📡

Lie-O-Meter TLC AI Lab · Truth Engine

LIVE DATA

⚡ No Spin · No Bias · No Party · Just the Math

The Lie-O-Meter

Track every promise made. Clock every claim stated. We pull from all sources, strip editorial narrative, and show you one thing: do their actions match their words?

—

False Claims Tracked

—

Promises Made

—

Avg. Kept Rate

—

Public Figures

🏆 Accountability Index — All Figures Ranked by Score

Politicians are ranked by promise-keep rate. Journalists are ranked by factual claim accuracy rate. Both use the same 0–100 accountability index — higher = less reliable. Party affiliation does not affect score calculation.

📰 Facts Feed — Narrative Stripped AI Filtered

Editorial opinion and political framing removed. Only verifiable facts and direct quotes remain. Bias label = source outlet lean — not a judgment on the facts. Sources span the full spectrum to counteract selection bias.

🔬

What Is the Lie-O-Meter?

Mission, scope, and what this tool does NOT claim

The Lie-O-Meter is a single-question accountability tracker: do a person's actions match their words? That's it. It does not evaluate whether their politics are good or bad. It does not take sides. It is not a partisan tool.

A politician you personally disagree with who keeps all their promises will score well. A politician you love who breaks promises and makes false claims will score poorly. The math is the same regardless of party, ideology, or policy position. This is accounting — not opinion.

For journalists, the question is different but equally neutral: are the factual claims they make on-air or in print accurate? A journalist can have strong opinions and still report facts correctly. We only score provably false or misleading factual claims — not commentary, analysis, or opinion pieces.

📐

The Scoring Formula

Transparent, mechanical, and applied identically to all figures

Politicians — Promise Accountability Index

Score = ( Broken × 1.0 + Lies × 1.5 + Partial × 0.4 ) / ( Total Scored Items × 1.5 ) × 100

— PENDING items are excluded from scoring entirely until sufficient time has elapsed (≥2 years for domestic policy, ≥4 years for systemic change promises).
— BLOCKED items (promise blocked by court ruling, legislative opposition, or constitutional limit) receive a 0.5× weight reduction vs. a voluntarily broken promise.
— Score represents the weighted percentage of unreliable behavior relative to the maximum possible bad-faith score.

Journalists — Factual Accuracy Index

Score = ( Fabricated × 1.5 + Misleading × 1.0 + Incomplete × 0.4 ) / ( Total Checked Claims × 1.5 ) × 100

— Same formula as politicians — Fabricated = LIE (×1.5), Misleading = BROKEN (×1.0), Incomplete = PARTIAL (×0.4). The system is identical; only the label language differs to match journalistic context.
— Only verifiable factual claims tracked — not opinion, analysis, or commentary.
— Corrections issued within 48 hours receive a 0.5× weight reduction (same BLOCKED discount logic).
— Sample is not exhaustive: we track notable, high-reach claims independently fact-checked by ≥2 sources.

Worked Example: Scoring a Politician

Example: 10 items total (2 pending = excluded, 8 scored)

3 × KEPT+0.0

2 × BROKEN (voluntary)+2.0

1 × BLOCKED (external force)+0.5

1 × LIE (confirmed false)+1.5

1 × PARTIAL+0.4

Total weight: 4.4 / Max possible (8×1.5=12)≈ 37%

🏷️

Category Definitions

Precise criteria for each status label — no subjective interpretation

✅ KEPT

The stated commitment was fulfilled. Outcome matches stated intent within a reasonable scope.

Requires: documented outcome that satisfies the core of the stated commitment. Cross-referenced with at least 2 independent sources.

❌ BROKEN

The commitment was not fulfilled and the figure had the authority and opportunity to fulfill it but did not act.

Requires: sufficient time elapsed + figure held relevant position + no documented blocking force + no documented attempt.

⚠️ PARTIAL

The commitment was partially fulfilled — either in reduced scope, inconsistent execution, or fulfilled by different means than stated.

Requires: documented partial action. Used when "accomplished differently" applies — credit is given for genuine progress.

🚨 LIE

A specific, verifiable factual claim was made that was provably false at the time it was made, not merely optimistic or aspirational.

