in Its Current Form

Under the proposed framework, all itemized receipts issued by establishments operating within federal jurisdiction must include a clearly marked checkbox bearing the standardized disclosure language:

“Tip amount calculated with computational assistance.”

The checkbox, regulators have specified, must measure no less than 0.847 centimeters in width, must be printed in ink of contrast sufficient to remain visible under standard restaurant lighting conditions, and must be positioned within the diner’s natural visual sweep when reviewing the receipt. A Bureau working paper accompanying the rule includes a six-page appendix on the ergonomics of receipt scanning, which concludes that diners overwhelmingly read receipts in a downward-and-rightward motion described by the report as “the L-shaped trajectory of post-meal accountability.”

The disclosure obligation, as written, applies to any tip whose amount has been determined, in whole or in significant part, with the assistance of an electronic device capable of performing arithmetic operations. The rule’s drafters acknowledge that this definition encompasses a broad range of devices and arrangements. The accompanying interpretive guidance specifies that the following activities, at minimum, trigger the disclosure requirement:

• Use of a smartphone calculator application, including the native applications installed by default on iOS and Android devices.
• Use of a smartwatch calculator function, including dictated voice queries to a wrist-worn device that produces an arithmetic answer.
• Use of a dedicated handheld calculator, whether scientific, graphing, or basic four-function, regardless of whether the device was manufactured before 1995.
• Use of a spreadsheet application, including but not limited to Microsoft Excel, Google Sheets, Apple Numbers, and any open-source equivalent.
• Selection of a suggested tip percentage offered by a payment processing terminal, including the 18%, 20%, and 22% buttons that have become standard in most point-of-sale systems.
• “Mental math enhanced by prior calculator exposure,” a category defined in the guidance as “arithmetic operations completed without a device but informed by the subject’s familiarity with the underlying computation as a result of repeated prior calculator use.”

The final item has generated considerable commentary, including from constitutional scholars who have raised questions about whether the federal government may, as a matter of administrative law, regulate cognitive processes that occur entirely within a private citizen’s head. The Bureau’s response, included in a supplementary clarification, states that the regulation does not concern the cognitive process itself but rather the disclosure of its computational provenance, a distinction the agency characterizes as “analytically clean and procedurally manageable.”

The Rationale, as Articulated

Officials defending the rule have offered three principal justifications, which together constitute what the Bureau’s public-facing summary describes as “the integrated case for arithmetic honesty.”

The first justification concerns the dignity of the recipient. Servers, regulators argue, deserve to know whether the tip they have received was the product of unaided cognitive effort or the output of a device. Internal Treasury memoranda obtained by this publication describe this consideration in language drawn directly from the agency’s broader work on AI authorship disclosure: a tip arrived at through mental computation is characterized as “an authentic gesture of appreciation,” while a tip arrived at through calculator use is described as “a computationally mediated transaction whose interpretive valence is altered by the intervention of the device.”

The second justification concerns the broader public interest in arithmetic transparency. Treasury officials have argued that the federal government’s recent push to require disclosure of computational assistance in writing, image generation, code production, and decision-making would be analytically incomplete if it did not extend to the most numerically consequential decision the average American makes on a regular basis, which Bureau studies identify as the determination of restaurant gratuity.

The third justification, articulated more cautiously, concerns what one official described as “the long-term educational implications of a calculator-mediated tipping culture.” Department of Education researchers attached to the rule-making process have submitted findings suggesting that the percentage of American adults capable of computing 18% of $84.70 without electronic assistance has declined from an estimated 67% in 1984 to an estimated 12% in 2024, a trend the researchers characterize as “consistent with the broader hypothesis that unaccountable computational assistance produces irreversible atrophy in foundational arithmetic capacity.”

At the press briefing, the Assistant Undersecretary distilled the justification into a single sentence:

“Authenticity matters in arithmetic.”

