Factual Foundations Report Last updated: 2026-06-28 14:00 BST

Factual Foundations
A Portal for Source-Linked Political Reporting

Compiled with MiniMax M3 (latest). Source-linked; no auto-publish. Model version history.

An AI-curated, source-linked, bias-disclosed portal for contested political topics. One page per topic, periodically refreshed, fallacy-audited, and designed to anchor public discussion in evidence rather than spin.

The Problem with Current News

Most news consumers today face a paradox. There has never been more raw information available — yet opinion formation based on fact has never been harder. This report documents why existing mainline sources systematically fail to provide what readers need to form their own opinions on contested political questions, and outlines what a better system would look like.

The core failure: Mainstream news sources are optimized for engagement and ideological alignment with their existing audience — not for providing the kind of full-context factual record that enables readers to think clearly.
Share this section:

Five Recurring Failures

  1. Selective context omission — facts that explain a reaction are elided, leaving readers to wonder why anyone cares about an event whose causes are not described.
  2. False equivalence — well-documented conclusions and political positions are presented as "two sides" with equal weight, even when evidence is asymmetric.
  3. Source-blackout by default — readers are not shown what sources are not covering. The information that "X outlet has not reported on this" is itself crucial.
  4. Source-bias opacity — outlets operate without disclosing their known editorial biases. A Manhattan Institute study of 28,000+ Wikipedia articles found right-of-center figures associated with more negative language than left-of-center — but the average reader doesn't know that.
  5. One-and-done reporting — events are reported once, then archived. No mechanism for retrospective correction, follow-up, or re-examination when new evidence emerges.

Sample: Why the Volhynia Case Is a Microcosm

The June 2026 dispute over Poland's revocation of Zelensky's Order of the White Eagle is a perfect case study. Every mainline outlet reported the event. Almost none reported the causal context that makes the event intelligible. Below is a summary of what each source actually said versus what an informed reader needed to know.

Share this section:
SourceWhat it reportedWhat it omitted
AP News "Nawrocki stripped Zelensky of the Order. Support for Ukraine unchanged." Why the UPA is controversial. No specific atrocity context. No exhumation dispute. No historical depth. A reader knows what happened but not why.
BBC News "Zelensky stripped of Polish honour over WW2 name of army unit." Specific methods of UPA violence. Exhumation ban history. Polish self-defence narrative. Still insufficient context for understanding Polish anger.
Wikipedia (English) Detailed Atrocities section, named perpetrators, scholarly consensus. Left-leaning systemic bias (Manhattan Institute 2024). False-equivalence between Polish genocide designation and Ukrainian "tragedy" framing.
Wikipedia (Ukrainian) Polish reprisals detailed. Sahryń 800+ killed, Berest 200+ killed. No description of UPA killing methods. The article simply does not mention that UPA fighters sawed people alive, impaled babies on pitchforks, or crucified priests. This is a deliberate editorial choice, not a difference of historiography.
Kyiv Independent / Kyiv Post Ukrainian official position. Sybiha: "strategic mistake." Minimal UPA atrocity detail. Pro-Ukraine framing presented as fact.
The pattern: Every source above is factually accurate about what it covers. Each one omits a different piece. A reader who only consumes one source gets a partial picture. A reader who consumes all of them still does not encounter the "exhumation ban lasted 7 years" or "the Ukrainian Wikipedia article has no atrocities section" — these are meta-facts that no single source surfaces.

The Factual Foundations Portal: What It Should Be

What is needed is not a new news outlet but a different kind of artifact for each contested topic: a self-contained, source-linked, bias-disclosed page that lets a reader see the full factual picture in 20 minutes — and that updates as new evidence emerges.

Share this section:

Core Design Principles

🔍

Source-First

Every factual claim has a clickable citation. No unsourced assertion. The page is a network of links, not a wall of authority.

📐

Bias-Disclosed

Every referenced outlet has its known editorial bias documented. The Manhattan Institute study, AER study, Wikipedia co-founder statements are surfaced for each source.

🌍

Multi-Language

Cross-check articles in multiple languages. A topic that has Polish, English, and Ukrainian Wikipedia editions has three different editorial choices; surfacing them is the point.

🔁

Self-Updating

An AI agent periodically re-fetches, re-summarises, and flags new evidence. The page is a living artifact, not a one-time publication.

💬

Discourse-Ready

Each section has a share button, a comment surface, and embedded X/Twitter. Discussions stay anchored to the factual record.

🚫

Fallacy-Audited

A dedicated section for the most common fallacies used to dismiss the topic. The reader comes pre-loaded with the rhetorical moves they will encounter.

The most labor-intensive part of maintaining a "factual foundations" page is keeping it current. New events, new exhumations, new witness statements, new political developments — all of these should update the underlying record. A human researcher cannot realistically do this for more than a handful of topics. An AI agent can.