Requires: claim was factual not aspirational + independently debunked by 2+ non-partisan sources + figure had access to correct information.

⏳ PENDING

Commitment made recently or with a future timeline — insufficient time to fairly evaluate outcome.

Excluded from score calculation. Re-evaluated when timeline elapses or when action/inaction becomes clear.

🔒 BLOCKED

Promise was actively attempted but blocked by an external force: court ruling, opposing legislative majority, constitutional constraint, or international body.

Applied as a nuance tag (not a standalone status). Receives 0.5× weight reduction in scoring. Applies equally to all figures, all parties.

⚖️

The Hard Cases: Where Honest Scoring Gets Complicated

Four categories of ambiguity — and how we handle each, consistently

Accountability scoring isn't always binary. These four categories represent genuine complexity — and we apply the same framework regardless of who the figure is.

🔒 Case 1: Externally Blocked Promises

A promise was made sincerely, action was attempted, but an external force (courts, opposing legislature, constitutional limits) prevented fulfillment. This is categorically different from choosing not to act. We reduce the score weight but do not excuse it entirely — the promise was still made and not kept.

Biden — Student loan cancellation: SCOTUS ruled the executive lacked authority. Action was attempted. Score: BROKEN + BLOCKED tag (0.5× weight).

Obama — Close Guantanamo: Congress passed the Consolidated Appropriations Act blocking transfer funds repeatedly. Action was attempted. Score: BROKEN + BLOCKED tag.

Trump — ACA repeal: Senate vote failed 51-49 (including Republican senators). Score: BROKEN + BLOCKED tag.

DeSantis — Property insurance reform: Market forces and reinsurer pullout partially blocked outcomes. Score: BROKEN (no BLOCKED tag — this was a policy design failure, not an external block).

🔊 Case 2: Political Hyperbole vs. Factual Claims

"Democrats are communists." "This is the greatest economy in history." "Republicans want to destroy democracy." These are rhetorical statements — political speech operating in the register of exaggeration, not factual assertion. We do NOT score hyperbole as lies. We flag it as HYPERBOLE and note what the verifiable underlying concern is, if any. Both parties use hyperbole constantly. We treat it identically.

MTG — "Democrats are communists": Political hyperbole. Not scored as lie. Tagged HYPERBOLE.

AOC — "The world will end in 12 years if we don't address climate change": Rhetorical framing of a real finding (IPCC's 12-year window for 1.5°C pathway). Scored PARTIAL, not lie.

Various — "Greatest/worst [anything] in history": Superlatives without citation are hyperbole. Scored PARTIAL at most.

↗️ Case 3: Accomplished — But Differently Than Promised

A promise was fulfilled in spirit but through different mechanisms than stated. This gets a PARTIAL score — not BROKEN. Credit is given for genuine outcomes. The original framing matters: if a figure claimed a specific method that was wrong, that's flagged under the delivery portion, but the outcome is acknowledged.

Trump — "Renegotiate NAFTA": USMCA was signed. Different name, updated terms, genuinely renegotiated. Score: PARTIAL (not BROKEN). Credit given for outcome.

Biden — "Restore alliances": Rejoined Paris Agreement, NATO engagement increased. Score: KEPT for that specific commitment.

Sanders — "Medicare for All": Not passed, but provisions expanded Medicaid and ACA subsidies through IRA. Score: PARTIAL.

🚨 Case 4: Genuine Deception — Intent Matters

A LIE requires that the claim was: (1) factual, not aspirational; (2) verifiably false; (3) made by someone with access to the correct information. "If you like your doctor you can keep your doctor" — this was a specific, factual claim made repeatedly after internal briefings indicated it wasn't fully true. That's a lie. "I will build 1,000 miles of wall" — that's an aspirational commitment. Scored BROKEN if not done, not LIE.

Obama — "If you like your doctor, you can keep your doctor": ACA architects knew grandfathering was incomplete. Repeated post-briefing. Score: LIE.

Biden — "I was arrested protesting": No records. No contemporaneous witnesses. Score: LIE.

