Don't trust.
Verify.

Throughout the site you'll see small orange badges: don't trust, verify×DON'T TRUST, VERIFYClaim: This is what a verification badge looks like when you click it.Verify at: Don’t Trust. Verify.Try it: click the badge to toggle this panel open or closed. Click a badge to open a panel with the exact claim, where to verify it, and why it is flagged.

Why every claim has a source

Bitcoin's core design principle is that you don't have to trust anyone. You run a node, you verify the blocks, you confirm your balance yourself. Nothing is taken on faith. Every transaction in the whole 17-year history is checkable by anyone.

Primary-data verification worksheet: copy these fields before sharing a claim.
1. Claim and source: write the exact claim and page URL. Open the original institution’s table or series, not just a chart screenshot, book summary or homepage. Record series ID, title, publisher, download/retrieval date and observation dates.
2. Definitions: record population, units, frequency, seasonal adjustment, base year, real or nominal basis, and whether values are observations or projections. Check revisions and whether the latest year is complete.
3. Worked audit: test “unchanged dollars lost 87.1% of CPI purchasing power from 1971 to 2024.” BLS published annual-average CPI-U, U.S. city average, all items, not seasonally adjusted, 1982–84 = 100, is 40.5 for 1971 and 313.689 for 2024. Use both published annual averages, not a December index at one endpoint. Calculate 313.689 / 40.5 = 7.7454: the price-level increase is 674.5%. The inverse calculation (1 − 40.5 / 313.689) × 100 gives 87.1% purchasing-power loss, rounded to one decimal. These are different percentages, not annual inflation rates. The example excludes interest, wages, tax and household-specific spending.^[1][2][3]
4. Comparable assets and wages: use common dates and frequency; for nominal X, rebase real values as 100 × (X_t / CPI_t) / (X_base / CPI_base). Do not deflate an already-real series twice. Record dividends, rent, costs and tax assumptions; a price-only index is not a total return. LES1252881600Q is quarterly median weekly real earnings starting in 1979, not hourly pay since 1971. If the overlapping data do not support the comparison, leave it blank instead of filling in plausible values.^[4]
5. Causation and forecast: a shared 1971 start date does not prove that ending convertibility caused all later inflation. List plausible alternative causes and evidence that would distinguish them. For a price forecast, record model version, publication date, target date, interval, assumptions and out-of-sample results; do not score a revised model as if it had been published before the outcome.
6. Conflicts and decision: record holdings, sponsorship and affiliate compensation. A disclosure is not validation of the math. Reject guaranteed-return or guaranteed-recovery language, preserve the source file, and write a verdict: supported for these dates/units; conditional on these assumptions; or unverified: do not rely on it.^[5][6]

The internet is full of finance content that's confidently wrong, and AI chatbots now generate it at scale. The only defense is: check the source. See AI Financial Advice for why model output in particular demands verification.

How to read a verification badge

Click any don't trust, verify×DON'T TRUST, VERIFYClaim: The badge expands to show you the specific claim, the source, and why it's flagged.Verify at: This page ↗Badges are interactive. Open them, read the source, close them with the × or the Escape key. badge and the panel below it shows:

• The claim: the specific statement that needs verification, in plain terms.
• The source: link to the actual data producer, legal text, issuer document or original research relevant to the claim. A book or peer-reviewed paper may be secondary for a particular statistic. Confirm that the cited passage supports the exact claim and period, not merely the same subject.
• Why it's flagged: most common reasons are "rates change," "data updates monthly," "law may have been amended," or "self-reported figure."

Keyboard users: tab to a badge and press Enter or Space to open it. Press Escape to close. Mobile: tap to open, tap again to close, or tap elsewhere on the page.

What gets flagged

Any claim where the right answer today might not be the right answer next month. Categories that get flagged routinely:

• Dollar figures that update: IRS contribution limits, HYSA APYsAnnual Percentage Yield (APY)An annualized measure of deposit interest that includes the effect of compounding. It is not an inflation-adjusted or after-tax return.Full definition, ETFExchange-Traded Fund (ETF)A basket of investments (stocks, bonds, or Bitcoin) that trades on a stock exchange like a single share. AUM and expense ratios, current prices.
• Moving statistics: Bitcoin CAGR over rolling windows, volatility figures, mining hash rateHash rateHow many hashes (guesses at the puzzle that wins a block) a miner makes per second. The network hash rate is the total across all miners: the higher it is, the harder and more expensive it would be for any attacker to overpower the network.Full definition, Sharpe ratios.
• Legal and tax rules: IRS notices, revenue rulings, state tax laws, pending regulatory changes.
• Self-reported industry data: Bitcoin Mining Council energy mix surveys, corporate treasury holdings, ETF self-disclosures.
• Quotations: where exact wording matters and the source material is second-hand or summarized. This site paraphrases and flags rather than fabricating a quote.

What doesn't get flagged

Claims that are stable and already have a cited primary source in the page's sources block don't carry badges. The badge is a prompt to re-check; a stable citation doesn't need one.

Opinion and framing must be labeled, but factual premises remain checkable. “Bitcoin could hedge fiscal-policy risk” is a thesis; “it cannot lose if held long enough” is an unsupported return assurance, not an opinion exempt from scrutiny. Bitcoin has no guaranteed price or purchasing power.^[6]

Report a bad source

For questions, corrections, or source disputes:

Email hello@fiatisfake.org with:

• The URL of the page
• The specific claim
• The primary source you believe is correct (with link if possible)

No registration, no form, no account. Just email. Updates land fast when a claim on this site is wrong.