Seeding is a projection, and every projection carries assumptions

What a seed is actually for

Seeding exists to stop the strongest entrants meeting each other early, which protects the competitive shape of a tournament rather than rewarding any individual. Because the goal is structural, the seeding list only has to be approximately right about relative strength to do its job adequately. The problem is that spectators read it as a ranking of likely winners, which asks far more precision of it than the mechanism was ever designed to deliver.

A seed number is therefore best understood as a position in a bracket-building exercise rather than a prediction about a particular player's fortnight. That distinction matters because it changes what counts as a seeding failure, and most so-called upsets are nothing of the kind.

Rankings are a decaying memory

Ranking systems accumulate points over a rolling period, which means a player's position reflects an average of the recent past rather than their present standard. A player improving quickly is therefore under-ranked, since the system is still carrying results from a period when they were a weaker competitor. A player declining is over-ranked for the same reason, holding points earned in circumstances that no longer describe how they are competing.

The lag is largest immediately after any long absence, because the accumulation window contains a stretch during which no points could be earned at all. Seeding built on that foundation inherits every one of these distortions, and it inherits them silently.

Why surface weighting exists

Some tournaments adjust seeding to account for performance on the surface being played, because a general ranking blends surfaces that reward very different games. The adjustment is an admission that the underlying ranking answers a broader question than the one a single event actually needs answered. Any such weighting introduces its own choices about how far back to look and how heavily recent surface-specific evidence should count.

Those choices are defensible but arbitrary, and different weightings would produce visibly different brackets from exactly the same set of entrants. The existence of the debate is itself informative, since it shows that seeding is a modelling decision rather than a measurement.

Where the projection breaks

Seeding assumes that relative strength is roughly stable across the fortnight, which fails whenever a player's condition or confidence is changing rapidly. It also assumes strength is transitive, when in practice a lower-ranked player can be a poor matchup for someone several places above them. Entry rules add another wrinkle, since players returning from absence may enter through protected mechanisms that place them outside the seeding entirely.

The combined effect is a bracket where the genuinely dangerous unseeded entrants are concentrated unpredictably rather than spread evenly. That concentration is the main reason one quarter of a draw can be substantially harder than another despite identical seeding on paper.

Reading a seeding list usefully

The useful reading treats the seeding as a rough ordering with wide error bars rather than a ranked list of expectations for the tournament. Attention is better spent on which unseeded entrants have games that trouble specific seeds, since that is where the bracket's real asymmetry lives. It is also worth noting how recently each seed's points were earned, because a position built largely on older results carries more uncertainty.

None of this makes seeding a bad system, since it performs its structural job perfectly well across a whole tournament. It simply means the number beside a name is a piece of tournament administration and not a verdict on how the fortnight will unfold.