Rubric 1.0.0 · column 8 of 13
Language preferences
navigator.languages, in order. The order matters — it is part of the value, not a set.
Under rubric 1.0.0 this column is worth between 1.56 and 8.38 bits, and the likeliest outcome the table knows — en-US — costs 1.56. Those figures are assumptions printed before your run. The counted box beside them is not: it is how many of the devices these sites have read share your value.
How a browser produces the language list
The list is assembled from the browser language preference and the operating system locale settings, in the order the user ranked them, and it is sent both as a JavaScript array and as the Accept-Language request header. Two surfaces, one setting: a tool that rewrites the array in the page and leaves the header alone has produced a disagreement between two copies of the same fact, which is cheaper to detect than anything on this chart.
How rubric 1.0.0 prices the language list
Four outcomes driven almost entirely by length. The exact strings en-US and en-US, en are assumed at 0.34 and cost 1.56 bits. Any other single-entry list drops to 0.03 and 5.06 bits. Two entries sit at 0.012 for 6.38 bits, and three or more at 0.003 for 8.38 bits — the second-highest single reading this rubric can produce. An empty list is not scored at all: the rule returns null and the pen lifts.
The spread between the cheapest and the dearest outcome is 6.82 bits, and that spread is the whole of what this column can ever do to a total. Where the four band edges sit →
Every outcome this rule can produce
This is the printed envelope, written out. The pen cannot land outside it, because these are
all the answers rule languages has.
| Outcome | Assumed p | −log₂p |
|---|---|---|
| en-US modal outcome — the printed line | 0.340 | 1.56 |
| de-DE any single non-en-US list | 0.030 | 5.06 |
| de-DE, en any two-entry list | 0.012 | 6.38 |
| de-DE, en-GB, fr any list of three or more | 0.0030 | 8.38 |
What changing the language list costs
Trimming a three-language list to en-US saves 6.82 raw bits, the largest single saving available anywhere on this form. It is also the change most likely to contradict something: your time zone, your proxy exit country, and the Accept-Language header your requests carry. Language preferences are grouped with time zone, so if the zone is the louder of the two you keep only 35% of the saving anyway — and you keep all of the contradiction.
The rubric's own reasoning, verbatim
A single `en-US` is by far the most common list. Every additional language multiplies the possibilities, so longer lists are treated as much rarer.
What moves it
Browser language settings. Emptying the list entirely does not help: an empty navigator.languages is itself an automation indicator that this site's sibling tools flag.
Where this prior is weakest
Ordering is treated as significant but length is what actually drives the prior, so `de-DE, en` and `en, de-DE` are priced identically at 6.4 bits when they are not equally common. A measured table would separate them.
Correlation
Grouped with Time zone under "locale". Whichever of them carries the most surprisal in a given pass takes full weight; the others take 35%. The chart draws the short stroke rather than describing it.
How the correlation discount works →
Challenge it
If you have a source that contradicts this assumption, the honest thing is to say so in
public against this specific prior. Open a challenge
against languages →
What people ask about the language list
Answered from rubric 1.0.0 itself. Where an answer quotes a probability, that probability is a constant in the shipped scoring code and is printed in the table above.
- Why does a long language list cost so much?
- The rule assumes that each additional entry multiplies the number of possible lists, so it prices three or more entries at 0.003, or 8.38 bits. That is an argument about combinatorics, not a count of anyone.
- Does an empty language list lower my score?
- It removes the column from the sum entirely, because the rule declines to read an empty list. It also produces an empty navigator.languages, which is one of the oldest automation indicators there is and costs far more than the bits it saved.
- Does the order of my languages matter?
- It is part of the value, so two orderings of the same pair are different strings. Under rubric 1.0.0 they are nonetheless priced identically, because length rather than order drives the prior. That is a stated weakness of this rule.
Where to read next
- The prior for time zone — the other reading the locale group discounts this one against.
- The complete prior table for all 13 columns — every rule, every outcome, and the rationale the rubric ships with each one.
- Assumed and counted: what a prior is, what the counted column beside it measures, and why the two are never added.
- Compare two passes for the delta on this column — change one setting and read what actually moved, instead of guessing from a total.
Where the pens move
Knowing which prior is weak is only useful if something can actually move the signal.
Engine level, not JS level. Donut Browser spoofs through CDP injection in its Wayfern Chromium fork, so the value a page reads is the value the engine reports — there is no patched accessor to catch.
It refuses to contradict itself. A profile whose proxy exit geography, time zone or language disagrees with its fingerprint does not launch. That is a pre-launch gate in the app, not advice in a manual.
Open, and quiet. AGPL-3.0, desktop app and sync server both in the repository, and no analytics SDK anywhere in the source. macOS, Windows, Linux.