The Elo Lie — Chess Ratings Explained

2882
The highest chess rating ever recorded
Magnus Carlsen. May 2014. The peak of human chess achievement, measured in a single number.
Scroll

Bobby Fischer peaked at 2785.
Lower number. Must have been worse.

Right?

Fischer
1972
2785
Peak FIDE rating
+125
Points over #2
vs
Carlsen
2014
2882
Peak FIDE rating
+52
Points over #2

Fischer was more dominant relative to his era.
The raw number doesn't tell you that.

♔ ♛

The number is a lie.

Act I

The Gap

Forget the number itself. The gap between #1 and #2 measures dominance. How far ahead the best player stood from everyone else. It tells a completely different story.

Morphy appears

1858. Paul Morphy arrives in Europe and demolishes everyone. His Chessmetrics rating soars to 2799, a 158-point gap over the next best player. Chess has its first superstar.

Gap over #2: 158 points

Steinitz builds a fortress

1873–1888. Wilhelm Steinitz dominates for 15 years. His gap peaks at 186 points in 1888. A chasm no one alive could cross. He wasn't just winning. He was playing a different game.

Peak gap: 186 points (1888)

Capablanca's peak

1920. José Raúl Capablanca reaches 2887. 217 points above #2. The largest gap in the entire dataset. The "Chess Machine" loses only 34 serious games in his entire career.

Gap over #2: 217 points. The record

Fischer obliterates the field

1971. Bobby Fischer's Candidates run. 6-0 against Taimanov, 6-0 against Larsen, then crushes Petrosian. His Chessmetrics rating hits 2914. The gap: 184 points. Nobody is close.

Rating: 2914 · Gap: 184 points

Kasparov: the long reign

1985–2000. Garry Kasparov holds #1 for nearly two decades. His gap is large but never Steinitz-large. The field is deeper now. More strong players means smaller gaps, even for the greatest.

Peak gap: 104 points (1992)

Carlsen and the crowded top

2014. Carlsen hits 2882, the highest FIDE rating ever. But his gap over #2 is just 52 points. Several players are above 2800 simultaneously. The top has never been more crowded.

Gap over #2: 52 points. Smallest among greats

The full picture

Watch the gap shrink over 150 years. Not because champions got weaker, but because the field got deeper. More players, better preparation, engine training. Dominance itself became harder.

♞ ♞

But why did the gap shrink?

Act II

The Pool

Elo is a closed economy. Rating points are transferred between players, not created. Except the player pool keeps growing, pumping new points into the system. More players means more inflation.

Rated players worldwide

A gentleman's game

1970. FIDE's first rating list covers an estimated few hundred players. Chess is a niche pursuit. Soviet grandmasters, Western enthusiasts, a handful of prodigies. The rating pool is tiny.

A few hundred rated players

Fischer sparks a boom

The 1972 World Championship, Cold War on 64 squares, puts chess on front pages worldwide. The player pool starts growing. By 1985: ~4,600. By 1993: ~15,000.

15,000 rated players by 1993

FIDE lowers the floor

FIDE systematically lowers the minimum published rating: 2000 in 1993, 1800 in 2001, 1600 in 2006, and 1000 from July 2012. Each drop floods the pool with new players and new points.

Rating floor: 2200 → 1000

The online explosion

Chess.com, Lichess, and the pandemic boom drive FIDE registrations. By 2019: 352,000 standard-rated players. By 2025: over 500,000. The pool is hundreds of times larger than Fischer's era.

500,000+ rated players in 2025

The inflation engine

Every new player entering the pool brings rating points with them. As the pool expanded by orders of magnitude, the top ratings climbed with it. A 2700 in 1975 (1 player) and a 2700 in 2025 (~30 players) are not the same achievement.

♜ ♜

The number didn't change.
Everything around it did.

Act III

The Shape Shifts

This is the chart nobody shows you. Not just the top rating, but the entire distribution of rated players. Watch the bell curve shift rightward over 50 years. The whole landscape of chess ratings has moved.

Top players in pool

1900: a thin peak

51 players in the dataset. The distribution is thin, peaked around 2300. The tail barely reaches 2800. Lasker sits alone at the far right. A statistical outlier.

51 players · #1: Lasker

1950: the Soviet machine

181 players. The USSR dominates chess, producing a generation of strong players. The curve widens. The center shifts right. More players above 2500 than existed in total in 1900.

181 players · Peak: 2821 (Botvinnik)

1970: Fischer's mountain

500 players tracked. The distribution fills out. A proper bell curve now. But Fischer is so far right he's almost off the chart. The gap between him and the curve is the gap between genius and everyone else.

500 players · Fischer at 2860. An outlier

1990: the curve shifts

The entire distribution has moved right. The minimum in the top 500 has risen from 2229 to 2448. The center of mass shifts 200 points. This is inflation, visible in the shape itself.

Floor of top 500: 2229 (1970) → 2448 (1990)

2000: the new normal

The floor hits 2477. The peak has shifted from ~2350 to ~2500. What was once exceptional is now ordinary. The entire scale has moved, and the number at the top moved with it.

2500 in 2000 ≈ 2350 in 1970
Act IV

The 2700 Club

2700 used to mean you were one of the best players alive. Now it means you're one of ~30. The threshold didn't move. The population behind it did.

Players rated 2700+

1979: just one

Anatoly Karpov. The only player on Earth rated above 2700. The 2700 club has exactly one member.

2700+ club: 1 player

1994: a handful

Kasparov, Karpov, Anand, Kramnik, Ivanchuk, Shirov. Six players above 2700. Still exclusive. Still elite. Still meaningful.

2700+ club: 6 players

2009: the flood

33 players above 2700. The club isn't exclusive anymore. 2700 no longer means "one of the best alive." It means "strong grandmaster." The bar has been silently redefined.

2700+ club: 33 players

2025: the new baseline

~30 players above 2700 at any given time. 133 players have broken 2700 at some point in history. Only 15 have ever broken 2800. The real elite threshold has shifted to 2800.

2700+ ever: 133 · 2800+ ever: 15
♚ ♕

So if the number lies,
what's the truth?

Act V

The Correction

Jeff Sonas built Chessmetrics, a system designed to compare players across eras on a level playing field. Same methodology, applied consistently from 1851 to the present. It tells a different story.

Player Chessmetrics Peak Gap over #2 Year
Capablanca 2921 +217 1920
Fischer 2914 +184 1971
Lasker 2886 +119 1910
Kasparov 2886 +104 1992
Carlsen 2882 +52 2014
Morphy 2840 +158 1864

Notice: Carlsen's FIDE 2882 and his Chessmetrics 2882 are the same number, but in Chessmetrics, he ranks 5th, not 1st.

The system wasn't designed to crown a winner. It was designed to show what the raw number hides.

Act VI

So Who's the Best?

It depends on what you measure. Pick a metric. Each one crowns a different king.

The number was never the answer. The question was.

2882 isn't a fact about Magnus Carlsen. It's a fact about the system that measured him. A system built for a few hundred players, now serving 500,000. A system that's been patched, adjusted, and debated for over 60 years.

The greatest chess player of all time depends entirely on what you think "greatest" means. The Elo rating never told you that. Now you know.

Buy me a coffee