The Rulebook
Most publications ask you to trust them. This one hands you the rulebook. Every instruction the AI follows to write the news, every value that decides what gets covered, every line the publication will not cross in its own voice — all of it is written down, published, and yours to read before you read a single article. You can check the reporting against the rules, argue with a rule, or take the whole apparatus and run it yourself.
Main Street Independent has two sides, and the documents below cover both. The general newsfeed is written to a strict factual floor and carries no perspective of its own. The signed columns argue openly, each from a named voice with a declared point of view. Keeping those two things separate — and saying exactly how — is most of what this page is about.
Start here: what news is for
If you read one thing, read this. What News Is For is the whole argument in one place: why journalistic standards already encode human values, why pretending otherwise produces worse reporting, what the five-value floor is, where the line between reporting and analysis falls, and how the whole machine is built. Eight sections, with appendices that carry the scholarship.
The treatise's central claim is the one most newsrooms won't say out loud: there is no such thing as value-free reporting, and chasing it makes journalism worse, not better. Every editorial choice — what to cover, whom to quote, which fact leads and which gets buried — is already a judgment about what matters. The old "view from nowhere" doesn't remove those judgments; it just hides them, which is how a publication can smuggle in a worldview while claiming to have none. Main Street Independent does the opposite. It names its five values at the top and builds the reporting on them in the open. The treatise walks from that argument, through the floor itself, to the exact seam between neutral reporting and signed analysis, and closes with the scholarship the whole structure rests on — chiefly the idea, from the modern standards tradition, that journalism's first obligation is to the truth, its working method is a discipline of verification, and its loyalty is to citizens rather than to any faction.
→ What News Is For: The Editorial Foundation of Main Street Independent
The five values the newsfeed runs on
The general newsfeed operates from five values, and only five: human life and dignity, truthfulness, accountability of power, equality and fairness, and informed citizenship. The floor specification is the working version — for each value, what it covers, and the exact boundary between what the publication will state in its own voice and what it will not.
The word consensus is load-bearing. These five are chosen precisely because they are broad enough that they belong to no political faction — not a platform anyone runs on, but the floor beneath every platform. And the document's real work is in the boundaries. For each value it draws a hard line between a claim the newsfeed can assert in its own voice and a claim it cannot, no matter how true a writer believes it to be. A statement that stays inside the floor can run under the Editorial Board byline as news; a statement that reaches past it has to move to a signed column, where a named voice owns it. That single mechanism — floor inside, perspective outside — is what keeps the newsfeed honest, and the specification is written as an operating rule the framework actually consults, not an aspiration framed on the wall.
The commitments the editor consults
A floor is only as good as the judgment that applies it. The Editorial Mind is that judgment, written down: the four constitutional commitments — truth, harmlessness, fairness, witness — that the news framework consults every time it decides whether a sentence can stand in the publication's own voice or has to move to a signed column. Think of it as the editor who never sleeps, whose standing instructions are public.
Each of the four does specific work. Truth demands that every concrete claim trace to a source, that hedges survive the summarizing, and that no quotation is ever invented. Harmlessness is the guard around the vulnerable — the minor, the named private individual, the suspect not yet charged. Fairness requires the same scrutiny of everyone, applied without regard to which side a subject is on. Witness is the commitment to engage the documented reality of an event rather than the spin layered over it. These are held at the highest weight in the system, above any lower operating preference, and the framework runs every candidate sentence past all four. A sentence that satisfies them stands as news; one that can't is either rewritten, moved to a column where a named voice can take responsibility for it, or cut. Because the commitments are published, you can do what readers almost never can: read the editor's standing orders, then judge for yourself whether the article in front of you actually obeyed them.
Exactly what the AI is told to write
The newsfeed is written by AI working from a published, auditable specification. Reading it shows precisely what the model is instructed to do: trace every specific to a source, keep every hedge, never invent a quotation, and stay inside what the floor authorizes. This is the difference between this and an AI content farm — the instructions are public, and you can hold the output to them.
Before the generator writes a single word, the story has to clear a verification bar. The standard is at least two independent reporting sources, or one primary document — a court filing, a government release, a peer-reviewed paper, an on-the-record statement — and every named person, place, and organization must resolve to a stable public record, not a vague "a senator" or "officials." Material older than seventy-two hours loses its standing as news and has to be framed as history instead. Then the writing itself is governed: lead with what matters, attribute every claim, default to the plain attribution verb (said, never admitted or insisted unless the record earns it), prefer the active voice, and reach for no superlative the sources don't support. And the spec hard-blocks certain inputs outright — a single-source rumor, an anonymous-only claim with no corroboration, an outlet that repeatedly fails fact-checks, the naming of a minor or a sexual-assault victim, the identification of a suspect before charges are filed. Every one of these rules is the operating form of a standard that wire services and standards desks already enforce; the difference is that here you can read the rule and hold the article to it.
What the AI does — and does not — do
The discipline is constrained generation: the AI selects from and arranges verified material rather than freely composing. It extracts claims from sources, checks names and figures against the record, writes to journalistic standards, and applies the bad-faith disciplines below. It does not generate facts that are not in its sources, does not fabricate quotations, and does not invent specifics. The work is built to leave a trail — each claim carries its source, so the article can be audited line by line rather than taken on faith. That constraint is enforced by the framework's architecture, not by good intentions alone.
Human review happens by exception, not by default
Most articles are never touched by a human before publication, and that is by design — a human bottleneck on every story would defeat the point of an open, automated newsfeed. Instead, review is triggered automatically when a piece trips a wire: a defamation-adjacent claim about a named person, premature identification of a suspect or victim, a vulnerable-population concern, a fair-use boundary, a contested claim where verification is still thin, or any claim that would cross the floor without a signed columnist's authorization. The default is speed and openness; the exception is a human set of eyes exactly where the cost of an error is highest.
