# The Elo Numbers Lie. Here's What's Actually Happening to Chess Skill.
Everyone says modern chess is stronger. The ratings prove it, right? Elo averages keep climbing. Players hit 2700 faster than ever. But Dr. Robert Howard's research into decades of rating data reveals something uncomfortable: we might be measuring the wrong thing entirely.
The core problem is Elo inflation. As the player pool expands and becomes more competitive, rating numbers drift upward like currency losing value. A 2600 today does not play the chess that a 2600 played in 1995. This is not controversial among statisticians. It is controversial among players who have those ratings.
Howard examined historical data across multiple decades. The pattern is clear. Players are not necessarily stronger in absolute terms. They are better at specific things. They memorize deeper opening theory. They calculate tactical sequences faster because computers showed them the patterns. They avoid catastrophic blunders more consistently. But do they understand the game better? Do they find ideas that humans had not found before? The evidence murkier.
This distinction matters because it changes how we think about Kasparov, Fischer, and Carlsen. Comparing their peak strength across eras has always been pointless. Now we have data showing why.
The research also challenges the romantic notion that chess talent is fixed at birth. Learning curves tell a different story. A player's improvement trajectory depends heavily on access to coaching, computer training, and play against strong opposition. Natural talent matters. It is not everything. Howard found that deliberate practice and environmental factors account for more variance than most chess culture admits. The kids grinding eight hours a day on Lichess are not discovering some innate advantage. They are doing what works.
Age at peak performance is another revelation. Players traditionally peaked in their early thirties. That window has narrowed. Top players now flame out or stagnate in their late twenties more often than before. The explanation is computational. Once you have absorbed what computers teach, the marginal returns disappear. You reach a ceiling faster because there is a ceiling now. Kasparov played opponents who had intuitions he did not possess. Modern super-GMs play computers that have already explored all reasonable moves in any position. The learning curve flattens.
The computer revolution does not appear in these statistics as a sudden rupture. It shows up as a slope change. Before engines became dominant analytical tools, chess improvement accelerated. After, it accelerated again, but differently. Players improved faster at narrow, tactical tasks. They improved slower at strategic depth and originality. Computers are excellent teachers of what they know. They are useless at teaching what they do not know yet.
One finding deserves particular attention: the current cohort of young players (under 25) is not stronger than the cohort that emerged in the 1990s and 2000s. They are better prepared. They arrive at the board with more memorized lines and tactical pattern recognition. But when forced into original positions, the data suggests they struggle more than players from earlier eras did. This could explain why recent world championship matches have often been decided by one side making an error rather than one side demonstrating superiority.
The implications for how we train players are substantial. Coaching focused entirely on engine analysis and opening preparation might be reaching diminishing returns. Players are already saturated with that input. What they lack is the kind of positional instruction and strategic intuition that requires human teachers, historical game study, and uncomputer positions. A talented 15-year-old in 2024 knows more opening theory than Karpov did. That kid probably understands less about pawn structures.
None of this diminishes modern chess. It simply describes it more accurately. The question "are players getting stronger?" is the wrong question. The better questions are "stronger at what?" and "what are we sacrificing to get there?"
As ratings keep climbing and players keep getting younger, Howard's research suggests we should stop taking the numbers as gospel. They measure something. They do not measure what we think they measure.