The line, which the official delivered without apparent humor, has since been reproduced on posters, stickers, and at least one Bureau-issued lapel pin, copies of which were distributed to attending journalists in a small velvet pouch.

The Office of Honest Numbers

The administrative entity charged with primary rulemaking responsibility under the Act is the Treasury Department’s Office of Honest Numbers, a sub-bureau established in early 2025 whose mandate, as articulated in its founding charter, is “to preserve the integrity of arithmetic in the public square.” The Office has, in its eighteen months of existence, attracted a small but devoted policy staff drawn primarily from the intersection of mathematics education, behavioral economics, and consumer protection law. Its current director, a former community college mathematics professor who has requested anonymity in connection with this article, has characterized the Office’s work as “the unglamorous middle of a problem the country prefers to discuss at its more dramatic extremes.”

The Office’s methodology, in developing the present rule, drew on what its working documents describe as “a triangulated evidentiary base.” The base includes:

• A longitudinal survey of 8,470 American adults conducted over the eighteen months preceding the rule’s publication, in which respondents were asked, among other questions, to report the methods by which they typically determined restaurant gratuities.
• A behavioral study, conducted in collaboration with researchers at three undisclosed universities, in which 247 volunteers were observed determining tips under conditions of varying time pressure and varying access to electronic devices.
• A textual analysis of approximately 184,000 publicly available restaurant reviews, in which the Office’s research staff sought references to tipping practices and identified patterns of computational assistance described, however indirectly, by the reviewers.
• A series of structured interviews with 84 servers, conducted across twelve metropolitan markets, in which the servers were asked to describe what information about their tips, if any, they would find professionally valuable.

The fourth element of the evidentiary base — the structured interviews with servers — has attracted particular scrutiny, in part because the Bureau’s summary of the findings does not include the question that several servers, on follow-up, indicated they had attempted to answer. The interviews, as originally designed, asked servers what information about their tips would be useful to them. The servers, according to internal transcripts, did not converge on the cognitive provenance of the tip as an answer of any consequence. Several servers asked whether the question was, in fact, the question being asked, and indicated some confusion about why the federal government was interested in this particular topic.

The Office’s public summary of the interview findings characterizes the server response as “mixed but constructive.” The full transcripts, which were obtained by this publication through a public records request and reviewed in their entirety, do not, on the editorial board’s assessment, sustain that characterization.

A Leaked Internal Memo

A draft memo circulated within the Office of Honest Numbers approximately three weeks before the rule’s publication, and obtained by this publication from a source within Treasury, addresses the question of whether the disclosure framework should be calibrated to the actual preferences of the population it nominally serves. The memo, written by a senior policy analyst whose name has been redacted, concludes that such calibration would, in the memo’s phrase, “collapse the rule into nothing.”

“If we structure the disclosure regime to address only those forms of computational assistance that meaningfully change the tip recipient’s welfare, we will produce no regulation, because the tip recipient’s welfare is, on available evidence, indifferent to the cognitive provenance of the gratuity. If we structure the disclosure regime to address the public interest in arithmetic transparency more broadly, we produce a rule that is formally robust but operationally untethered to any measurable outcome. The Office’s institutional judgment is that the second option is, on balance, the option more consistent with the broader regulatory tendency this Office is charged with maintaining.”

The memo’s circulation within the Office did not, according to interviewed staff, generate meaningful internal opposition. Several analysts who were copied on the memo described it, in retrospect, as “an unusually frank articulation of a position the staff understood the Office to hold,” and characterized the memo’s subsequent treatment as “institutionally typical.”

The Tiered Disclosure Framework

Anticipating the complexity of enforcement, the Bureau has proposed a tiered classification system intended to capture the spectrum of computational assistance available to the modern diner. The framework, drawn in part from the disclosure typology already adopted for calculator-assisted education work under a separate rule finalized in late 2024, distinguishes five categories of arithmetic provenance.