Share this section:

The Proposed Workflow

  1. Daily / weekly re-fetch: The agent re-pulls all cited sources via SearXNG + web_extract. Compares new content to existing report.
  2. New-evidence flag: Any new facts, new statements, new exhumation findings, or new editorial framing are flagged with timestamps and added to a "Recently updated" section.
  3. Cross-source comparison update: The cross-source table is regenerated, with new framing differences surfaced.
  4. Fallacy review: The fallacy section is checked against recent discourse — are new rhetorical moves being used? Are new ones documented?
  5. Human-readable diff: A short summary of what changed and why is appended, so a reader can spot what the AI updated.

What the AI Can and Cannot Do

Can doCannot do
Re-fetch and compare versions Tell the reader which version is "true" without human judgment
Surface new sources that meet the standard Force a partisan reader to update their priors
Detect and document fallacious rhetoric Prevent rhetorical misuse of the page
Maintain source-quality metadata Resolve contested scholarly debates on its own
Update timeline, cross-source table, source grid Decide which disputed fact is "in" and which is "out" of the report
Design principle: The AI is a research assistant and compositor, not an authority. Every claim it adds is cited. Every editorial choice is documented. A human reader can verify or reject any AI-generated update by following the citation. This is the only way the system can scale without becoming a new form of unaccountable authority.

Sharing, Discussion, X/Twitter Integration

For the portal to function, the page must be discussion-friendly. A reader who finishes the page should be able to share a specific section to X/Twitter, see what others have said about that specific section, and add their own analysis with citations.

Share this section:

Section Anchors

Every section is a deep-linkable URL. The "exhumation dispute" section is its own URL. The "UPA atrocities" section is its own URL. A reader sharing a section to X/Twitter does not share the whole 10,000-word page — they share the one part that has the facts they want to discuss. This is the difference between a portal and a dump.

Embedded Discourse

Where primary sources (X/Twitter posts, official statements) exist, they are embedded alongside the editorial synthesis. The page is a fact+context surface, not a curated monologue.

DT
Donald Tusk
@donaldtusk · Jun 19, 2026
Konflikt między Polską i Ukrainą cieszy Putina i szokuje naszych sojuszników. Zadaniem prezydentów Zelenskiego i Nawrockiego jest tonowanie emocji, a nie podsycanie napięcia. Linia frontu przebiega gdzie indziej.
7:15 PM · Jun 19, 2026
English translation + FF context

Translation: "The conflict between Poland and Ukraine delights Putin and shocks our allies. The task of Presidents Zelensky and Nawrocki is to calm emotions, not to inflame tension. The front line runs elsewhere."

What this says vs. what it implies: On the surface, Tusk appears to "criticise both sides" by saying both presidents should de-escalate. In context, the framing is pro-Ukrainian accommodation: Tusk calls the Polish-Ukrainian fracture a Putin win, and his "de-escalation" recommendation favours accepting Ukraine's positions on the UPA issue rather than confronting them. This is the position the Polish opposition (PO, Tusk's party) and the Ukrainian government have aligned on throughout the dispute.

Why this matters for FF: quoting Tusk's tweet as "criticising both sides" without the framing above reads as a neutral call for peace, when the position is structurally a softer Polish response to Ukrainian state honours. The two readings are not equivalent. A hostile reader can take the surface text and ignore the structural position; a careful reader needs both.

Comment Thread Per Section

Each section gets its own comment surface. Comments are anchored to the section, not to the page. This prevents the standard problem of "I disagree with the article" when the commenter only read the headline. Discussion stays anchored to the specific factual claim being discussed.

The Fallacy Handbook: Anticipating Dismissals

Any factual report on a contested political topic will be dismissed. The most common dismissals are not based on engaging with the evidence — they are rhetorical moves designed to skip the engagement. Below are the fallacies most often used to dismiss the kind of factual record this portal produces, with explanations and counters.

