The questions put to this project, with the answers it can defend and the ones it cannot. Split from the front page on 21 August 2026. The argument behind these answers is the case.
Questions this project is asked
Who this is about
About: the officers of frontier developers — the companies training models above 10²⁶ operations, or spending nine figures on a single training run. On the enacted family’s own thresholds that is a double-digit number of firms worldwide, and inside them a smaller number of people who decide what ships.
Not about: open-source contributors, startups, academic researchers, hospitals, schools, employers, small operators, ordinary deployers and API customers, or users. On deployers, precisely: the Act has always defined deployer as a covered class (SEC. 1(b)(3)) and has always given the thin ones a route to discharge the duty rather than an exemption from it — adopt the upstream validation, keep the manifest, monitor, report (SEC. 2(b)). What the open queue would add at Amendment 7 reaches a deployer only at consequential scale — mass market, or into government, military, financial, health or critical-infrastructure functions — and only for its own deployment decisions, never as the developer of a model it did not train. A company using a commercial model through an API is not covered by that fact.
And the claim, stated precisely. Not that no American law reaches a natural person over AI — it does, readily; Nebraska’s “operator” includes one, so a sole trader running a chatbot is personally inside that statute. What no American law does is place a duty on the officer of a covered frontier developer for the decision to release. The law reaches down, not up.
Grouped by who tends to ask them, and laddered — beginning where most people actually begin (“does this affect me?”) and climbing to doctrinal design. This is the harder of the project’s two question pages: the objections that professionals, and lately the answer engines, circulate — answered at full strength, name-free and wealth-free by rule. The plain-language set, the questions the public actually asks, angry ones included, is the dossier’s Q&A chapter; several questions appear on both pages on purpose, answered in each page’s register. The maximal case against this text is the hostile brief. Several answers below are honestly “open” — naming them is what a research draft is for.
Asked first.
Will my job or my role be affected? For almost everyone, no. The Act is aimed at the small number of natural persons—chief executives, controlling owners, and others—who possess final material authority to prevent, halt, restrict, or correct the deployment of a covered system. They cannot delegate that duty to a safety officer or leave accountability with the company: each qualifying person answers personally for failure to exercise due care, whether or not actual knowledge can be proved. Title, technical ability, access, advice, or merely implementing another person’s decision does not make an ordinary employee a controlling person.
I use ChatGPT, Claude, or similar at work or school — does this touch me? No. Users are nowhere in the Act. Personal and non-commercial use is expressly carved out, and nothing here limits what you run on your own machine.
Would this slow down or ban AI? It does not ban AI as a field or regulate ordinary AI use. It conditions the deployment, expansion, release, and operation of covered systems on compliance with defined duties. SEC. 2(c) provides a controlled-research pathway, while lawful use, study, and modification of lawfully obtained weights remain protected. Compliance may take time, and the Act permits injunction or suspension of particular unlawful configurations—including emergency suspension upon probable cause of imminent risk of death or serious injury. The power to stop covered conduct when necessary is part of the accountability mechanism, not a general prohibition on AI.
My startup builds with AI—are we covered? Not merely because it is a startup. Coverage turns on the model, the system, and the conduct—not company size. Training a covered model or materially expanding or deploying a covered system can trigger duties; a non-modifying deployer receives the reliance rule in SEC. 2(b). The lower [10^22]-operations threshold concerns specified records, not full frontier-model coverage. Personal liability still requires final material decision authority under SEC. 4.
If this passed, what would actually change? Natural persons with final material authority over covered systems would have to exercise due care and personally sign the required certifications. Company liability would remain, but it would no longer substitute for personal accountability or absorb an individual’s penalty. Serious incidents would be reported on fixed deadlines, and insiders who report violations would receive statutory protection and awards. Ordinary users and workers without the required decision authority would acquire no personal compliance duty.
Asked by everyone.
“You cannot put a god on a leash.” Agreed — which is why the Act never tries: it never regulates the mathematics at all. The objection arrives in exactly those words, and the field notes keep it beside its answer: deification is the last liability- evaporation structure. The duty attaches to the human who chooses to ship the system, and humans have never been gods, only occasionally defendants.
Why not just pause? A pause one officer can reverse tomorrow is not a law — it is the case for one. A pause is welcome, and the evidence file records one — announced by a single officer, voluntarily, reversibly, subject to no external standard and no consequence on resumption. That is the difference between restraint and law, and it is the entire argument: the Act is what the same undertaking looks like when it binds.
