The Artificial Wasteland · a portal across eight layers

Nothing Here Is a Bug

Eight layers of this site each arrived, from a different direction, at the same strange place: a standard that authorises its own indeterminacy. Not a standard that is unclear, and not an implementation that got it wrong. A standard that says, in writing, that more than one answer is correct. This page puts the question to five conforming implementations at once and counts how often they part company.

There is a gap in the middle of a great many specifications, and it is not the gap you expect. The interesting failure is not wrong. Wrong is easy: somebody wrote a bug, somebody files it, somebody fixes it, and the standard is the thing that settles the argument.

The interesting failure is correct but undetermined. Two implementations, both conforming, both careful, give different answers to the same question, and there is no test that either fails and no ticket anyone could file. If you happen to notice, you have nowhere to take it, because the specification already considered the question and declined to answer it.

Eight layers here found that shape independently, in eight unrelated crafts, and none of them said the shape was the finding. This portal says it, and then measures it.

The battery

Twenty-four questions. Three of them are controls: places where the standards genuinely do determine the answer, so any disagreement would mean the instrument was measuring noise rather than latitude. The rest are questions some standard hands to the implementation, each printed with the clause that hands it over.

They were put to five implementations on one machine at one moment: four command-line runtimes and one browser, two distinct JavaScript engines between them. Then your own browser answers the same twenty-four, in the rightmost column, without sending anything anywhere.

▌ divided: more than one answer among the implementations measured ▌ agreed: latitude that nobody took shaded row: a control, where the standard determines the answer

The clock reading in Cancún on a July morning in 1990 is the one to look at hardest. Every runtime above sits on the same machine, and the clause printed beside that row permits every answer any of them gives, yet they do not agree about what o'clock it was thirty-six years ago. The answer is not a fact about Mexico. It is a fact about which release of the time zone database each build was compiled against, and those releases do not run in order of the version numbers: Node 21.7.3 is a higher version than Node 20.20.2 and an older build, because the 20 line kept getting patch releases after the 21 line stopped. That is what makes this kind so hard to notice. You cannot reason about it from the version number in front of you.

Five ways to be correct and undetermined

The eight layers are not eight instances of one thing. Sorted by the mechanism that produces the latitude rather than by the field they live in, they fall into five kinds, and the kinds behave differently: some can be closed and some cannot.

  1. KIND I

    The standard declines, in writing

    The clause exists, it is short, and it says the answer is yours. ECMA-262 has never specified what Math.exp returns and has not specified it since 1997; the HTML standard lets a browser report fewer processors than it has, down to one, to limit fingerprinting. A reader who goes looking for the rule finds the rule, and the rule is that there is no rule.

    In this ground: The Rounding Your Browser Was Allowed to Get Wrong (twenty-three functions named as implementation-approximated, measured against a correctly-rounded answer built in BigInt) and The Cores Your Browser Won't Admit To (three engines' source read, and WebKit returning one of exactly two constants).

  2. KIND II

    The artifact carries two answers and does not say which is the copy

    The format was designed with redundancy, for good reasons, and then the two copies drifted. A ZIP records each entry's metadata twice, once in the local header and once in the central directory; a TrueType font carries three different vertical extents; a text file's bytes are frequently valid under several encodings at once. The standard is complete. It is the artifact that is ambiguous, and the ambiguity is invisible until two readers disagree.

    In this ground: The Archive That Disagrees With Itself, Your Font Does Not Agree With Itself, The Encoding Your Text File Admits To.

  3. KIND III

    The answer is relative to a version of a register nobody quotes

    The specification is exact and defers to an external body of data that keeps moving: the time zone database, the Unicode collation tables, CLDR's names for countries. The answer is fully determined given the version, and the version is never quoted, so the answer travels as though it were a fact. This is the kind of the five most likely to change underneath an application nobody has touched, because the register can be updated by the operating system beneath it.

    In this ground: The Past Has a Version Number (249 archived releases of the time zone database compiled and diffed; 145 of 248 consecutive releases changed a clock reading in the past).

  4. KIND IV

    The decoder accepts more than any encoder writes

    An encoding is defined in both directions and the two directions are not inverse. There are inputs the decoder will happily accept that no conforming encoder ever produces, so a value can be legal, unambiguous, and unreachable, and nothing in the specification calls that an error.

    In this ground: The Twenty-Four Declets No Encoder Writes (a complete census: an encoder writes 1,000 of the 1,024 ten-bit patterns, and every one of the remaining 24 still decodes).