Share this section:
1. "It was written by AI, so it's unreliable" (Genetic Fallacy)
Dismissing a claim based on the origin of the source rather than the content of the claim.
"I don't care what it says, it's AI-generated, so it's hallucinated / propaganda / can't be trusted."
Every factual claim has a citation. If you disagree with a specific claim, identify it and provide a counter-source. The LLM is the retrieval and composition tool, not the authority — the cited sources are. A statistical report isn't less valid because Python ran the calculations.
2. "It's Russian propaganda" (Poisoning the Well)
Pre-emptively dismissing a source or claim by associating it with a discredited origin, without engaging the content.
"The exhumation dispute is being pushed by Russian propaganda to divide allies."
The exhumation ban is documented by IPN, Notes from Poland, intellinews, and Ukrainian government statements — not Russian propaganda. Russia's interest in seeing Ukraine-Poland fracture is real, but pointing at Russia's motives does not invalidate the underlying facts. The Polish Institute of National Remembrance is not a Russian asset.
3. "Right-wing rhetoric" / "PiS propaganda" (Ad Hominem by Association)
Dismissing a claim by associating its source or claimant with a political faction, rather than engaging the claim.
"Nawrocki is PiS, so anything he says about UPA is just political posturing."
Nawrocki was elected in 2025 on a platform that included historical reckoning. He is also an academic historian who headed the IPN — a position that requires engagement with primary sources regardless of party affiliation. The claim that his statement about UPA is purely political can be tested against the historical record. It cannot be dismissed by association.
4. "What about Pilsudski? / What about AK?" (Whataboutism / False Equivalence)
Deflecting from a specific documented action by pointing to a different action, often unrelated or incomparable, by the other side.
"Why do you only talk about Volhynia? Pilsudski did X. AK did Y. Why don't you focus on Polish atrocities instead?"
Pilsudski (1867–1935) died eight years before the Volhynia massacres. He was a statesman, not a military commander conducting ethnic cleansing. The Polish Army (AK) is documented to have committed some retaliatory actions during 1943–45 — the IPN has investigated and acknowledged this. None of this is a defense of the 1943 UPA atrocities. Both sides committing crimes does not mean both sides committed the same crimes. See the dedicated fallacy section in the Volhynia report.
5. "You're assuming the person is Russian / PiS / fascist" (Loaded Question / False Attribution)
Discrediting an argument by attributing it to a discredited group without evidence, especially in response to someone whose politics you don't actually know.
"Anyone who criticises Ukraine is clearly a Russian troll or far-right."
A claim's validity is not determined by the politics of the person making it. The Polish president, Polish IPN, Polish survivors, and the Polish government are not Russian trolls or far-right. The Ukrainian Institute of National Remembrance has also acknowledged the 1943 UPA actions against Poles. The fact that multiple parties across the political spectrum make the same claim is evidence the claim is real, not evidence of conspiracy.
6. "You have to see the Ukrainian perspective / context" (Context Inflation as Counter)
Using "context" as a rhetorical shield to avoid the actual claim. Always paired with a vague appeal to "the other side."
"Yes, but you have to understand the historical context from the Ukrainian perspective — Bandera was fighting for independence."
The portal does provide the Ukrainian context. It is the point of the report. But "context" does not retroactively justify sawing children alive. The Ukrainian perspective on why the UPA was formed is included — it does not make the atrocities false.
7. "Why now? / timing suspicious" (Ad Hominem Circumstantial)
Rejecting a claim based on the timing or circumstances of when it is made, rather than its content.
"This is being raised now only to embarrass Zelensky during the war / to help Russia / for political gain."
The timing of a fact is not the fact. Nawrocki has been clear that Poland's military support for Ukraine is unchanged. The Polish position on the UPA is consistent with Polish policy for 80 years and was acknowledged in 2016 by the Polish parliament. The fact that Ukraine is at war with Russia does not change what UPA did in 1943.
Rule of thumb: If a dismissal cannot be stated as "X specific claim in this report is wrong because Y specific source says Z," it is almost certainly a fallacy. A factual page anticipates these moves and pre-empts them.
13. Appeal to Model Bias - "Grok is biased - don't trust what an AI says about Volhynia."
Dismissing a fact by claiming an AI is biased, after that AI independently confirmed the cited evidence.
"Grok is biased - don't trust what an AI says about Volhynia."
Genetic fallacy at one remove. The AI returned the same conclusion that the cited primary sources support (Motyka's book, the IPN archive, BBC reports, Notes from Poland articles, and the Ukrainian government's own statements). The right next move is to engage those sources, not to dismiss the model that confirmed them. If 'Grok is biased' actually means 'Grok agrees with you and I do not', state that directly and address the underlying disagreement.
14. Motte and Bailey - "The Polish government is not honouring UPA - we are merely noting that historical events have complex interpretations. (After publicly invoking Bandera as a hero.)"
Claiming an extreme position publicly, then retreating to a more defensible one when challenged.
"The Polish government is not honouring UPA - we are merely noting that historical events have complex interpretations. (After publicly invoking Bandera as a hero.)"
State the original claim plainly. The arguer cannot stake out an extreme claim, then retreat to a defensible one after critique. If the modest claim is what they 'really meant', say it first; the retreat is not the original position. The Polish-Ukrainian dispute over the Order of the White Eagle did not begin with 'historical events have complex interpretations' - it began with Zelensky honouring a unit explicitly named after UPA. That is the claim on the table.
15. Straw Man - "FF says Volhynia is a microcosm proving all Ukrainian nationalism is fascist. (It does not say that.)"
Misrepresenting the position being critiqued so it can be attacked more easily.
"FF says Volhynia is a microcosm proving all Ukrainian nationalism is fascist. (It does not say that.)"
Restate the original claim exactly. The straw man is a simpler or more extreme version of the actual position. If you are rebutting 'FF says Volhynia is a microcosm', that is not what FF says. The actual claim, per the methodology page, is that the documented record on UPA conduct and the 2017-2024 exhumation ban justifies Polish wariness toward Ukrainian state honours for UPA-linked units. Attacking a simplified version of that does not rebut the original.
16. Tone Policing - "Even if true, the way you write about this is so aggressive that reasonable people dismiss you. Try being more diplomatic."
Rejecting a claim because of the tone in which it is delivered, rather than the content of the claim.
"Even if true, the way you write about this is so aggressive that reasonable people dismiss you. Try being more diplomatic."
Tone is not evidence. The truth of a documented action does not depend on the politeness of the document. A photographic record of atrocities is what it is regardless of how it is described. The demand for 'more diplomatic' framing of atrocities is itself a tactical move used by people who cannot rebut the evidence. Politeness is a courtesy, not a refutation.
17. Bulverism - "Of course FF presents Polish survivors as victims - FF is hosted on a Polish-majority server, so it has a Polish perspective."
Dismissing an argument by speculating about why the arguer holds it, without engaging the content.
"Of course FF presents Polish survivors as victims - FF is hosted on a Polish-majority server, so it has a Polish perspective."
Speculating about motive is not rebuttal. The location of the server hosting a fact does not change whether the fact is true. Motive-inquiry is legitimate when backed by evidence (e.g. 'here is a specific citation that contradicts your claim'); motive-stacking is dismissive when used as a substitute for engagement. If the argument is biased, identify the bias in the specific claim, not in the alleged bias of the speaker.
18. No True Scotsman - "Real Ukrainians don't honour UPA. The real Ukraine respects historical truth. You are talking about fringe elements."
Defending a generalised claim by exempting inconvenient counter-examples from the group.
"Real Ukrainians don't honour UPA. The real Ukraine respects historical truth. You are talking about fringe elements."
Reject the purity gate. If the action was committed by people claiming the identity in question, the identity-membership claim does not erase the action. The Ukrainian government awarded a unit the title 'Heroes of the UPA' - the President signed the decree. The President is, by the most generous definition, 'real Ukraine'. A purity-test response ('real Ukrainians don't do that') does not change the document. Define 'real' by the documented record, not by an idealised version.

