Rubric 1.0.0 · column 3 of 13
Device pixel ratio
The ratio between physical device pixels and CSS pixels. 1 on a standard display, 2 on most retina-class hardware, fractional under OS display scaling.
Under rubric 1.0.0 this column is worth between 1.00 and 5.64 bits, and the likeliest outcome the table knows — 1 — costs 1.00. 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 device pixel ratio
The number is a quotient, not a hardware property: physical pixels divided by the CSS pixels the browser has decided to lay out in. It is produced jointly by the panel, the operating system scale factor and the browser zoom level, which is why it is one of the few signals a visitor can change by accident. On most engines a zoom level other than 100% changes it immediately, so a reading of 1.1 can mean an unusual display scale or a browser that was zoomed once and never set back.
How rubric 1.0.0 prices the device pixel ratio
Three shapes, four outcomes. Exactly 1 is assumed at 0.5 and costs 1.00 bit. Exactly 2 is assumed at 0.28 and costs 1.84. Any other whole ratio — 3, 4 — drops to 0.04 and 4.64 bits. Anything fractional lands in a single bucket at 0.02, or 5.64 bits, whether it is the extremely ordinary 1.25 that Windows scaling at 125% produces or a value like 1.7639 that almost nothing else would produce.
The spread between the cheapest and the dearest outcome is 4.64 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 dpr has.
| Outcome | Assumed p | −log₂p |
|---|---|---|
| 1 modal outcome — the printed line | 0.500 | 1.00 |
| 2 | 0.280 | 1.84 |
| 3 any other whole ratio | 0.040 | 4.64 |
| 1.25 any fractional ratio from OS display scaling | 0.020 | 5.64 |
What changing the device pixel ratio costs
Setting the ratio to a flat 1 or 2 is the cheapest correction on the whole form and it saves at most 4.64 bits. The cost is arithmetic: the pixel ratio has to make sense against the screen resolution it sits beside, because both are read in the same pass and both feed the display group. A 1512x982 screen at a ratio of 1 describes a physical panel that no vendor ships, and that mismatch is checkable by anyone who cares to check.
The rubric's own reasoning, verbatim
Effectively three buckets — 1, 2, and everything else. Fractional ratios come from OS display scaling and are individually uncommon.
What moves it
Changing the OS display scale moves it. Browser zoom moves it too on many engines, which is worth knowing before you conclude a tool changed something.
Where this prior is weakest
The fractional bucket is a single prior covering every value from 1.1 to 3.5. Windows scaling at 125% is enormously common and is priced the same as 1.7639, which is nearly unique. This bucket is too coarse and we would replace it first.
Correlation
Grouped with Screen resolution and Color depth under "display". 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 dpr →
What people ask about the device pixel ratio
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 did my score change when I zoomed the page?
- Because browser zoom changes window.devicePixelRatio on most engines. A zoomed page can move this column from the 1.00-bit bucket into the fractional bucket at 5.64 bits without anything about your machine having changed.
- Is a fractional pixel ratio bad?
- Under this rubric it costs 5.64 bits, the same as any other fractional value. Whether that matters depends on whether the rest of your display group is louder, since only the strongest member of the display group counts in full.
- Should I force my pixel ratio to 1?
- Only if the screen resolution you report is one a real display produces at that ratio. The two are read together, and a display geometry that does not exist is a worse outcome than the four bits you saved.
Where to read next
- The prior for screen resolution — the other reading the display group discounts this one against.
- The prior for color depth — the other reading the display 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.