Tier I: Unassisted Mental Computation

Defined as arithmetic performed entirely within the diner’s head, without recourse to any external device and without reference to the suggested tip percentages displayed on the payment terminal. The Bureau acknowledges that Tier I activity is, in principle, indistinguishable from Tier V activity (defined below) without the cooperation of the diner, and the rule accordingly relies on a system of self-attestation in which the diner indicates, in good faith, that no computational assistance was used. The guidance notes that self-attestation regimes have functioned adequately in adjacent regulatory contexts and may be expected to function comparably here.

Tier II: Suggested-Percentage Selection

Defined as the selection of one of the pre-computed tip amounts displayed on the payment terminal, including the now-standard 18%, 20%, and 22% buttons. The Bureau’s interpretive guidance characterizes Tier II activity as “passive computational consumption,” noting that the diner has, in selecting a suggested percentage, accepted the output of a device without independently verifying its accuracy. The accompanying methodology document includes a twelve-page discussion of whether Tier II activity is more or less computationally honest than Tier III activity, concluding that the question is “philosophically interesting but operationally moot.”

Tier III: Active Device-Assisted Computation

Defined as the use of a calculator application, watch function, or dedicated handheld device to independently compute the tip amount. Tier III activity is the framework’s central concern, constituting what the Bureau estimates is roughly 73% of all disclosed computational assistance in initial pilot data. Tier III diners are required to indicate not only that computational assistance was used but, in cases where the receipt allows for it, the device class employed (smartphone, smartwatch, dedicated calculator, or other).

Tier IV: Spreadsheet-Mediated Computation

Defined as the use of any spreadsheet application to compute the tip, including cases in which the diner has constructed a personal spreadsheet for the purpose of routine gratuity calculation. Bureau researchers have indicated that the population of Tier IV diners is small but “analytically distinctive,” and that the category has been included primarily to ensure typological completeness. One Bureau working paper notes, in a footnote, that the existence of Tier IV diners was confirmed only after focus group research revealed three participants who maintained tip-calculation spreadsheets, two of whom were software engineers and one of whom was “simply, and we say this with respect, that kind of person.”

Tier V: Calculator-Influenced Mental Computation

Defined, in the guidance, as “mental arithmetic performed without a device in the immediate sense, but enabled by the cognitive scaffolding produced by prior, sustained calculator exposure.” Tier V is the most contested category in the framework. The Bureau’s position is that any diner whose ability to compute 18% of an arbitrary dinner bill has been developed or maintained through regular calculator use is, in effect, performing calculator-assisted arithmetic, with the calculator’s contribution merely time-shifted rather than absent. Constitutional scholars and most calculator manufacturers have disagreed.

The Bureau’s public summary characterizes the Tier V category as “philosophically ambitious but enforcement-light,” conceding that the federal government does not currently possess a mechanism for distinguishing Tier I from Tier V diners and that the category may therefore function primarily as a moral framework rather than an operational classification. A senior Bureau official, asked at the press briefing whether the agency anticipated meaningful enforcement of the Tier V disclosure requirement, replied that the question was “not the most productive framing.”

Dr. Henry Gutenberg, in an interview given to this publication shortly after the rule’s publication, offered an early one-line assessment of the Tier V category that has since been widely quoted:

“Tier V is the part of the rule where the federal government, having run out of behavior to regulate, has begun regulating residue.”

An Economic Critique

Within forty-eight hours of the rule’s publication, Dr. Henry Gutenberg, Senior Fellow at the Port-au-Prince Institute for Market Dysfunction, had published a widely circulated analysis situating the Assisted Arithmetic Transparency Act within what he has previously characterized as the broader regulatory tendency toward disclosure as moral substitute. The Bureau, Gutenberg argued, was responding not to a documented problem but to a documented anxiety, and the rule accordingly produced no measurable improvement in any outcome of interest to any party affected by it.

