Home Videos Games Podcastle

International game theory clashes


#1

Two of my favourite games - Twilight Struggle and Through the Ages - have relatively recently had widely-accepted Western-origin game theories overturned soon after a good software implementation came out, due to the impact of high-level play by Chinese players.

Looking at Twilight Struggle, many years of international English-language competition had resulted in a meta that, broadly speaking, valued working towards a favourable endgame board state while thinning the deck to neutralise your opponent’s events.

Then the Chinese players came along, fighting tooth and nail for every victory point, always with an eye to overturning dominance of a region exactly when necessary, aiming for early wins, and actively pursuing the Space Race… and they won, and kept winning. 10+ years of theory overturned.

More recently, the new edition of Through the Ages came out, and despite wide-ranging sweeping changes to the game, most English-language players still concluded that, as in the previous edition, science and military were more reliable ways to win 2-player games, and this largely influenced 3-4 players games too. Indeed, falling behind in military was widely regarded as a mistake only made by very green players and many new players complained that not being able pursue a peaceful route to victory was a flaw in the game(!)

Then came the Chinese players, and bam, suddenly early strong culture pushes were revealed as not only viable, but more reliable and effective under many circumstances. Many assumptions were overturned, and theories shook up.

I’m sure this has happened over the years for games like Go and Chess too, but I don’t really know much about the history of these games. Anyone have any insights into these games, or others?

I just find it fascinating that the language barrier can result in these wildly different and contradicting game theories being developed over years and years, and only really put to the test when different groups of players clash on the international stage.


#2

I tend to think this is less (explicitly) a language thing and more to do with insular thinking. Competitive meta is only as good as the consensus decides it is. When a tiny minority dictates the rote actions of the herd, it’s easy to feel confident in your superiority. This is probably just a case of new eyes looking to poke holes (and succeeding).

[EDIT] Wanted to clarify that I meant it’s a language thing only insofar as it was the barrier, and to add that it’s a Chinese thing only insofar as scale REALLY counts here.


#4

Yeah everyone playing each other would have got themselves in a loop and reinforced strategies over and over and shown as dominant.

I wonder if there are “crap” strategies the lose against the western pattern - and lost favour due to their specific weakness to that pattern - would beat the Chinese one.

Edit; I rememberi a long time ago hearing about a smash brothers documentary (called the smash brothers on youtube) were the fixed patterns were set in east and west coast america.


#5

This is fascinating stuff. I’ve played loads of Through the Ages but mostly against a small group of people plus the inbuilt app AI. I would have thought we’d developed our own meta, but it seems to conform to what you describe as Western. I’d love to know where your data comes from. I generally don’t read online strategy stuff so I’m out of touch I guess.

With respect to Twilight Struggle, I’m a long way off having the experience to appreciate the subtleties of different approaches, though I have read quite a bit of the Twilight Strategy web site, which is probably responsible for a lot of Western meta.


#6

Thanks. Through the Ages seems to be having a mini-revolution over on the BGG forums, catalysed by some of the most prolific Chinese players, and specifically Gary Pan, who seems quite influential among Chinese players. He basically took on all challengers in 1v1 matches, and I haven’t yet seen anyone seriously contest his claim to be “the best player of 2-player TtA of all time”. The key point is that numerous well-known and well-respected players are getting beaten by what they describe as “unorthodox” strategies, which begin to get outlined on https://boardgamegeek.com/thread/2132925/who-best-tta-player-world-me-or-petri/page/3

Regarding Twilight Struggle, I think the “Theory” approach, based on the website written by Theory, contains a lot of discredited ideas even prior to the Chinese impact (to be fair, it was written a long time ago), but even so, the Western world champion, Ziemowit, still championed the long-game approach, and to some extent still does. It took the evidence of not just one, but many Chinese players, dominating the GLICKO rankings on the Playdek app, and the explanations of their strategies, largely provided by a Chinese player named Sankt, before the majority of strong players really starting reassessing their priorities. I don’t think this is necessarily the best summary, because it doesn’t focus on which strategies are “new”, just on what works, but if you click through the links it’s the first one I found just now: https://boardgamegeek.com/thread/1990968/rethinking-twilight-strategy-summary

So, in each case, there has a been a dominant “western” meta developed over years, and proved and trialled in “world” championships around the western world, that has been shaken up by an influx of strong Chinese players, and analysis and explanation of their play by a select few Chinese players with the English communication skills and desire to connect (or gain recognition, or whatever).

I spend too much time on BGG, I’d prefer to be playing =)


#7

Thanks for sending me to that thread. It’s one of the most entertaining things I’ve read on BGG for ages. I’m not sure it totally supports the claim that China produces a completely new meta, but I can well believe that China produces the best players. I’m resisting the temptation to get relentlessly squished by GaryPan myself.