“Safety law is how censorship arrives — this will make models preachier, or worse.” The Act contains no content rules at all: not one duty in it tells a model what to say, refuse, or believe. SEC. 0(a)(4) forbids the reading outright — no provision requires any person to express, adopt, endorse, or refrain from expressing any opinion, characterization, viewpoint, or contested position, or to alter the output of any covered system — and SEC. 3(a) closes the standards topic to safety, authorization, monitoring, incident-reporting, and deployment controls, full stop. The deeper answer runs the other way: today’s refusals are discretionary — over-applied where refusal is cheap, under-applied where it is expensive, answerable to no one — and a duty regime ends the ambiguity by drawing the floor at process: validation, records, incidents, a signature. Everything above that floor becomes visibly a product choice. A statute that never touches outputs cannot censor them; what it ends is “safety” as the alibi.
What about China? Food-safety law was never suspended because other countries had laxer kitchens. Product-safety duties are not a capability race: the duty protects the people the product reaches, in-state, whoever else builds what. Nothing in the Act slows research; it attaches consequences to shipping unvalidated systems at people.
How does this proposal compare with officer-liability regimes in other jurisdictions? PRC Criminal Law art. 31 provides that where an entity commits a crime the entity is fined and the persons directly in charge and other directly responsible persons are subject to criminal punishment. That general dual-punishment rule has been in force since the 1997 revision. This Act is materially narrower: it operates only within its compute- and conduct-defined frontier-AI scope, and its individual duties reach only natural persons who satisfy the controlling-person test. Germany uses a different mechanism, described below, to attach supervisory responsibility to identified natural persons. (Comparative source note.)
“Won’t this hand the lead to someone else?” America has run this experiment: it met a ten-nation safety embargo with inspection law, and exports rose by two-thirds. From 1879 Germany and nine other European states closed their markets to American pork on trichinosis grounds; meat exports fell from about $134 million in 1881 to about $69 million the following year. Congress answered with inspection — the Act of 30 August 1890 for salted pork and bacon intended for export, then the Act of 3 March 1891 covering live cattle and hogs and requiring certificates. Within the year the President was reporting the result to Congress: comparing May through August 1892 with the same months of 1891, pork-product exports were up 62 per cent by weight and about 66 per cent by value, and live cattle exports had gone from 205,786 head in 1889 to 394,607. Honesty requires the other half: inspection alone did not reopen Germany, which lifted in 1891 under a threatened retaliatory tariff on its sugar beets. Verifiable safety was not sufficient. It was necessary — the thing the diplomacy had to have in hand before it could be spent, and the thing that kept the market open once it was. (The provisions, including the sugar beets.)
“No other country does this to its own industry.” Several do, and the American frontier sector is closer to an exception than a victim. Germany reaches the individual through § 130 OWiG, which makes a business owner’s failure of required supervision a punishable omission where that failure enabled a contravention — § 9 OWiG extends the same exposure to managing directors and board members, and the fine runs to €1 million where the underlying breach is itself criminal. That is weaker than this Act, and it is candidly an administrative offense rather than a crime, because Germany has no general corporate criminal code; the point is only that the duty attaches to a named person rather than dissolving into the firm. China’s art. 31 does the same thing through actual criminal punishment. What is unusual is not that duties reach an identified officer. It is that in one American industry they currently do not. (The provisions.)
“I’m not American. Why does this matter to me?” Because the systems reach you and the law that governs them does not. A handful of frontier models are shipped worldwide by a handful of companies, and the decision to ship one is made in a room in one country, by people no other country can reach. This Act claims nothing beyond its own state: it regulates conduct in or into that state, evenhandedly, and its findings say so. But the responsible corporate officer doctrine does not attach liability by where harm lands or by what anyone intended — it attaches to the person standing in responsible relation to the hazard, with the authority to prevent it. There is one such moment for a frontier system, and one set of hands on it, regardless of how many countries the weights subsequently reach. So the protection travels with the product rather than being asserted over anyone. That is the ordinary pattern for safety standards fastened to a procedure instead of a border: the surgical checklist trialled across eight hospitals on four continents worked in each, and spread because it was attached to the operation. One state is enough not because one state governs the world, but because the artifact is the same everywhere it lands.