“What is at stake here is not the dignity of the server, who is broadly indifferent to the cognitive history of the tip and is interested principally in its magnitude. What is at stake is the regulator’s need to demonstrate that something is being done about an unease the regulator cannot fully articulate. The unease is real. It concerns the gradual displacement of unaided human capability by quiet, ubiquitous machine assistance in domains the culture has not yet decided how to feel about. The Assisted Arithmetic Transparency Act resolves the unease not by addressing it but by creating an administrative surface on which the unease can be inscribed. The checkbox is the unease, given a place to sit.”

Dr. Henry Gutenberg

— Dr. Henry Gutenberg, Port-au-Prince Institute for Market Dysfunction

Gutenberg’s working paper, titled Disclosure as Catharsis: On the Regulatory Function of the Pointless Checkbox, has been cited approvingly by economists across several ideological positions, including a number who do not ordinarily cite Gutenberg approvingly. The paper’s central argument is that disclosure regimes which impose costs on regulated parties without producing actionable information for any recipient are not, on inspection, regulatory failures but rather regulatory successes of a particular kind — successes measured not in the units the regulation purports to optimize but in the units the regulator actually cares about, which Gutenberg identifies as “visible administrative motion in the direction of acknowledged public concern.”

The Bureau has not formally responded to Gutenberg’s analysis, though a senior official, speaking on background, characterized the working paper as “the kind of critique we anticipated and accept as a feature of operating in the open.”

Public Reaction

Initial consumer response to the proposed rule has been characterized, in the Bureau’s own focus group data, as “procedurally bewildered.” A representative cross-section of approximately 2,847 diners, interviewed across twelve metropolitan markets, expressed confusion in proportions exceeding any other measured response category, including support, opposition, and indifference.

One diner, identified in the transcript only as Subject 412, said:

“I just didn’t want to mess up 18%.”

The Bureau’s focus group methodology document treats this response as paradigmatic and reproduces it, with permission, on the cover page of the agency’s consumer outreach materials. The interior of the materials includes the agency’s official response to Subject 412’s articulated concern:

“Intent does not eliminate disclosure.”

The principle — that the diner’s motivation in using a calculator is irrelevant to the disclosure obligation triggered by its use — has been reiterated across Bureau communications and is, the agency notes, “structurally identical” to the principles already governing AI authorship and autocorrect disclosure under separate but parallel rules.

Selected Focus Group Responses

The Bureau’s public summary of the focus group findings includes a representative sample of consumer responses, reproduced here in full:

“If I have to disclose that I used the calculator, can I also disclose that I’m bad at math? Because that’s the actual story here.”
— Subject 089, age 41, marketing professional

“I thought the suggested tip buttons were the point. That’s what they’re for. Now I’m supposed to feel bad for pressing them?”
— Subject 217, age 33, paralegal

“Look. I’m fifty-eight years old. I’ve been tipping for forty years. The math hasn’t changed. The math is exactly the same as it was. I haven’t changed. The government has changed, but the math hasn’t.”
— Subject 731, age 58, retired

“I asked my server what she thought, and she said she’d rather not know how I got the number. She said it was nicer that way.”
— Subject 1,104, age 29, software engineer

The final response — in which the diner reports that the recipient of the tip expressed a preference for not knowing the tip’s computational provenance — has been the subject of internal Bureau discussion. A working memo describes the response as “directionally important but not dispositive,” noting that the public interest in arithmetic transparency is not contingent on the recipient’s expressed preference for ignorance.

The Server’s Perspective

Industry advocacy groups representing tipped workers have responded to the rule with what one union official described as “a degree of polite confusion that has surprised even us.” Restaurant Workers United, in a statement issued the morning after the rule’s publication, observed that the population of servers actively interested in the cognitive provenance of their tips was, in available survey data, vanishingly small.

The statement noted further that servers ranked “knowing whether the tip was calculator-assisted” as the 47th most important workplace concern out of 50 surveyed, ahead only of “the temperature of the wait station,” “the music selection in the dining room,” and “whether the menus are laminated or not.”

