Seeding formulas claim to measure readiness and mostly measure history

What a seeding formula is doing

A formula converts a general ranking into a surface-specific order by adding weight to results achieved on the relevant surface within a defined window. The window length, the weighting and the treatment of missing results are all decisions made by whoever designed the system. Those decisions are not derived from evidence about how much surface history predicts surface performance, because that evidence would be hard to assemble.

They are chosen to produce an order that looks reasonable to the people who must defend it, which is a different objective. Presenting the result as a more accurate measure conceals that a judgement has been embedded in an arithmetic wrapper.

The sample the formula draws on

Grass tournaments occupy a short block, so a player surface record within any window is built from a small number of matches. Small samples produce unstable estimates, and a player who was injured during one block carries a gap that the formula must interpret somehow. However the gap is handled, the handling is another choice with consequences for the resulting order.

A player returning from absence is therefore seeded on evidence that is missing rather than negative, and the two are treated as the same. That is a defensible convention and it is not a measurement of anything about the player current ability.

Why any formula is contestable

Several different weightings are equally defensible on their face, and they produce materially different orders near the boundaries that matter. Since the boundaries determine who meets whom, small formula choices produce large differences in the tournament that is actually played. A system whose output is sensitive to arbitrary parameter choices is not delivering the objectivity that justified its adoption.

This is not an argument for abandoning it, since the alternative of unstructured judgement is worse in obvious ways. It is an argument against describing the output as though it had been discovered rather than constructed.

The self-confirming loop

Seeding determines the draw, the draw determines the difficulty of the early path, and the path affects how far a player progresses. Progress then feeds back into the ranking that will be used the following year, which reinforces the original assessment. A player seeded generously receives a path that makes the seeding look correct, and a player seeded harshly receives the reverse.

The loop is not large enough to determine careers, and it is large enough to make the record a poor test of the formula. Any evaluation using subsequent results is therefore partly evaluating the effect of its own decisions.

A more honest framing

The formula could be described as a convention adopted to reduce disputes and to produce draws that most participants will accept as reasonable. That description is accurate, defensible and immune to the objection that the weights are arbitrary, because conventions are allowed to be. The trouble comes from claiming that the order reflects who is genuinely best prepared for the surface.

That claim invites exactly the scrutiny the formula cannot survive, and it was never necessary for the system to function. Most defences of seeding would be stronger if they claimed less. A convention needs only to be applied consistently and announced in advance, and neither of those requires it to be true about anything.