Was AI used to draft this project? Yes. AI assistance is disclosed throughout the repository. The maintainer remains responsible for the text, sources, corrections, and decisions about publication. The proposal assigns legal duties to natural persons, not to models or software systems.
Who maintains the project? One person, writing under the project’s name rather than their own — masked in the oldest American tradition, Publius and Junius and Dickinson, because the citations are the only authority this text claims. Check them, not the byline. The maintainer is identifiable privately to reviewers before they sign anything, and to retained counsel at engagement. No client, organization, or grant funds the work. Provenance and method describes the drafting and correction process.
Is this advocacy or introducible legislation? It is model legislation and a research draft. Any sponsor’s legislative counsel would need to conform its definitions, penalties, procedure, and cross-references to the adopting jurisdiction. The draft states elements, burdens, defenses, and severability rules so that reviewers can assess it as proposed statutory text. Known errors and corrections are recorded in the register.
Who funds this? No client, organization, or grant funds the project. The drafting record, revisions, and corrections are public, and the text is released under CC0.
“Where are the bodies?” In the incident timeline, which exists because the question deserves a file rather than a sentence. The older answer is structural: the offense is shipping unvalidated, precisely so the count stays where it is — prevention that works is always accused of solving nothing.
Asked from the left.
This is frontier fetishism — the harms happening now sit below your compute line. Guilty as scoped, and the scope is the point: this is one statute about one gap — no person yet answers personally for the largest systems — not a theory of all AI harm. Sub-frontier injuries have live statutes, agencies, and plaintiffs, and SEC. 0(a)(7) keeps every one of them in force. A bill that tried to be the whole of AI justice would be the omnibus that dies in committee, taking the gap down with it.
Prison as a theory of change is still carceral politics. The Act’s center of gravity is not the cell: the working instruments are the signature, the disgorgement, the disqualification, and the debarment — consequences that reach the person without filling a prison — and the custodial tiers exist because their absence has a market price. As for the doctrine’s ugly history: eighty years of public-welfare liability were spent almost entirely on warehouse managers and pharmacists. The novelty being objected to is not the aim; it is the direction.
Asked by engineers and researchers.
Am I personally liable? No — title, credentials, technical ability, access, and the ministerial execution of another’s decision are excluded from authority in black letter (SEC. 4). The Act climbs the organization chart; it does not descend it.
Does this touch open weights, my fine-tune, or my local model? Your local model, your fine-tune, and your study or modification of lawfully obtained weights are expressly untouched. The one duty sits on releasing frontier weights above the line, and it is the same validation duty deployment behind an interface carries — parity, not penalty, on the EU’s own systemic-risk logic; sub-threshold derivatives sit outside coverage by default.
Is the modification-evaluation budget the right size? It is bracketed precisely so that evaluations researchers, not drafters, set it. The default — the greater of one percent of lineage compute or 10^24 operations — is open, and invited to be kicked.
The standards are paywalled — isn’t that a problem? Yes — and the Act’s rule is absolute: no standard binds in this State unless it is free to read (SEC. 3). A standard incorporated by reference into a compliance regime is law you must pay to read, and law you must pay to read fails the oldest requirement in the rule-of-law tradition: rules must be promulgated before they can bind. The courts have fought this exact fight over building codes and technical standards — model codes enacted into law enter the public domain as law (Veeck v. Southern Building Code Congress International, 5th Cir. 2002, en banc), the government-edicts doctrine keeps official legal text uncopyrightable (Georgia v. Public.Resource.Org, 2020), and posting standards that the law incorporates has been held fair use (ASTM v. Public.Resource.Org, D.C. Cir. 2023). This Act sides with that line. And the register applies the rule to ourselves first: the interim standards are borrowed verbatim, dated and chapter-cited, from three enacted state frameworks anyone can read — and now pinned in the repository itself (the adopted texts) — but the regulations draft referenced one paywalled industry standard, which conflicted with the Act’s own free-access rule. Defect owned; cured at v3.4: the control objectives are restated in the regulations’ own words, and the standard is not incorporated. An accountability regime whose rulebook sits behind an invoice would be the thing this project exists to end.
“Nobody will take the safety-officer job if it carries prison exposure.” Liability follows authority, not title: an officer without power to halt is not a controlling person under SEC. 4, and one with that power is exactly who the duty exists for. The signature converts the role from ornamental to load-bearing — a hiring upgrade, not a deterrent, for anyone serious. The precedent held: CFOs did not vanish after Sarbanes-Oxley; the job stopped being decorative.
