The games I’m thinking of are abstract strategy board games such as chess, which I’m superficially familiar with, and go, which I play on a regular basis; but the same logic might also apply to others, such as checkers.
I’m wondering whether there is a long-term tendency for the overall level of play and skill in the communities playing these games to increase over time. I wouldn’t be surprised if there were; after all, players benefit from the experience of observing and analysing high-level championship and tournament games played by others, as well as by the ample literature that keeps being published.
The usual numerical methods of capturing a player’s skill, such as Elo numbers in chess or the dan/kyu ranking system in go, wouldn’t capture such a tendency - these numbers are relative, in the sense that they are derived from assessing a player’s score against the score of other players he or she has won or lost against, so a general increase in the overall level of play across the community would leave individual player’s scores unaffected (much like the Flynn effect, since IQ is also only a number that expresses how well you do in an IQ test relative to others).
So it’s difficult to detect such an effect mathematically, and obviously we can’t have today’s players play games against the champions of a generation ago. But if the effect existed, it should certainly be noticeable to a skilled observer. So does it exist? Would, say, a Capablanca today be considered just a good but not outstanding player, as opposed to the shining figure he was 100 years ago?
I would think that, yes, as long as a game remains popular, that over time standards will gradually improve, and for the reason that you suggested: that great players and analysts continually improve our understanding of the game.
We also have computers using either brute force or deep learning algorithms to analyze games and suggest improvements. If this seems like cheating, bear in mind one probable reason for the standard, IQ-based, Flynn effect, is access to information technologies to allow us to more efficiently communicate and build on ideas.
For the question about Capablanca, that kind of question gets asked a lot, not just in strategy games but in sports too. It’s extremely hard to say. Capablanca was a genius and no doubt if he grew up in our era, with our more extensive knowledge of the game, he would have reached even greater heights of chess brilliance.
Could he be world champion though? I think it’s impossible to say – there are too many confounding factors.
Ratings like Elo can in fact be used to compare between times, as long as people of different ages play against each other (which they do). And Elo ratings have, in fact, increased with time. I once saw somewhere an animated graph of the top 10 chess players for the past 200 years or some such: Colored threads winding up and down, with every so often a thread dropping off the bottom or ending, and new threads coming in from the bottom or starting anew, and there was a definite and significant upwards trend.
For the past 50 or so years, comparison across time has gotten even better, because we now have players who have been on the scene for that long and whose skill level can be absolutely known to be constant with time, and who will continue to be on the scene at the same skill level for centuries to come.
Disagree with this. As Schnitte correctly points out, ELO is a relative measure, but beyond that, there are a number of factors that cause ELO inflation or deflation. The extent to which we can correct for this is debateable. Some measures like “chessmetrics” claim that the highest inflation-adjusted one-year ELO was achieved by Bobby Fischer.
Just to make sure I follow: you mean we can compare human players to chess computers using a standardized configuration?
Ratings like Elo can in fact be used to compare between times, as long as people of different ages play against each other
Is that true? Sure, you can have old A play young B, and then later have B, when he’s older, play young C. B would then be a link between A’s and C’s generation. But in the meantime, B’s strength need not have stayed constant, nor developed in any way that could be captured in absolute terms numerically (you can measure it as an Elo number, sure, but again I’m not sure that would capture a gradual improvement across the board).
In any individual case, no, you couldn’t be sure that B’s true skill level remained constant, but with many people spanning the generations, that should average out.
But has anyone ever actually done the experiment, of taking an early rated chess program, dusting off an old PDP-11 or whatever, and playing it against enough modern opponents to get a good Elo rating for it? That way, we could say definitively whether ratings were inflating.
It’s only really possible for games that have programmed AI that can still be run on modern systems to the same specifications that they originally were, as @Chronos has pointed out. I don’t know if there’s ever been any attempt to dust off old programs and see how they fare against newer programs or meat-based players whose skill in the current system is well understood.
I feel it’s very likely such an effect exists. The analogy I would use would be standing on the shoulders of giants. Morphy never had a chance to learn from Capablanca or Fisher or Carlsen. But Capablanca could learn from Morphy, Fisher could learn from Morphy and Capablanca, and Carlson can learn from Morphy and Capablanca and Fisher. And as the OP pointed out, modern players have the opportunity to learn from computer play.
One thing you have with video games that you don’t have with Chess or Go is that people tend to ditch older games for newer games. People constantly come in and out of the Chess world in a flow that basically balances out over time, but video games tend to have a lot of players when they’re new, then a big decrease over time with few new players. This means that, in addition to players getting better, a lot of the players are just going to play for a bit, then move on, and very few people who just want to mess around with a game are going to keep playing a video game ten years later.
Yeah. It seems almost certain that the chess community as a whole has learned strategies and techniques over the centuries, leading to an increase in skill level across the board (so that novices now know more than novices “then” and grandmasters now know more than grandmasters then).
Aren’t novices by definition people who basically only know the rules of the game? Like, I’m a novice at Go in that I can follow the rules, but whatever techniques have developed in Go over the centuries I’m totally lost to despite having tried to read about them, simply because I wasn’t serious about learning. With Chess, there was a time I was serious about learning, and learned all sorts of things that would give me a massive advantage over someone who just learned the rules now. If you haven’t sat down and seriously studied a book on Chess, you’re not getting any advantage from the past few hundred years of study.
I’m thinking of a novice as a little higher than an absolute beginner - someone who has read the rulebook that has a few bits of advice about the relative values of pieces and the importance of control of the center - stuff that’s basic-basic today - but wasn’t always.