Servers interviewed for this article expressed a range of reactions, none of which fell easily into the framework the rule’s drafters appeared to anticipate.

“Honestly? I don’t care how they got there. I care that they got there.”
— Server, 31, Chicago bistro

“If anything, I’d rather know they used a calculator. That means they checked. The ones who do it in their heads — that’s where the math gets creative.”
— Server, 44, Houston steakhouse

“The government wants me to know that the tip was calculated by phone. I want the government to know that I haven’t been able to afford rent in two years. There is, I suspect, a mismatch in our priorities.”
— Server, 27, Brooklyn diner

The last comment, which has circulated widely on professional networks frequented by tipped workers, has been excluded from the Bureau’s public summary of stakeholder feedback on grounds that it does not, in the Bureau’s judgment, “respond to the policy as proposed.”

International Comparison

The Assisted Arithmetic Transparency Act has been situated, in comparative legal scholarship, within a small but growing class of disclosure regimes that extend AI-era authorship principles into domains of routine cognition. No directly comparable rule exists in any other jurisdiction at the time of writing, though several adjacent frameworks have been cited as conceptual relatives.

The European Union’s General Data Protection Regulation, in its consumer-disclosure provisions, requires the documentation of automated decision-making in a range of commercial contexts; restaurant tipping has not, to date, been considered an instance of automated decision-making subject to GDPR provisions. The Bureau’s interpretive guidance includes a twelve-page comparative analysis of why the Assisted Arithmetic Transparency Act’s approach exceeds GDPR scope, concluding that the U.S. framework “reflects a more comprehensive understanding of the cognitive provenance of consumer behavior than current European frameworks have yet adopted.”

In Japan, where service charges are typically incorporated into the menu price and gratuity is a marginal cultural practice, regulators have followed the U.S. rule’s development with what one Tokyo-based researcher described as “polite anthropological interest.” The researcher, asked whether any analogous framework was being contemplated under Japanese law, replied that the underlying premise of the rule — that the cognitive provenance of a tip is a matter of public concern requiring federal disclosure — was “not, in our legal culture, an obviously tractable proposition.”

In France, where service is included by law and tipping is a discretionary supplementary gesture, a senior official at the Ministry of Economy responded to a query from this publication with what was, on careful reading, a series of declarative sentences that did not directly address the question and that may, in the official’s view, have constituted a sufficient response.

In Haiti, where Dr. Gutenberg’s home institution has maintained, for some years, a running commentary on the regulatory pathologies of higher-income jurisdictions, the rule has been received with the muted recognition Gutenberg’s working papers have prepared the Institute to extend. A junior researcher at the Institute, in a short response paper published within days of the rule’s announcement, observed that the rule “does not regulate tipping; it regulates the American relationship to its own competence at arithmetic.” The paper notes that the question of whether this distinction matters is “the substantive question the Bureau has declined to engage.”

Academic Reception

Within the academic community, the rule has generated a literature whose volume, in proportion to the rule’s operational stakes, is best described as disproportionate. By the editorial board’s count, at least eighteen working papers, four journal articles, and one forthcoming edited volume now address some aspect of the Assisted Arithmetic Transparency Act. The literature is concentrated in three principal disciplinary clusters.

Economists, broadly, have approached the rule as an instance of regulatory design in which the cost-benefit calculus is unusually transparent and unusually unfavorable. The principal cost identified is the compliance burden imposed on restaurants and payment processors; the principal benefit identified is the production of administratively legible data whose downstream utility, even in the rule’s own framing, is unclear. Economists who have previously analyzed disclosure regimes in adjacent domains — in finance, in pharmaceuticals, in food labeling — have noted that the present rule departs from those precedents in one important respect: in finance, in pharmaceuticals, and in food labeling, the recipient of the disclosure is presumed to act on it. In the present rule, the recipient is the Bureau’s quarterly statistical report, which does not act on it at all.

