Two kinds of soft, two different effects

Two blurry photos are not blurry in the same way, even when they look equally soft to the eye. Motion blur and defocus blur come from different physical causes, and they leave different marks on the image a model actually processes.

Two different smears

Motion blur happens when the subject or the camera moves during the exposure, smearing detail in the direction of that movement. It has a direction and, often, a fairly consistent streak length across the frame, because the motion that caused it was usually one continuous movement. Defocus blur happens when the lens is not focused at the subject's distance, spreading each point of light into a small disc rather than a streak, with no preferred direction at all. To a person, both just look "not sharp." To a model reading spatial-frequency information - the fine, high-frequency detail that edges and texture depend on - the two are structurally different signals, because a directional smear and a symmetric disc remove different frequencies in different patterns across the image.

Why the distinction reaches the score

Texture-sensitive axes read high-frequency detail as part of what they respond to, and both blur types remove that detail, but not identically. Motion blur tends to preserve some edge information along its direction of travel while destroying it perpendicular to that direction, so a model can sometimes still register a strong edge that happens to run parallel to the motion. Defocus blur removes detail more uniformly in every direction, which tends to read as a flatter, more generally "soft" image overall. The sensitivity is documented: Vasiljevic, Chakrabarti and Shakhnarovich (2016) found recognition networks lose significant accuracy on images blurred by "defocus, or subject or camera motion", and that training on blurred examples improves robustness. A scorer that never saw much blur during training has no such cushion, and it will not tell you which kind it is reacting to.

Why one gets caught more reliably

Cheap pre-scoring blur checks generally work by measuring the amount of high-frequency content left in an image, on the reasoning that a sharp photo has more of it than a blurry one. Defocus blur removes that content fairly evenly, which is exactly the pattern such a check is built to catch. Motion blur's directional, partially-preserved edges can slip past a check tuned for uniform softness, particularly at shorter blur lengths, because there is still meaningful high-frequency content in the frame, just not in every direction. What quality gates do before scoring even starts is its own subject and worth reading for the mechanism; the point here is narrower - the two blur types are not equally likely to be caught by the same threshold, and a motion-blurred photo can reach the scoring model when an equivalently soft defocused one would not.

What to take from it

If a retake looks about as sharp as the original but scores differently on a texture-related axis, the type of blur, not just its amount, is worth considering - a slight motion smear and a slight defocus are not interchangeable inputs even at similar apparent softness. The broader list of variables that move a score between similar-looking attempts covers blur alongside lighting, angle and crop, and this is the detail underneath that one entry. Reading one result in light of that spread, rather than reacting to it in isolation, is exactly what Penis Rater writes about for users trying to make sense of a single number. Practical shooting advice, like how to avoid either kind, belongs to a tool's own guide rather than to the mechanism: Rate Cock's photo guide covers that ground. A human reviewer usually just says "this is blurry" without needing to diagnose which kind - Rate Penis's judging notes describe a process built around that kind of plain read - and anyone comparing repeat shots for consistency will find Measure My Cock's gear guidance covers stabilising a shot from the equipment side, which addresses both blur types at once.