The first reading

If you take one reading and stop, it is probably the best number you were ever going to get. That is not a compliment to your technique - it is a predictable pattern, and knowing it changes how a single reading should be treated.

Why the first attempt tends to run high

Nobody presses a ruler into their own body with a fixed, pre-decided force. The pressure, the exact angle, and where the endpoint gets called all involve small choices, and on a first attempt those choices are made without any prior reading to compare against - there is nothing yet pulling the outcome toward consistency. Left with a genuinely free choice and no reference point, most people's hands settle on the combination of angle and pressure that produces the more favourable number, without any conscious intention to do so. This is not dishonesty; it is closer to the same effect that makes a first guess at a jar of coins skew optimistic before anyone starts counting. The same pattern shows up in any self-administered measurement with no fixed protocol behind it yet - a first attempt at anything tends to land wherever the free choices happen to fall, and only repetition narrows that down.

Regression on the readings that follow

Once a first number exists, it becomes a reference the next attempt is implicitly measured against, and the readings after it tend to drift back toward whatever the method actually supports rather than repeating the first result. Some of that drift is fatigue or reduced motivation to optimise the position on a second or third try. Some of it is a purely statistical effect: an unusually favourable outcome, however it was produced, is followed on average by something closer to the middle, simply because extremes are less likely to repeat than moderate results are. Both push the same direction, which is why a sequence of careful readings taken in one session tends to start high and settle rather than wander randomly around a stable centre.

What this means for a single reading

A single measurement, taken once and recorded, is more likely than not to be the flattering end of what your own technique produces - not necessarily wildly so, but the bias runs one direction rather than none. It is a different problem from the viewing-angle distortion that shifts where a self-measurer reads the tip mark against the scale, described separately, though the two can compound on that first attempt.

The fix is repeats, not more effort on the first try

Trying harder to be objective on a single reading does not solve this, because the bias operates below the level of conscious choice. What works is what the standard method already specifies: take several readings in the session rather than one, and use the median rather than the first or the highest. Discard nothing - a reading that looks like an outlier is data about your own spread, and stating that spread honestly as an uncertainty is more useful than pretending a single confident number has none. The point is not to find the "real" reading among several candidates; it is to let a first-attempt bias get diluted by readings that were not taken under the same novelty.

Why this is worth separating from other reading problems

This is a different failure from misreading a mark between millimetre lines, and different again from the angle a self-measurer views their own ruler from - each has its own fix, and treating them as one problem tends to leave at least one uncorrected. None of it is about assessment rather than measurement: the number a ruler gives you is a length, full stop, not a verdict, and a verdict is what services built for that purpose provide - a rating drawn from photographs or a person's direct opinion answer a different kind of question, and an image model estimating from a photo has no readings to regress across in the first place, since it never took a physical measurement at all. If a subjective read is genuinely what you are after, Rate Cock is built for exactly that; a ruler's first-reading bias is a narrower problem than the one it solves.