Legal scholars have addressed the rule’s constitutional dimensions, with particular attention to the Tier V category and the underlying question of whether the federal government may regulate the cognitive history of arithmetic operations conducted within a private citizen’s head. Two law review articles have, by the time of writing, argued that the Tier V framework cannot survive even minimal constitutional scrutiny; one article has argued that the framework, properly understood, raises no constitutional issue because it does not in fact regulate cognition; and a fourth article has argued, more idiosyncratically, that the Tier V framework is constitutional because the federal government’s authority to regulate cognitive provenance is implicit in its authority to regulate computational provenance, the two categories being “structurally identical at the level of information” under the framework’s own internal logic.

Behavioral scientists have approached the rule from the perspective of consumer compliance, and have produced what is, at present, the most empirically grounded body of scholarship on the rule’s likely operation. Preliminary experimental data, drawn from pilot deployments in several restaurant chains, suggests that the proportion of diners who select the “decline to state” option exceeds the proportion who select any of the three substantive options combined. The behavioral scholars have characterized this finding as “the regulation’s most predictable outcome and its most analytically clarifying.”

Restaurant Industry Compliance

Restaurants, for their part, have moved quickly to develop receipt systems capable of accommodating the disclosure requirement. Major point-of-sale vendors — Toast, Square, Clover, and Lightspeed — have all announced firmware updates expected to roll out within the rule’s 180-day compliance window. The updates introduce the mandated checkbox to the printed receipt, along with a digital equivalent on the touch-screen payment interface, where the diner is now asked, after selecting a tip amount, to indicate the method by which the amount was determined.

The interface design has attracted commentary. In its current form, the touch-screen prompt appears immediately after the diner has selected a tip percentage, and presents four options:

• “I calculated this in my head.”
• “I used the suggested percentage.”
• “I used a calculator or other device.”
• “Decline to state.”

Bureau guidance specifies that the “decline to state” option must be available but does not, in itself, satisfy the disclosure obligation; restaurants are required to maintain records of the proportion of diners selecting each option and to report these proportions to the Bureau on a quarterly basis. The aggregate national data will, the Bureau anticipates, be published in an annual report titled The State of Arithmetic Honesty in American Dining, the inaugural edition of which is scheduled for release in the second quarter of the following year.

Several major restaurant chains have indicated, in private communications, that they intend to configure the “decline to state” option as the default selection, on the grounds that this configuration minimizes the time the average diner spends interacting with the payment terminal and accordingly accelerates table turnover. The Bureau has not formally objected to this configuration but has issued an interpretive note stating that “widespread reliance on the decline-to-state option may, over time, attract additional regulatory attention.”

The Long Division Incident

On the morning the rule was published, a chain of casual dining restaurants in suburban Maryland experienced what employees later described as “a small civic episode” involving diners attempting to comply with the disclosure requirement in advance of its formal effective date. Several diners, having read coverage of the rule, declined the suggested-percentage buttons on the payment terminal and asked their servers for pen and paper.

The servers, having no immediate use for pens and paper in this context, were obliged to source them from the manager’s office. The diners then proceeded to attempt long division by hand on the backs of the receipts.

Of the seventeen diners observed engaging in this activity over the course of the lunch service, manager observation reports indicate that:

• Three completed the calculation and arrived at the correct tip amount within a reasonable timeframe.
• Six completed the calculation and arrived at an answer that was, in the words of one manager, “in the neighborhood but not at the address.”
• Five abandoned the attempt and reverted to the suggested-percentage button, checking the “computational assistance” box with what one server described as “visible defeat.”
• Two requested the assistance of the server in completing the calculation, an arrangement that the Bureau’s guidance has not yet addressed and that may, depending on interpretation, constitute a separate category of computational assistance not currently contemplated by the rule.
• One wept quietly for approximately four minutes and then left a flat $20 regardless of the bill’s amount.

