01The rules for this page
Anything that would change a reader's judgement or their actions gets logged here after it is fixed, with the date, the page, what was wrong, why it was wrong, and what it says now.
- Logged: factual errors, phrasing that would mislead an action, tools returning a wrong verdict, and links that are dead or point somewhere unintended.
- Not logged: typos, layout adjustments, pure copy-editing, and additions that do not alter a conclusion.
- Nothing is fixed quietly. We do not delete an incorrect passage and pretend it was never published.
A note on scope: this site launched in Chinese first, and the English section was published afterwards. The entries below occurred during that build, on the Chinese pages; the English pages carried the corrected version from the day they went up. They are logged here because the underlying errors — and the reasoning behind the fixes — apply to both language versions.
02The log
2026-08-26 · The address validator misread a class of Solana address as Bitcoin
Page: address format validator
What was wrong: the tool chose its branch from the first character plus the length. One class of Solana address — for instance the system address made up of thirty-two "1" characters — satisfies both "starts with 1" and "length within the Bitcoin range", so it was routed into the Bitcoin Base58Check branch and reported as "looks like a legacy Bitcoin address, but the checksum fails". A perfectly valid address was flagged as broken.
Why that mattered: reporting a valid address as faulty makes people think they copied it wrong, which can push them to go and find another address from somewhere less reliable — a worse outcome than the original confusion.
What it does now: branches are chosen by the byte length after Base58 decoding — 32 bytes is Solana, 25 bytes goes to the Bitcoin and Tron branches. Those lengths do not overlap, so there is no ambiguity. Before publishing, the change was re-checked against a set of known addresses: the Bitcoin genesis address, P2SH, Bech32, Taproot, the official EIP-55 examples, Tron and Solana, plus a single-character-altered variant of each.
2026-08-26 · The glossary keyword filter never appeared at all
Page: Web3 glossary (both language versions)
What was wrong: the filter script looked for the term list under the id gl-list, while the page marked it up as list. Finding nothing, the script returned immediately — so the filter box stayed hidden, and the line reading "this page has a keyword filter that needs JavaScript to appear" stayed on screen. Readers with JavaScript running were being told they had it switched off.
Why that mattered: more than a missing feature. The page described a capability that did not exist and blamed the reader's browser settings for its absence — our defect, presented as their problem.
What it does now: the script selects the list by the semantic selector dl.glossary rather than that specific id. Both language versions were then exercised in a real browser: the filter box appears, the notice is removed, the initial count reads 24 entries, typing "gas" narrows it to 3, and a term with no match hides every entry.
2026-08-26 · The wallet setup article omitted the optional passphrase
Page: setting up your first wallet
What was wrong: the section on writing down a seed phrase covered only the twelve words and made no mention of the optional passphrase some wallets offer during backup, sometimes called the thirteenth word.
Why that mattered: a reader meeting that option in their wallet would not know whether it was required or optional, or what enabling it implies. Worse, some people reason that "an extra layer cannot hurt", set one casually, and forget it — which produces the same outcome as losing the seed phrase.
What it says now: a section was added explaining what it does, that it has no recovery either, that it must be stored separately from the phrase, and why we recommend not using it on a first wallet.
2026-08-26 · A broken title tag on one page
Page: moving coins from an exchange into your wallet
What was wrong: the closing <title> tag on that page had been typed as the closing punctuation of an attribute instead, leaving the title element unclosed.
Why that mattered: the title is what appears in search results and browser tabs, and an unclosed tag can cause following content to be rendered as part of it.
What it says now: the correct </title> was restored, and a tag-closure check was run across every page on the site before publication.
03Telling us about a problem
Write to [email protected] naming the page, the passage and what you think is wrong. If you have a source, include it.
Information in this field ages quickly: exchanges redesign, fees are adjusted, wallet interfaces move — and accurate material slowly stops being accurate. We cannot track every change in real time, but we fix what we find and log it here as it happened. The reasoning behind that policy is in About this site.