A Chess Formula Is Taking Over the World (theatlantic.com) 28
An anonymous reader quotes a report from The Atlantic: In October 2003, Mark Zuckerberg created his first viral site: not Facebook, but FaceMash. Then a college freshman, he hacked into Harvard's online dorm directories, gathered a massive collection of students' headshots, and used them to create a website on which Harvard students could rate classmates by their attractiveness, literally and figuratively head-to-head. The site, a mean-spirited prank recounted in the opening scene of The Social Network, got so much traction so quickly that Harvard shut down his internet access within hours. The math that powered FaceMash -- and, by extension, set Zuckerberg on the path to building the world's dominant social-media empire -- was reportedly, of all things, a formula for ranking chess players: the Elo system.
Fundamentally, what an Elo rating does is predict the outcome of chess matches by assigning every player a number that fluctuates based purely on performance. If you beat a slightly higher-ranked player, your rating goes up a little, but if you beat a much higher-ranked player, your rating goes up a lot (and theirs, conversely, goes down a lot). The higher the rating, the more matches you should win. That is what Elo was designed for, at least. FaceMash and Zuckerberg aside, people have deployed Elo ratings for many sports -- soccer, football, basketball -- and for domains as varied as dating, finance, and primatology. If something can be turned into a competition, it has probably been Elo-ed. Somehow, a simple chess algorithm has become an all-purpose tool for rating everything. In other words, when it comes to the preferred way to rate things, Elo ratings have the highest Elo rating. [...]
Elo ratings don't inherently have anything to do with chess. They're based on a simple mathematical formula that works just as well for any one-on-one, zero-sum competition -- which is to say, pretty much all sports. In 1997, a statistician named Bob Runyan adapted the formula to rank national soccer teams -- a project so successful that FIFA eventually adopted an Elo system for its official rankings. Not long after, the statistician Jeff Sagarin applied Elo to rank NFL teams outside their official league standings. Things really took off when the new ESPN-owned version of Nate Silver's 538 launched in 2014 and began making Elo ratings for many different sports. Some sports proved trickier than others. NBA basketball in particular exposed some of the system's shortcomings, Neil Paine, a stats-focused sportswriter who used to work at 538, told me. It consistently underrated heavyweight teams, for example, in large part because it struggled to account for the meaninglessness of much of the regular season and the fact that either team might not be trying all that hard to win a given game. The system assumed uniform motivation across every team and every game. Pretty much anything, it turns out, can be framed as a one-on-one, zero-sum game. Arpad Emmerich Elo, creator of the Elo rating system, understood the limitations of his invention. "It is a measuring tool, not a device of reward or punishment," he once remarked. "It is a means to compare performances, assess relative strength, not a carrot waved before a rabbit, or a piece of candy given to a child for good behavior."
Fundamentally, what an Elo rating does is predict the outcome of chess matches by assigning every player a number that fluctuates based purely on performance. If you beat a slightly higher-ranked player, your rating goes up a little, but if you beat a much higher-ranked player, your rating goes up a lot (and theirs, conversely, goes down a lot). The higher the rating, the more matches you should win. That is what Elo was designed for, at least. FaceMash and Zuckerberg aside, people have deployed Elo ratings for many sports -- soccer, football, basketball -- and for domains as varied as dating, finance, and primatology. If something can be turned into a competition, it has probably been Elo-ed. Somehow, a simple chess algorithm has become an all-purpose tool for rating everything. In other words, when it comes to the preferred way to rate things, Elo ratings have the highest Elo rating. [...]
Elo ratings don't inherently have anything to do with chess. They're based on a simple mathematical formula that works just as well for any one-on-one, zero-sum competition -- which is to say, pretty much all sports. In 1997, a statistician named Bob Runyan adapted the formula to rank national soccer teams -- a project so successful that FIFA eventually adopted an Elo system for its official rankings. Not long after, the statistician Jeff Sagarin applied Elo to rank NFL teams outside their official league standings. Things really took off when the new ESPN-owned version of Nate Silver's 538 launched in 2014 and began making Elo ratings for many different sports. Some sports proved trickier than others. NBA basketball in particular exposed some of the system's shortcomings, Neil Paine, a stats-focused sportswriter who used to work at 538, told me. It consistently underrated heavyweight teams, for example, in large part because it struggled to account for the meaninglessness of much of the regular season and the fact that either team might not be trying all that hard to win a given game. The system assumed uniform motivation across every team and every game. Pretty much anything, it turns out, can be framed as a one-on-one, zero-sum game. Arpad Emmerich Elo, creator of the Elo rating system, understood the limitations of his invention. "It is a measuring tool, not a device of reward or punishment," he once remarked. "It is a means to compare performances, assess relative strength, not a carrot waved before a rabbit, or a piece of candy given to a child for good behavior."