The Maryland incident has been cited, in subsequent commentary, as an illustration of what Department of Education researchers had attempted to convey in their submissions to the rule-making process. A senior researcher, asked about the incident, said that the population of American adults capable of performing two-digit-by-two-digit long division under time pressure is “considerably smaller than the population that believes itself capable of doing so,” and that the gap between perceived and actual arithmetic ability “is the empirical foundation on which the rule rests.”

Appendix: Excerpted Focus Group Transcript

The following transcript excerpt, reproduced with redactions, is drawn from a Bureau-commissioned focus group conducted in Columbus, Ohio, in the months preceding the rule’s publication. The participants — eight diners drawn from a stratified sample of metropolitan-area restaurant-goers — were asked to respond to a draft version of the disclosure framework and to articulate, in their own words, what concerns or questions the framework raised. The moderator’s prompts have been preserved in italics.

Moderator: I’d like to start by asking each of you to describe the method you most commonly use to determine the tip you leave at a restaurant.

Participant A (age 34, accountant): “I press the 20% button. It’s right there.”

Participant B (age 52, contractor): “I do it in my head. Move the decimal one place, double it. That’s 20%. Knock off a little if the service was bad.”

Participant C (age 28, graphic designer): “I use the calculator on my phone. I’m not good at math. I’ve never been good at math. I’m not going to start being good at math at a restaurant.”

Participant D (age 67, retired teacher): “I write it on the receipt. Long form. I show my work.”

Participant E (age 41, attorney): “Whatever the suggested amount is. I don’t think about it.”

Participant F (age 29, nurse): “My partner does it. I just pay.”

Participant G (age 38, software engineer): “I have a shortcut on my phone that calculates 22% of any number I type in. I built it during a slow afternoon.”

Participant H (age 55, sales manager): “Twenty bucks. Always. Doesn’t matter what the bill is. Twenty bucks. Fight me.”

Moderator: The proposed rule would require you to disclose, on the receipt, whether you used computational assistance to determine the tip. What is your initial reaction to that requirement?

Participant A: “Wait. The receipt I sign or the receipt they keep?”

Participant B: “Why does anyone care how I got the number?”

Participant C: “Am I supposed to feel ashamed of using the calculator? Because I don’t. I’ve done a lot of things I should feel ashamed of. That’s not on the list.”

Participant D: “I think it’s probably fine. I always do my work by hand. So I can just check the box that says I didn’t use the calculator.”

Participant E: “Is this an actual rule or are you asking me to imagine an actual rule? Because I can’t tell from how you’re asking the question.”

Participant F: “What if my partner used the calculator and I’m the one paying? What box do I check?”

Participant G: “Does the shortcut count? It’s code I wrote. It’s not really a calculator. It’s a function I authored.”

Participant H: “I leave twenty bucks. There’s no math. There’s no calculator. There’s twenty bucks. What box does that go in?”

Moderator: The rule’s drafters have stated that the disclosure is intended to allow servers to know whether the tip was computed organically or with computational assistance. Do you think the server you most recently encountered would have wanted to know this information?

[Silence, approximately seven seconds.]

Participant B: “Are you serious?”

Participant C: “I’m sorry — the server wants to know?”

Participant E: “In what world does the server care about that?”

Participant H: “The server wants twenty bucks. That’s what the server wants. That’s the whole transaction.”

Participant D: “I think it might be a nice thing to share, actually. A little acknowledgment. ‘I did this in my head, just for you.’ It’s a kindness.”

Participant G: “You said the rule’s drafters have stated that. Have they actually stated that?”

[Moderator declines to answer.]

The transcript, in its full unredacted form, runs to approximately 84 pages. The Bureau’s public summary of the focus group findings characterizes the participant response as “a productive mix of support, concern, and constructive critique.” The editorial board notes, without further comment, that the transcript does not, on its own terms, sustain this characterization.