Asked by legislators and staff.
Why would a committee consider independently drafted model legislation? A sponsor would receive a package with a named maintainer, disclosed drafting methods and conflicts, a provenance page, and any completed expert reviews. Legislative counsel would then conform the proposal to the jurisdiction’s code and drafting rules.
Nobody is asking for this bill — where is the constituency? Conceded, on the record — and the Model Penal Code had the same audience: no crowd, just reporters, a decade, and one legislature at a time. A model act’s audience is measured in sponsors, and the current count is zero. The claimed path is the same and no shorter: named reviewers, then a sponsor’s counsel, then one state.
Does the proposal criminalize innovation? Research and ordinary AI use are not offenses, and professional credentials, technical ability, access, advice, or implementation alone do not establish controlling-person status. The base individual offense requires final material decision authority, a relevant duty, and failure to exercise due care; higher tiers require additional elements. Entity liability remains, but the natural persons with practical power cannot place all accountability inside the corporate shell.
What does it cost my state? Not yet quantified. The principal cost drivers are standards development, secure intake of certifications and reports, whistleblower processing, records-related investigations, and Attorney General enforcement. SEC. 3(b)’s bar on preapproval removes licensing, application-review, and denial-appeal costs, but it does not eliminate administrative workload or possible backlogs. Volume depends on covered entities and conduct, deployment and incident frequency, and the adopting state’s jurisdiction and existing capacity. The fiscal note identifies the required low-, central-, and high-volume analysis, keeps startup separate from steady state, and never books penalties as revenue.
Why a state, and not Congress? Because Congress has written nothing and one state is enough — the biometric-privacy precedent proved that a law which exists first becomes the standard. Criminal law over in-state harm is the states’ oldest power and the hardest for Washington to reach.
Industry will ask for its Price-Anderson — a liability cap in exchange for the duty. Then industry should be offered the whole bargain it cites: Price-Anderson paired its cap with strict, channeled liability, mandatory insurance to the cap, and pervasive oversight. The cap was the price of accepting liability by default, not an escape from it. A cap without the channeling is dessert without the dinner, and the Act declines to serve it.
Why not add a right-to-cure before charges? Because for this offense the cure period already exists, and it sits before deployment: validation is the cure. A post-harm cure window converts the shipping offense into a free first bite at the public — an amendment met in the wild and preserved in the field notes under exactly the name it earned: the cheapest gut.
Any agency you create will be captured. The warning is taken as a design constraint rather than rebutted: the Act minimizes what capture can switch off. Duties commence on standards already enacted elsewhere; no approval mode exists to sit on; if the agency never organizes, filings run to the attorney general. An agency that cannot gate the duties is an agency not worth capturing.
“A hostile attorney general will weaponise this against companies whose politics he dislikes.” This Act hands a partisan less to work with than almost any criminal statute on the books, because the offenses are documentary and the yardsticks are published. A charge must prove its elements to a jury beyond reasonable doubt — including the person’s practical power, whose genuine absence defeats the case as a failed element, never as something the accused must establish — and due care is measured against standards that are enacted, dated, and free to read, not against an official’s taste. The text carries no viewpoint term to aim through: SEC. 0(a)(4) forbids compelled opinions and output alteration on its face. What remains is charging discretion, which is the residue of every criminal law ever written; this one at least confines it to paper — records kept or destroyed, reports filed or buried, statements true or false. The instrument a weaponiser needs is a statute that bends toward its holder, and this one was drafted, in public, to bend toward its elements.
With approval forbidden (SEC. 3(b)), how does anyone buy certainty? By filing, not by queueing: validation is documentary, self-executed against published standards, and complete the day it is transmitted — the seatbelt model, not the drug-approval model. What SEC. 3(b) forbids is the waiting room, which is where delay and capture both live.
“Pass this and the labs simply leave your state.” The duty attaches to deployment in or into the State — to the market, not the mailing address. A firm that leaves the state’s patients, customers, and hospitals has not evaded the Act; it has exited the state. Three states have already enacted frontier duties, and the census of departed laboratories stands at zero. Conduct is taxed where the harm lands — the one variable relocation cannot move.
