The Meta Game

I want to talk about metagames. But before I can tell you that story, I want to tell you a completely different story.

Pokemon Showdown is a competitive Pokemon simulator. For any audience members who are unfamiliar, Pokemon is a franchise with a long history and many genres (video games of varying genres, card games, anime, movies, books, merch, etc). But at the heart of this massive media franchise are "core Pokemon games" which are video games with a fairly specific format. Each of them places you as a Pokemon trainer in a world of Pokemon. While there have recently been changes in how you navigate the world or what your goal is, all of the core games are usually unified by the battle system.

The battle system, in fine JRPG tradition, is played in a separate screen from the overworld and is a form of turn based combat. Up to 6 Pokemon or "mons" for each player battle it out until only one side is left standing. In the games, this is quite simple. These games are meant for children. All the standard JRPG techniques work (grinding, hitting weaknesses, item spam, etc). But there's one small little complicating factor: the battles can be multiplayer. And where there's the possibility of challenging others, there are people who will take things very seriously. In other words, they'll desire to be the very best, like no one ever was.

A full analysis of the effect of tryharding on video games and the ethics of such is beyond this piece but maybe something I will write about eventually. If you're curious, go watch this for more on that.

Pokemon tournaments go way back in e-sports terms, right to the birth of the concept, although in my expirience it isn't what a lot of people think of when they say "e-sports". Tournaments are where the breakdown between designing a game for the average player and designing a game for the top ranked competitive player becomes apparent.

See, when you catch a Pokemon, it has 6 stats that define it. These stats are calculated from 3 values:

At least those 3 are used in modern versions of Pokemon. Most things in Pokemon work differently if you go back far enough.

So naturally, if you're preparing a team for a tournament, you want a mon with stats as high as possible. BST is unchangeable and only factors into selecting which mons to use. EVs can be set manually with a little work. But how do you max IVs? IVs are, at least in older gens, completely random. The only way to max them is to spend hours catching or breeding as many of that mon as possible until you get one that is perfect. Then you have to grind it to level up, maybe evolve through weird means, set the EVs, and get the moves you want. And then they went and complicated things by adding more mechanics like natures and egg moves and TMs and mons that are just really annoying to catch (cough cough gen 4 feebas).

If you're a casual player, you won't notice any of this. The only practical effect is that different Pokemon seem to have different stats, in a seemingly random but well distributed fashion. You can play through the entire game while only understanding this at the level of "higher levels means bigger stats" and "different species are good at different things".

but training up a full "competitively viable" team took hours. Many hours. And that meant that if you played 2 matches and immediately realized you need to replace a mon, you'd be at it for hours more. It's a bad system for casual players, weeding out anyone who can't put the time in, regardless of their skill. It also mandates understanding these mechanics on a deeper level, necessitating reading external guides and creating yet more divides between casual and competitive players.

I should add that in recent games, steps have been taken to fix these problems. Tools for checking and improving IVs have slowly been added and levelling up and grinding EVs got faster. But go back 10 years, no such things existed.

Really, there were two connected problems. One was that training teams took too long, meaning that enjoying the full range of competitive battle was out of the range of players without the time to sink into it, denying them an intended mode of play. The other was that top rank players had to either hack or put in way too much work, just to test a team out. So the players did the obvious thing and improvised.

Pokemon simulators copy the battle system of Pokemon exactly, but without the rest of the game attached. Instead of training a team, you just tell the simulator what you want. iT checks if it would be a legal Pokemon team and if so, gives it to you instantly. Add some ladder functionality and hey! You get Pokemon Showdown.

Obviously the history is a little more spotted than that, with a lot of fun twists and turns and maybe the odd bit of potential legal action. Showdown wasn't the first simulator by far, but it did inherit a lot from those that came before it. These days, it's comprehensive, covers all gens, has many battle formats including modded ones, is open source, and anyone can run their own server (although most people use the main one for the benefit of larger ladders). As well, it is strongly connected to Smogon, a group of competitive Pokemon enthusiasts who study all this.

Their mechanics research team are dedicated to discovering and then replicating weird quirks of the mechanics into the simulator. My personal favourite is the discovery of the overflow bug. If you can do enough damage in a single hit, the target will take no damage.

The interesting result of all this is that there's a thriving community of Pokemon players, all doing competitive Pokemon online, in a way that Nintendo never intended and perhaps only begrudgingly puts up with.