On the Use of an LLM to Compile This Portal

An LLM compiled this page. See the Model Version History section in the methodology page for which model was used at the time of writing. As with the underlying factual pages, the standard fallback dismissal is "it's AI, therefore it's unreliable." This is a genetic fallacy. The LLM is a retrieval, reading, and composition tool, not an authority. Every factual claim on this page is cited. Every editorial decision is documented. A human researcher with the same tools would produce the same factual content.

Counter to the "it's AI" dismissal: "Which specific factual claim on this page is incorrect, and what is your source for the correction?" If the person using this dismissal cannot answer, they are not engaging in good-faith discourse.

Sources Referenced

Why no links? Sources are listed by name only, not as hyperlinks. Three reasons: (1) Wikipedia entries are editorially controlled by the same community that shapes the bias we are documenting - linking to a Wikipedia article amplifies its SEO weight and effectively endorses it as a source; (2) several sources we name here (Ukrainian Wikipedia, Wire services with state bias) are cited specifically because they fail some standard - linking to them is a contradiction; (3) the cited primary sources (Motyka's book, IPN archive, BBC, Notes from Poland) are accessible by name search in any browser, so the link is unnecessary for navigation. If you find a source you cannot locate by name, that is a finding worth noting in the report itself.
Manhattan Institute
Research
"Is Wikipedia Politically Biased?" 2024 study, 28,000+ articles.
American Economic Review
Peer-reviewed
First empirical measurement of Wikipedia political slant (2012).
BBC / AP / Reuters
Wire services
Core reporting on 2026 Zelensky-Nawrocki dispute and Volhynia events.
IPN (Polish Institute of National Remembrance)
Government institute
Primary forensic documentation of UPA crimes and exhumation work.
Wikipedia (PL/EN/UK)
Encyclopedia
Three-edition comparison shows how the same topic is treated differently.
Notes from Poland
Independent journalism
Best English-language Polish-perspective coverage with embedded primary sources.
Ukrainska Pravda / UNN / Rubryka
News
June 2026 reporting on Puzhnyky and Huta Pieniacka exhumation results.