#8

Yeah, there isn’t the body of evidence as there is for Twilight Struggle, and arguably there isn’t the same scope for variation in the first place. The key things are that it only applies to 2-player games, and lacks a breakdown of actual teachable points, unlike the TS revolution. I do have to say that I have almost never seen a genuinely good player put out multiple (or any) Age I drama and libraries except when playing against really forgiving opposition. Even 3 early Iron is a little controversial, although there were good non-Chinese players advocating that before Gary Pan came along.

I’m looking forward to the promised implementation of actual ranked TtA games/tournaments instead of the current “play more to gain ranks” system.


#9

Oh, and I should add that Gary Pan has been at this for quite some time if you track his “challenge me” post back 2 years(!)


#10

I’m currently enjoying the fact that I can farm achievements without my actual skill level being made clear :slight_smile:. I’m planning to retire as soon as I hit 86 achievements, which will probably be soon after my 50th online game.


#11

I don’t know how many achievements I have, but I enjoyed completing all the challenges, and getting 2222 Domination, and a double hat-trick of online wins. I think the only difficult one left is 3333 Domination, but I’m not sure I can be bothered with that one!

I’m benkyo on the app, if you’d like a game! (2-4 players, I’m not bothered, but it has to be three turtle speed max, if I’m to keep up…)


#12

Done. 4-player ‘SUSD challenge’ awaits you, at 3-turtle speed


#13

I should probably add that I also had the Japanese Go scene in mind before I started writing, but forgot to mention it in the thread starter. I’m far from knowledgeable on the topic, but I had the impression that Japanese professional Go players had stagnated in their own meta for quite some time before Korean and Chinese professionals showed them up. Sorry, no sources or dates for this, it’s just something I vaguely remember from somewhere, somewhen.


#14

I want to believe it’s somehow cultural, but in both those cases doesn’t “Chinese players arrive” coincide with “game is available online”. Suddenly many, many more games are being played, and it’s far easier to try out new and weird strategies, because you can get a game any time.

Certainly in the case of TS too, that initial ‘Western’ meta seems somewhat influenced by the design objective of the game: it was designed to run to time with early victories being a rarity. That the strongest strategy is to not actually do that might be considered a design flaw.


#15

Oh hey, I was never intending to imply a cultural link. Just different large communities in relative isolation developing ideas in parallel. The game being available online (well, technically, both were widely available online for years prior to the events I’m describing) on a more popular platform is what brought the communities “together”.

Well, the Japanese Go thing is a different case. That was to do with pro communities finally having real international competitions, nothing to do with online gaming AFAIK.

I don’t know about TS being “designed” to go to final scoring. I’ve not seen any indication that that is the case.


#16

Indeed, there’s plenty of indication of the opposite. The 20-point margin rule, the DEFCON rule, and War Games to name three.


#17

Didn’t mean to imply you had either! Just wanted to ensure it was noted in case anyone inferred otherwise.

I don’t have anything to add specific to these titles being discussed, but I’m loving the conversation. I guess I will say that I find it interesting that this is a phenomenon common to just about any competition (friendly or otherwise) at a high level of play. Games (board/card/vid), sports, politics, science… sharing (or maybe enforcing) ideas basically always improves the whole. It’s cool.


#18

I’ve been following this, and it’s prevalent everywhere, I think.

There’s a case for this in sport. In rugby union, and its evolution from an amateur sport to a professional sport. The arrival of Jonah Lomu (16st, 6’4" winger) in the early 90s, at the cusp of professionalism, has a profound impact on the sport, leading one traumatised England captain to call him a “freak”. There are numerous books about that subject which are far more eloquent than I.

It’s in cricket too. A Test match is 5 days long, and some teams wouldn’t score more than 200 in a day. Then the Australians of the late 1990s showed up and said, let’s get to 350 in a day. Their reasoning was simple - it’s a tactic that hasn’t been seen in the game, and other teams will remain in the old mindset and be unprepared to make that kind of total. Australia would win by batting them out of the game. One player in particular, Adam Gilchrist, was a new type of wicketkeeper who batted destructively, making so many more runs per ball than anyone batting in his position, that other countries wouldn’t select wicketpkeepers if they couldn’t bat like a batsman. In T20 cricket, it’s all about the batting. In the IPL Sunrisers Hyderabad recruited (similar to NFL draft) really canny bowlers, and almost won (they lost to Chennai in the final) by restricting teams to 100/120 runs, down from a par of 180/200.

And so football (soccer) doesn’t feel left out, England haven’t really changed from their physical fast tempo game that won them the World Cup in 1966. All other footballing nations have moved on. With the deep lying playmaker (Italy 2006), tiki-taka (Spain 2010), gegenpress (Germany 2014) and counterattacking (France 2018) all proving effective and changing tactics in past World Cups.

Sports (and games) evolve, and that’s the beauty of it all.

I feel like I hijacked the thread, but different approaches - and finding them - are key to understanding game theory.


#19

Even in chess, Alphazero has sort of done something similar (assuming something similar is ‘change the expectations of experts’). Nobody expects old-school chess engines with understandable algorithms to outperform self-learning (and consequently highly mysterious) machines any more, like nobody has expected humans to outperform computers since Deep Blue.