Trump — "The 2020 election was stolen": 60+ courts dismissed for lack of evidence. Republican-led audits confirmed results. Told to top advisors it was wrong (per sworn testimony). Score: LIE.

Warren — Claimed Native American heritage for career purposes: DNA test showed <1/1024 Native American ancestry. Listed as minority hire at Harvard Law. Score: LIE.

📚

Our Sources

Cross-referenced across the political spectrum to counteract selection bias

No single fact-checking source is used for any rating. Every claim requires corroboration from at least two independent sources. We deliberately draw from sources with different editorial leans to reduce systematic bias.

PolitiFact

Pulitzer Prize-winning fact-checker. Promise tracker for all major politicians.

Known lean: Center-Left on selection bias. Counter-balanced with right-leaning sources.

Washington Post Fact Checker

Pinocchio rating system. Extensive promise and claim tracking.

Known lean: Center-Left editorially. Fact-check division operates with separate standards.

Associated Press

Wire service fact-checks. Broad coverage, internationally cross-referenced.

Generally considered Center. Wire service model reduces editorial influence.

C-SPAN Transcripts

Unedited government proceedings. Primary source — no editorial layer.

Non-partisan. Primary source. Used for all congressional and executive claims.

Congressional Record

Official record of all votes, bills, and statements in Congress.

Government document. Not interpretive. Used for voting record verification.

CISA / NIST (Election)

Federal agencies — used specifically for election integrity claims.

Non-partisan federal agencies. Used only for their specific domain expertise.

Heritage Foundation

Conservative think tank. Used for counter-balance and conservative policy analysis.

Known lean: Right. Used specifically to cross-check left-source baseline findings.

Media Matters / MRC

Left and Right media watchdogs respectively — used for journalist fact-checks.

Both have strong leans. Used only when claims are independently also confirmed by a neutral source.

Reuters

International wire service fact-checking division.

Generally Center. International editorial independence from US political scene.

⚠️

Known Limitations — We're Being Honest With You

Every methodology has gaps. Here are ours.

📊

Sample size is not equal across figures. Obama has 533 tracked promises (per PolitiFact's multi-year study). MTG has 39. A single high-profile lie affects a small sample dramatically more than a large one. We note sample size on each profile and are working to normalize this in v2.

⏱️

Time in office affects score. A figure who's been in office 30 years has more opportunities to keep and break promises than someone in their first term. PENDING items are excluded, but tenure context matters. We display context for each figure.

🗣️

Volume of statements varies wildly. High-output communicators make more claims and therefore produce more verifiable data — some accurate, some not. A figure who says less is harder to rate but isn't necessarily more honest. Low sample size = lower confidence in the score.

🔍

Fact-checkers have selection bias. They tend to check claims that are newsworthy — which skews toward controversial figures. We cross-reference across ideologically opposite sources to reduce this, but it cannot be eliminated entirely.

🌐

Context collapses in a single score. A 70% score doesn't tell you what kinds of promises were broken. Foreign policy promises broken may be more serious than domestic ones, or vice versa. The Tracker view is designed to let you look beneath the number.

🤖

Human judgment is still required at the edges. The scoring formula is mechanical, but categorizing a claim as BROKEN vs. BLOCKED vs. PARTIAL requires human editorial review. We document our reasoning for each item. We welcome dispute submissions.

📩

Dispute a Rating

If our data is wrong, we want to know

Every item on this tracker is open to challenge. If you believe a rating is incorrect — for any figure, any party, any ideology — we want to hear it. Disputes must include: (1) the specific claim being disputed, (2) a primary or credible secondary source that contradicts our rating, and (3) which category you believe it should be in and why.

We will review all disputes and update publicly. Corrections are logged — we don't quietly edit. The goal is accuracy, not a fixed narrative.

📧 disputes@tlcailab.com — include "LOM DISPUTE" in subject line

Lie-O-Meter is a TLC AI Lab project. Data is manually curated by editors and cross-referenced against public records, C-SPAN transcripts, voting records, and established fact-checking databases across the political spectrum. Each item is sourced. Promise statuses are determined by comparing the original stated commitment to verified, documented outcomes — using a mechanical, party-blind formula. This is not political opinion — it's accounting.