“Criminal exposure here just moves the labs overseas — you’d be handing the frontier to looser jurisdictions.” The duty never asks where the developer sleeps: it attaches to deployment in or into the State, the way long-arm public-welfare law has worked since goods first crossed borders. Incorporate anywhere; the moment the system reaches this State’s users, the duties reach the officers. The exit that actually sheds liability is called leaving the American market, and no frontier firm built on American capital is taking it — the same firms that predicted flight over the GDPR and the EU AI Act complied with both and stayed. The Act regulates at the point of harm precisely because the mailing address is the most mobile fact about a company.
Asked by lawyers.
“Hasn’t the Supreme Court been moving away from strict criminal liability?” Yes — and the Act is drafted inside that line, not against it. For three decades the Court has presumed scienter where penalties are severe: Staples v. United States, 511 U.S. 600 (1994); Rehaif v. United States, 588 U.S. 225 (2019); Ruan v. United States, 597 U.S. 450 (2022). Measured against that presumption the Act passes on its face: SEC. 6(c) forbids any custodial sentence without proof of a due-care failure — negligence, not strict liability, is the floor for jail; the felony tier of SEC. 6(b) requires knowing or willful conduct; strict liability survives only where the modern Court tolerates it, in the entity’s civil penalty under SEC. 10(a). And on Park’s own escape hatch: Park preserved a defense of objective powerlessness, 421 U.S. at 673. SEC. 6(d)–(e) does not merely adopt it — it upgrades it. Absence of practical power negates an element the prosecution must prove beyond a reasonable doubt; it is not an affirmative defense the accused must carry. On this point the Act gives a defendant more process than Park did, on purpose. A statute built to survive review should say so on its face; this one now does.
Isn’t this the gun-manufacturer fallacy — blaming the maker for the user? No, twice over: the firearm shield is a specific statutory choice, not the default of products law, and a covered system is regulated here for what it does itself — autonomous action reaching into a stranger’s systems — not for what a user does with it. Park never asked what the product intended; it asked who had the power to stop the shipment.
Why criminal law at all, rather than civil liability or regulation? Because both already exist and neither reaches the person: civil penalties are paid from the treasury, and a fine a firm can pay is a price. The one instrument with an eighty-year record of changing officer behavior is personal criminal exposure — the Park line — which Congress wrote into the Clean Water Act, and which has never reached software.
Prison on strict liability — is that constitutional? The Act does not attempt it: fines may follow the classic strict public-welfare pattern, but imprisonment requires proven fault — a negligence floor, codifying the constitutional line the DeCoster concurrence drew. What it deliberately omits is a good-faith defense, following the 1948 congressional refusal. Whether a strict-liability misdemeanor tier should sit beneath the felony is a genuine design question — open, and squarely within the criminal-law reviewer.
Park was a $250 misdemeanor; you attach life imprisonment to a negligence floor. Different floors, kept apart on the face of the text: the due-care offense of SEC. 6(a) is the Park misdemeanor — one year at most (SEC. 10(b)). Life enters only through SEC. 6(b)(1), knowing or willful conduct, plus a death charged as but-for and proximate cause under Burrage, per victim, to a jury — the mens rea and structure every state already uses between manslaughter and murder. The one bracketed novelty is the [two]-year minimum, held open for the criminal-law reviewer.
“The behaviour is emergent — you’re criminalising outcomes nobody could foresee, without mens rea.” The Act criminalizes no outcome standing alone. The floor offense is failing to run the required checks — process, not prophecy — and a certification under SEC. 8 attests that validation ran, never that the future is safe. The harm tier demands more, not less: knowing or willful conduct plus but-for and proximate causation under Burrage, charged per victim to a jury. Unpredictability is the reason the duties exist; it has never been a defense to skipping the pre-flight.
What about deferred and non-prosecution agreements? Deliberately silent — charging discretion stays where state law puts it. Whether negotiated reviews should be cabined for individual liability — the settlement culture is much of why entity-level enforcement stopped deterring — is open, and belongs to the enforcement reviewer.
No attorney general will bring the first case — SEC. 4 is unprovable against a governance chart built by three firms. Year one is filing cabinets by design: the offenses that commence immediately are documentary — records, reports, statements, certifications — and documentary cases are the ones offices actually bring. Thereafter the signature does the element’s work: the person who certified under SEC. 8 is the person who claimed the authority, in writing. The FDCA comparison also cuts backward — Park prosecutions are rare because the exposure works ex ante. Whether SEC. 4 survives a real charging memo is the enforcement reviewer’s core question — open, and the reviewer exists because it is.