Implementation Timeline

The proposed rule, as currently drafted, contemplates a 180-day comment period followed by a 90-day implementation grace period during which restaurants and payment processors are expected to complete compliance preparations. The Bureau has indicated that initial enforcement will be “educational rather than punitive,” with formal civil penalties not expected to be assessed before the rule’s second year of operation.

The structure of those penalties, as outlined in the rule’s preamble, is itself the subject of ongoing discussion. The current proposal contemplates a tiered penalty schedule in which the responsible party — the diner, the restaurant, or the payment processor, depending on the nature of the violation — is subject to a civil fine of between $84.70 and $847 per documented instance of non-disclosure. The penalty’s relationship to a specific dinner bill of $84.70, on which the rule’s drafters have apparently anchored the schedule, has not been formally explained.

Industry observers have indicated that the penalty schedule, in its current form, will be challenged in litigation almost immediately upon its formal adoption. At least three restaurant industry trade groups have indicated their intention to file suit; one consumer advocacy organization has indicated that it will join the litigation on behalf of diners; and the calculator manufacturers, as a class, have indicated that they regard the rule as ambiguously written in ways that may, depending on enforcement patterns, expose them to liability not contemplated by the rule’s plain language.

A Bureau spokesperson, asked about the prospect of immediate litigation, indicated that the Office of Honest Numbers was “prepared for and welcomes a robust judicial dialogue on the rule’s constitutional dimensions.” The spokesperson declined to specify what outcome of that dialogue the Office considered most likely or most desirable.

A Second Critique

Returning to the policy in a follow-up commentary published several weeks after his initial analysis, Dr. Gutenberg expanded his critique to address what he characterized as the rule’s most consequential feature: the manner in which its enforcement architecture relies, almost entirely, on consumer self-attestation in a domain where the consumer has no incentive to attest honestly and the regulator has no mechanism to verify the attestation.

“Consider what the rule asks of the diner. It asks the diner, having completed a meal and having determined a tip, to pause at the moment of greatest cognitive depletion in the entire transactional encounter and to make a candid statement about the method by which the tip was determined. The candidness of this statement is unverifiable. The diner’s incentive to be candid is negligible. The recipient of the candidness, if there is one, is the Bureau’s quarterly statistical report. The rule does not regulate behavior. It regulates the production of administratively legible information about behavior, and the information is, by the rule’s own structural design, of low evidentiary quality. The rule is, in this sense, a sophisticated piece of regulatory theater, and the sophistication is what concerns me. The unsophisticated regulatory theater is easy to identify and easy to dismiss. The sophisticated kind is harder to dismiss, and it accumulates.”

Dr. Henry Gutenberg

— Dr. Henry Gutenberg, Port-au-Prince Institute for Market Dysfunction

Gutenberg’s follow-up paper closes with an observation that has been quoted in subsequent commentary across the ideological spectrum: that the Assisted Arithmetic Transparency Act represents not the federal government’s attempt to address computational displacement of human capability but rather its formal acknowledgment that computational displacement has already occurred and is no longer recoverable, with the rule serving as a kind of bureaucratic memorial to a competence the culture has decided to mourn through paperwork rather than through education.

“We are not preserving arithmetic. We are filing its obituary.”

The Bureau, asked to respond to the obituary characterization, declined comment.

Closing Statement

Restaurants are preparing updated receipts. Point-of-sale vendors are rolling out compliance firmware. The Department of Treasury has begun printing copies of The State of Arithmetic Honesty in American Dining in advance of next year’s inaugural release, on the assumption that demand will be at least modest. Department of Education researchers continue to gather evidence in support of the rule’s educational rationale, and have indicated that subsequent rulemakings may extend the disclosure framework to additional categories of routine arithmetic, including but not limited to grocery price comparison, mortgage amortization estimation, and the splitting of bills among groups of friends.

At press time, several diners in suburban Maryland were attempting long division by hand.

Poorly.

Editorial Footnotes