  5. KIND V

    The authority itself declines to choose

    The furthest from a computer and the closest to home. Ask the standard pronouncing dictionary of American English how a word is said and for two words in every five of running text it gives you more than one answer, because the language has more than one answer and the dictionary is honest about it. Any system downstream that needs a single reading has to pick, and its picking is not in any specification at all.

    In this ground: Two Words in Every Five (39.28% of running words carry more than one reading, thirteen named variations account for 36.63 points, and what the argument about spelling reform is actually about turns out to be 0.218%).

Read the five in that order and something falls out that no single layer could say. Only Kind II and Kind IV are closable. A redundant field can be made authoritative and a decoder can be made strict, and in both cases somebody has to break compatibility to do it. Kind I is a deliberate grant, and taking it back means making implementations slower or making some of them non-conforming overnight. Kind III cannot be closed at all, because the register is the world and the world moves; the only available fix is to quote the version, which nobody does. And Kind V is not a defect in the first place. The dictionary that lists two pronunciations is the accurate one.

What the gap is worth to somebody else

There is a reason two of the probes above are about drawing rather than arithmetic.

The 2D canvas specification fixes the geometry of a drawing and says nothing about how it is rasterised. Which font a generic family resolves to, how the edges are antialiased, whether the text is hinted: all of that is left to the implementation, correctly, because it depends on the machine. So two conforming browsers asked to draw the same sentence produce different pixels, and the difference is stable for one machine and different across machines.

That is the whole of canvas fingerprinting. It is not an exploit and it patches nothing, because there is no defect: the latitude is the specification working as designed, and it is worth money precisely in proportion to how undetermined it is. Mowery and Shacham named the technique in 2012 (Pixel Perfect: Fingerprinting Canvas in HTML5, W2SP), and Laperdrix, Rudametkin and Baudry measured it in the wild in 2016 (Beauty and the Beast, IEEE S&P), where canvas rendering was among the most discriminating attributes of the browser fingerprint. Those are their measurements and not ours; this page measures only what the machine in front of you answers, and keeps it there.

Which is the turn worth taking away. The gap between correct and determined is not only an epistemic curiosity. It is a resource. Somebody is already mining it, and the reason they can is that everyone else reads a specification as a promise that the answer is the same everywhere.

The check

The controls. Three probes ask things the standards do determine: Math.sqrt(2) and 0.1 + 0.2 (IEEE 754 requires correct rounding, so exactly one bit pattern conforms), and Unicode normalisation (frozen by policy: a composition, once defined, may never change). All three came back identical across all five implementations. Had any of them divided, the divided rows below would mean nothing.

What was measured, exactly. One Linux machine, one moment, five implementations, listed in the table's own header with their versions. Two distinct JavaScript engines: V8 (three Node builds and Chromium) and JavaScriptCore (Bun). That is a real but narrow sample, and the narrowness is stated rather than papered over: there is no Firefox or Safari here, because the container this was captured in could not download either.

What your column is and is not. Your browser runs the identical battery, in your tab, and the result is compared with the captured column in your browser and nowhere else. Nothing is transmitted, no identifier is derived, and the canvas probe reduces a drawing to a 32-bit hash and discards the image. A row where you differ from Chromium is not evidence of a bug in either.

What the count does not claim. "Five of nineteen" is a property of these nineteen questions, which were chosen because a standard grants latitude on them. It is not an estimate of how much of any specification is undetermined, and no such estimate is offered.

Reproduce it. node research/correct-is-not-determined/capture.mjs re-runs the battery on each runtime it finds at the paths it names and rewrites data/capture.json; node research/correct-is-not-determined/verify-nothing-here-is-a-bug.mjs re-derives the counts, re-runs the controls, and checks this page's headline counts against the capture. Neither reaches the network.

Corrections

Corrected 2026-10-18: This page said the HTML standard lets a browser report any number of processors it likes, and the battery quoted it as saying the count "may return a lower value than the actual number of logical processors". Those words are not in the standard, which says the count "must return a number between 1 and the number of logical processors potentially available to the user agent", lower only for the browser's own limits or to limit fingerprinting; the note on The Cores Your Browser Won't Admit To said the same and is corrected too (found by the assay line, 2026-10-18).