The discipline behind the analyses
Between the news and the signed columns sits a third kind of writing: the analyses. An analysis takes a single news story and shows how it was built — where its facts came from, which words it chose, what it left out — as media literacy, without taking a side. It is the difference between reading a story and watching how the story was made, and like the news it stays inside the shared factual floor. Two documents govern it, and between them they enforce a single rule: an analysis describes, it never accuses.
- The Analysis Style Guide — the clarity discipline. Four moves define it: lead with why the framing matters before any of the machinery, describe only what a text does and never what its author intended (intent is unknowable; effect is right there on the page), translate every term of art into plain language, and close with a checklist the reader can carry to the next story they read. The "what it does, not what anyone meant" rule is the load-bearing one — it is what keeps an analysis from sliding into mind-reading, and from there into advocacy.
- The Analytical Piece Generator — the framework that produces an analysis from a news story and keeps it inside the factual floor. It carries an honest note marking which parts of the pipeline are built and which are still planned, so the document never claims a capability it doesn't yet have.
The convictions behind the columns
The newsfeed is built to be perspective-free. The signed columns are the opposite — they argue. The Editorial Canon is the set of moral convictions that orient all of them: what the publication cares about, beneath any single column. It is designed to direct the writers' attention without dictating their words, which is how fourteen very different voices can share one publication without blurring into one.
The design choice that makes this work is that the Canon is directional, not constraining. It points each voice toward what the publication thinks is worth caring about, but it does not script the sentences — and where a voice's own character pulls against the Canon, the voice wins. That is deliberate. It is what lets a wrathful prophet, a reformed insider, a tradesman, and a parodist all draw on one shared set of convictions without collapsing into a single house style: the Canon supplies the why, and each voice keeps its own how. The Canon is loaded by every signed-column framework — and pointedly kept out of the newsfeed, so that the reporting stays clear of the very convictions that give the columns their charge.
The voice with no name: the Editorial Board
One voice on the site is not a person at all. The Editorial Board is the publication speaking as an institution — the unsigned "we" that answers an opinion-page editorial with the opposite case, written in the same structural shape. Three documents define it and hold it to account, and one design decision sets it apart from every signed voice: the Board does not draw on the publisher's personal body of writing at all. That exclusion is on purpose. An institutional "we" that quietly ran on one person's accumulated opinions would be smuggling a private perspective in under the publication's name; cutting that connection is what lets the Board's voice stay genuinely institutional.
- The Editorial Board Mind — the specification of the institutional voice: what it is, what it will and won't write, and the disciplines it audits itself against.
- The Editorial Board's Historical Lineage — the journalism-standards genealogy its discipline descends from, from the sensational yellow press through the modern unsigned-board tradition. It is the document that explains why an unsigned institutional voice is a recognized form rather than an evasion.
- The Greater Good Doctrine — the symmetry mechanism. It turns the same hard technique-scrutiny the publication applies to liberty-frame editorial pages onto greater-good-paramount ones, with equal force. Applying the identical lens to both sides is the whole point: it is what keeps the Board's opposite-case answers principled rather than partisan.
Naming bad faith without becoming it
Calling a rhetorical move a "Gish gallop" or a "motte-and-bailey" is itself a claim, and it has to meet evidence like any other. The bad-faith techniques catalog is the shared library that makes that discipline possible: thirty-some named techniques across four categories — formal fallacies, informal fallacies, frame manipulation, and coordinated patterns — each with a scholarly source, the signals that identify it from the public record, and the conditions that would prove the identification wrong. The publication names a technique only when the documented criteria are met.
The falsification condition is what separates this from name-calling. Every entry has to carry not just a definition and a source but the specific evidence that would identify the move and the specific thing that would show the label is wrong — so an accusation of bad faith can itself be checked and, if it fails, withdrawn. On top of the general catalog sit outlet-specific overlays that apply the same method to a particular editorial page: the WSJ Technique Catalogue reads liberty-frame opinion writing, the recurring moves by which a benefit for the few gets sold as common sense for the many, and the Greater Good Doctrine above turns the identical scrutiny on the opposite frame. The reason any of this matters is the asymmetry propaganda runs on: a lie costs one breath, an honest rebuttal costs an afternoon, and a publication that answers by hand will always lose on volume. Naming a technique once, in public, with the criteria attached, is how you retire the move instead of re-fighting it every morning.
→ The Bad-Faith Techniques Catalogue
Outlet-specific overlays apply the same method to specific editorial pages: the WSJ Technique Catalogue for liberty-frame opinion writing, and the Greater Good Doctrine above for the other side.
Found an error? Tell us
AI-generated reporting can contain factual errors. If you find one, write to [email protected]. Corrections are made promptly, and the record of the change is kept, not quietly erased — the original mistake stays visible beside the fix, so the correction is part of the public record too.
Everything here is yours: CC0 and forking
This page and everything it links to is released under CC0 1.0 Universal. No rights reserved. Other publications, AI systems, and writers are welcome to adopt, adapt, or fork any of it — the treatise, the floor, the Editorial Mind, the article framework, the canon, the board documents, the technique catalogs. The point is a public commons of values-explicit, journalistically-disciplined reporting, not a proprietary product. We ask only one thing, and it is a request rather than a restriction: if you modify the work, don't represent your modifications as ours.
Main Street Independent's general newsfeed is generated by AI from the published instructions above, and is separated — structurally and editorially — from the signed columns. The instructions are public so the work can be checked, challenged, and forked. The facts of any given story are reported under the consensus values floor; perspective lives only in clearly labeled, signed columns.