Rubric 1.0.0 · column 1 of 13
Operating system
The operating system, read out of the user agent string rather than from any dedicated API. Six outcomes: Windows, macOS, Android, iOS, ChromeOS, Linux.
Under rubric 1.0.0 this column is worth between 0.69 and 5.64 bits, and the likeliest outcome the table knows — Windows — costs 0.69. 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 operating system
There is no operating-system API in a browser. The rule reads a token out of the user agent string the browser sends and volunteers to JavaScript — a string the vendor writes, freezes for years at a time, and increasingly lies in on purpose. The order of the tests matters and is visible in the shipped code: Windows first, then macOS, then Android, then iOS, then ChromeOS, then Linux. An Android user agent contains the word Linux, so if Linux were tested earlier every Android device on earth would score as a Linux desktop. That ordering is not a detail, it is the rule.
How rubric 1.0.0 prices the operating system
Six buckets, priced from broadly reported desktop share and rounded hard. Windows is assumed at 0.62 and costs 0.69 bits — the cheapest reading anywhere in this table. macOS is 0.18 and costs 2.47. Android is 0.10, iOS is 0.07, and ChromeOS and Linux share a floor of 0.02, which is 5.64 bits each. A browser that presents no recognisable token at all is not scored zero: the rule returns null, the pen lifts, and the column is excluded from the sum.
The spread between the cheapest and the dearest outcome is 4.95 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 platform has.
| Outcome | Assumed p | −log₂p |
|---|---|---|
| Windows modal outcome — the printed line | 0.620 | 0.69 |
| macOS | 0.180 | 2.47 |
| macOS | 0.180 | 2.47 |
| Android | 0.100 | 3.32 |
| ChromeOS | 0.020 | 5.64 |
| Linux | 0.020 | 5.64 |
What changing the operating system costs
This is the single cheapest signal to change and the single most expensive one to change badly. Declaring Windows while your GPU renderer names an Apple M-series adapter, your fonts include Helvetica Neue and your touch-point count matches a MacBook is not a lower score, it is a contradiction — and a contradiction is evidence of intent in a way that a rare value never is. Moving from Linux to Windows saves you 4.95 bits on this form and buys you a set of claims the rest of the machine has to keep.
The rubric's own reasoning, verbatim
Desktop OS share is the best-documented signal here. Windows dominates, macOS is a strong second, Linux is a long tail. Assumed from broadly reported desktop share, rounded hard.
What moves it
Almost nothing you do in a browser moves this. Changing the UA string moves it immediately and moves nothing else, which is exactly the pattern a consistency check looks for.
Where this prior is weakest
Desktop OS share is the best-documented figure in this whole table, and it is still a global average applied to a visitor who is not a global average. Anyone reading a scraping tool's landing page is far more likely to be on Linux than the world is. This prior is almost certainly too generous to Windows for this audience.
Correlation
The only member of the "platform" group, so it always takes full weight. That is worth knowing when you compare it against a grouped signal: no discount is protecting the total from 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 platform →
What people ask about the operating system
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 is Linux worth more bits than Windows here?
- Because the rule assumes it is rarer. Windows sits at an assumed 0.62 and Linux at the table floor of 0.02, so Linux costs 5.64 bits against Windows at 0.69. That is a statement about the assumption, not about how many Linux users exist.
- Does spoofing the user agent lower my total?
- It lowers this column and nothing else, because nothing else in the rubric reads the user agent. The saving is at most 4.95 bits, and it is undone the moment another signal disagrees with the operating system you just declared.
- What happens if the operating system cannot be identified?
- The rule returns null. The pen lifts, the column is listed as not read with its reason, and it contributes nothing at all — it is never scored as zero bits, which would quietly claim that an unreadable browser is an ordinary one.
Where to read next
- 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.