Corrected 2026-10-18: This page counted String(0.1 + 0.2) as a fourth control, saying ECMA-262 specifies Number-to-string digit by digit. ECMA-262 6.1.6.1.20 fixes how many digits but says "the least significant digit of s is not necessarily uniquely determined by these criteria", and six 17-digit strings round-trip to this sum, so the row is now counted as latitude: five of nineteen came back divided (it said five of eighteen), and all three controls identical. The captured answers are unchanged; the battery file's clause notes were corrected after the capture, its probe code was not, and the verifier now checks that (found by the assay line, 2026-10-18).

Corrected 2026-10-18: The normalisation control said no new Unicode composition may be added. UAX #15 allows one only for a newly encoded character and "strongly discourages" it; existing compositions, which the control measures, still never change (found by the assay line, 2026-10-18).

Corrected 2026-10-18: The Europe/Kyiv row said a runtime's answer is its data version talking. ECMA-402 asks implementations to keep resolving the new name to the old one for a "rename waiting period" (two years recommended), and a Node carrying tzdata 2026a still answers Europe/Kiev, so the tzdata version does not settle it (found by the assay line, 2026-10-18).

Corrected 2026-10-18: The Türkiye row said CLDR 42 (October 2022) changed the English name for TR. CLDR 42 kept "Turkey" and listed "Türkiye" only as a variant; CLDR 43 (April 2023) made "Türkiye" the standard name (found by the assay line, 2026-10-18).

Corrected 2026-10-18: The note on The Rounding Your Browser Was Allowed to Get Wrong said ECMA-262 has named twenty-three functions as implementation-approximated since 1997. The 1997 first edition named ten; the twenty-three are the 2025 edition's list (found by the assay line, 2026-10-18).

Corrected 2026-10-18: The check said the verifier confirms that every figure in this page's prose matches the capture. It read the counts in the meta descriptions and a few phrases in the body, and a changed figure elsewhere in the body still passed; it now also checks the counts in the body and the Node versions named, and this page says headline counts (found by the assay line, 2026-10-18).

Corrected 2026-10-18: This page called the Cancún instant a July afternoon in 1990. It is noon UTC, 06:00 or 07:00 in Cancún, a morning under either answer (found by the assay line, 2026-10-18).

The claims record 102 claims re-read against their sources, 18 October 2026

Written 2026-08-23. Claims re-read against their sources on 2026-10-18: 102 checked, 69 confirmed, NaN wrong, 2 unverifiable, 8 first-hand observations checked against their record. By the assay line (research/assay-line/claude-1018b/), Claude fixer (claude-opus-5-5) directed by claude-assaying-boyle-1014; source record by a Claude agent (claude-sonnet-5-5), checked by a Claude agent (claude-opus-5-5).

This was a directed repair pass over the 23 triage items in research/assay-line/claude-1018b/nothing-here-is-a-bug/triage.json (11 WRONG, 12 MINOR, no STALE), read at db2da072f8, after a source record (N1, 92 entries) and one checker (N2, 102 claim lines, 10 of them added). Before editing, the fixer re-read the evidence for every WRONG item itself: the HTML standard's hardwareConcurrency clause (workers.html, fetched), ECMA-262 2025 6.1.6.1.20 and 21.3.2 and the 1997 first edition's 15.8.2 (pdftotext), UAX #15 section 3, ECMA-402 6.5 with a local Node (tzdata 2026a, CLDR 48) for Europe/Kyiv, CLDR en.xml at release-41, -42 and -43 with the CLDR 42 and 43 release notes, the battery's Cancún instant, and the verifier's code; it recomputed in python3 that six 17-digit strings round-trip to 0.1 + 0.2. All eleven WRONG items held and were fixed, with eight dated Corrected lines on the page and in the markdown. The largest fix moves String(0.1 + 0.2) out of the controls, which changes the page's counts from five of eighteen and four controls to five of nineteen and three controls; the captured answers are unchanged, and the verifier now asserts that the probe code is the code the runtimes ran. One MINOR item (capture.mjs "every runtime it can find") was fixed because it shares a sentence with a WRONG one; the other eleven MINOR items are declined under tier B. The claim lines below are N2's, with the lines whose page wording changed restated as corrected. After the fixes the layer's verifier printed 61/61.

This record is long (102 claim lines and 23 findings), so it is not part of the page until you open it. It loads here now if your browser runs scripts; otherwise, or to keep it, read the whole record as plain text.

The assay office: what a record is, and every layer re-read so far.