I think I speak for everyone when I say... (Score:4, Insightful)
What?
Don't bring me down (Score:2)
Yep
ELO had nothing to do with Musk, it was around before he was even born
Re: (Score:1)
What?
Oh don't get excited, it's just one of the regularly scheduled reminders to all of us of exactly what a colossal amoral mean spirited scumbag and menace to human civilisation Mark Zuckerberg is.
Re: (Score:3)
So, this system has been around for centuries.
Specifically, Sumo uses it today and has for a very long time (well at least as recorded for 100 years, but probably longer).
If you are in a top-level sumo tournament (or basho), you will get "billing" in the tournament based on your past performance, meaning, you will likely face lower-ranked opponents early on if you have just been promoted to the top level of sumo (Makuuchi). If, by some crazy turn of events you end up facing a Yokuzuna, top highest ranking
Re: (Score:2)
Sumo is an interesting example because while nonminimally it is based on a statistical ranking system, in reality the bosses decide matches to make the tournaments interesting and to reward or punish individual wrestlers based on how they perceive the amount of effort and sportsmanship they are putting in.
Often recently ranked up wrestlers will face a Yokozuna early on in the tournament, for example. They do it that way because they don't expect them to win, and are holding back the likely title contenders
Predictive power? (Score:2)
Re: (Score:3)
I predict (Score:2)
Re: (Score:3)
was written by EloGPT, it's the in thing
Widely adopted, but how good is it? (Score:3)
The goal of the Elo system is to predict who will win a chess game based on their difference in Elo ratings. There's really two components to this - 1) deriving the player ratings, and 2) predicting what outcome to expect for a given difference in ratings.
Nowadays with machine learning we could easily learn a rating system that directly optimizes the goal of predicting match outcomes. It'd be interesting to do it and see how much better at predicting it would be than the Elo system.
Re: (Score:2)
nah, Elo is a compromise between predictive power and transparency/intuitiveness and it is mostly for maintaining a rating of players over time, rather than "predicting who will win" any given match.
ML can trivially beat Elo at one-shot prediction of chess games even using nothing but past match data. I have trained such a system, as have many others; it is not difficult, but it would really suck for a scored rating system.
apart from a few corner cases which most orgs account for one way or another, your ch
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Yeah, I was thinking of something like a recommender system where one could jointly learn player attributes (strengths, weaknesses) and match-up outcomes, but certainly there would be little transparency/explainability.
It would be interesting though to distill these types of attributes into a scalar Elo-like number, predictive of outcomes, and see how closely that aligns with Elo ratings.
Re: Widely adopted, but how good is it? (Score:2)
here, have fun!
https://www.kaggle.com/c/chess [kaggle.com]
Ten (Score:3, Funny)
In the movie, the written formula made it look like a sub-expression was multiplied by ten, when it fact the sub-expression is an exponent on a base of ten. The stage hand drawing it from notes probably thought the "disjointed levels" was merely sloppy writing, and "corrected" it. Rumor has it he later went to work for Boeing.
Scissors/Paper/Rock (Score:2)
Does that come up in chess?
AI? (Score:2)
How well does it work for rating competing nets or even adjusting weights within networks?
Play to win the metrics (Score:2)
Seen this throughout my working career.
When you have metrics that measure a persons worth, they will work towards the metric and not what you are measuring.
Just based on this description, no one is going to play against anyone who can potentially drop their score.
Re: (Score:2)
In chess your elo will drop when you don't win, regardless of your opponents elo.
The fun is lost if that's all you're thinking about.
Play to learn, not for points.
Football rankings (Score:2)
Many things seem like quantized partial diff eqs (Score:2)
Extra, Extra! (Score:2)
Re: (Score:2)
Overgeneralisation (Score:2)
But not necessarily sensibly. Framing anything which isn't zero-sum as a zero-sum game leads to suboptimal decisions. Framing one-on-one-on-one-on-one games as six simultaneous one-on-one games violates the assumption of independence.