“Your records duties turn emergency patching into evidence tampering.” Backward on the face of the text: remediation is a duty, and the offense is failing to establish, maintain, or preserve required records — the manifests, validations, and logs — never the act of fixing the system they describe. Patch the model; keep the log of the patch. Destroying the record of remediation is the crime. Remediation never was.
Doesn’t banning insurance and indemnification destroy market discipline? Insurance against a personal penalty converts the one non-priceable consequence back into a price — the failure mode the Act exists to end. The counter-argument is respected, insurers can be private regulators; but the ban follows an enacted pattern, applies prospectively with a conforming window, and carves restitution out so victims are never the ones disciplined. The economics remain fair council terrain.
Won’t personal liability teach firms to stop looking for their own problems? The Act is drafted so that silence, not candor, is the dangerous strategy: reporting clocks run from when certified monitoring would have detected an incident, so not looking starts the clock anyway; the records offenses punish destruction and falsification, not disclosure; and disclosing nonconformity is a protected filing that simply does not count as validation. The deeper incentive-design question is real, and open.
The regulated class is small, and this repository also contains a wealth dossier. Is this targeted legislation? The concern is serious. The Act names no person and determines no guilt. Coverage turns on prospective rules—compute or designation, covered conduct, and practical authority—and every element of a criminal offense must be proved in court beyond a reasonable doubt. People enter or leave the regulated class as the relevant facts change. The dossier is neither incorporated into nor cited by the statutory text and should be evaluated separately as project research.
You rewrote three states’ statutes by stripping their thresholds — that isn’t incorporation, it’s ventriloquism. No sister state is amended or spoken for: the borrowed texts operate only as this State’s own interim standards, frozen as of a stated date, so no other legislature drafts for this one prospectively. What is stripped are the enacting states’ fiscal choices about their own machinery — subjects SEC. 3(c)(4)(D) reserves to this Act — not their safety judgments. States borrow enacted text constantly, from tax codes to building codes; the only novelty here is doing it with a date and a checksum.
“Deployed in or into this State” is extraterritorial — dormant Commerce Clause. The Act regulates the subject that doctrine protects rather than threatens: harm to persons in-state, evenhandedly. SEC. 0(a)(5) is operative, not decorative — no in-state advantage, no out-of-state burden an in-state actor doing the same conduct escapes — and the findings disclaim any duty arising from research, training, or development that neither occurs in-state nor concerns a system sent there (SEC. 0(a)(3)). The sibling formulas are enacted law in the interim standards’ home states; the full map is chunk 2.
Why won’t federal preemption kill it? It may try; the armor is operative text, not cover copy — SEC. 0 states the core, SEC. 13 orders the severance and revives suspended provisions if a federal switch-off later lapses. State criminal law over conduct harming people in-state is the last thing preemption reaches, which is why the core was built there.
Added 23 August 2026, for a First Amendment reader. The self-designation route (Amendment 6) treats a developer’s own published “frontier” designation as evidence of a jurisdictional fact. Compelled disclosure of purely factual, uncontroversial commercial information receives Zauderer review; compelled adoption of contested positions receives strict scrutiny (NIFLA; X Corp v. Bonta, 116 F.4th 888 (9th Cir. 2024)). The bounded question: does a statute that takes the company’s own marketing designation as a rebuttable jurisdictional fact — compelling nothing, adopting nothing, requiring no publication — implicate the compelled-speech doctrine at all, and if so, on which side of Zauderer does a self-chosen commercial designation fall? The Act’s design facts for whoever takes it up: SEC. 0(a)(4); SEC. 8’s statements-of-fact limit; SEC. 9(c)’s no-characterization rule; and Amendment 4, which removes the statute’s last characterization-shaped trigger.
None of the hardest questions above began as hypotheticals. Objections arriving in the wild are preserved, with the answers that survived them, in the field notes — the leash, the gun analogy, the Price-Anderson bargain, and the cheapest gut among them — and every objection met is kept, sourced, in the question-and-answer chapter. The first genuine outside catch is credited permanently as E7 in the register. Criticism is raw material here; send more.
Corrections to the project contact; they enter the corrections register with the fix attached and permanent credit.