So with that out of the way, lets talk about metagames. A "metagame" is a commonly used word in competitive games. It's thrown around just as often by League of Legends players as it is by Magic: the Gathering players. But what does it actually mean? Well, very literally, "meta" means referring to itself or the genre. Therefore, a "metagame" is a game about games or a game about the concept of games.

A metagame is the game you play outside of and between the normal game. It covers your evolving strategies and decisions. To put this another way, the actual 'game' of Pokemon is the battle system. The 'metagame' is the team building. It is not the actual game, but it is deeply connected to the game. It is played around and through the actual game. The key element of the metagame is that it doesn't just rely on a good understanding of the individual battle mechanics, but also a comprehensive knowledge of other people. And I'd like to illustrate that with an example.

Say I've made a fighting game with 100 characters. Alice is a fast character who doesn't do much damage. Bob is a slow character who hits really hard. Eve is every other character, who are all kinda average and the same (I'm not a very creative designer).

I've completely failed to balance this game. Alice doesn't hit hard enough. She loses against Eve (98 of the characters) very reliably. To make this simple, lets say that she loses 100% of the time against Eve, ignoring player skill. Bob, on the other hand, hits way too hard. He wins against Eve 100% of the time. But, Alice is a little too fast. She is perfectly calibrated to go against Bob and can dodge his best attacks. Alice beats Bob 100% of the time. Any mirror match has a 50% win rate.

We can calculate expected win rates! Alice would expect to win against herself 50% of the time and against Bob 100% of the time. This gives her a net win rate of 1.5% out of the 100 possible opponents. Bob, by the same logic, has a 98.5% win rate. Eve has a neat 50% win rate. Clearly, the obvious strategy is just to play Bob.

But wait a minute. These win rates assume each character is being used with the same frequency. But if the dominant strategy is to play Bob, then won't more people start doing that? To calculate the actual win rates, you have to do a weighted average of win rate versus that character times the frequency you would expect to see that character.

And that means that as Bob's use rate climbs, some genius will realize that Alice, a character only capable of beating Bob is suddenly useful. And then Alice's use rate will climb in turn, until the Bobs realize that Eve is viable, and so on. Things will shift continuously.

This gets more complicated the more characters there are and the more player skill factors into it. 100% win rates for a character versus another character are usually derided as bad game design, for justifiable reasons. There are exceptions, of course. It's okay for one Pokemon to always beat another one because you get six of them. You can simply switch to one better suited for the match up. Switching being a key part of the game is one thing that catches a lot of casual players getting into competitive by surprise. It isn't uncommon to see very skilled players both switch for 2-3 turns in a row at a time, which is baffling to anyone who isn't familiar with the metagame what's going on.

Commonly, "the meta" is used a shorthand for whatever strategies and builds are common on the current competitive scene. I'm pretty sure this is a bad use of etymology, but it's too late to change it. Metas are really interesting to look at because they can completely change or break a lot of the assumptions that went into the game design. And that's why I brought up Pokemon earlier.

Most of these games, the balancing decisions are conducted behind the scenes. League developers may listen to the community, but ultimately the nerf or buff decisions are handed out from a sort of divinity. Pokemon Showdown, on the other hand, is in the unique position of being run by volunteers through a public forum. And that's means it's a massive repository of thinkpieces, essays, discussions, and more about the philosophy of metagames which anyone can just read.

There are currently 1000 Pokemon. Barring non-fully-evolved Pokemon which, with a couple key exceptions, are just weaker that fully-evolved Pokemon, there's probably something like 400-500 potentially competitively viable Pokemon. There are thousands (I think) of held items, thousands of potential moves, and hundreds of possible stat combinations. And most of them are bad.

Pokemon is designed firstly as a playable RPG. And that means that things are balanced by how hard they are to get in casual terms. Legendary Pokemon that are the crux of the storyline have better stats and stronger moves, so they feel more rewarding and climactic. Rarer Pokemon are stronger in general. And this means that the majority of Pokemon will do really badly in the meta.

They fixed this with tiers. Each tier bans Pokemon of over a certain strength. This means that each tier has a very different feel and most players will likely have a tier where their favourite Pokemon is competitive. This is also the heart of how Showdown shifts metagames. If a metagame is unbalanced, the easy solution is to ban a mon to only be used in the next highest tier.

The tiers are, in order from strongest to weakest, Ubers, Over Used (OU), Under Used (UU), Rarely Used (RU), Never Used (NU), ?? (PU), and Little Cup (LC). And, fascinatingly, each tier has a very different style. Little Cup skews very offensive due to the way it is structured. Defensive tactics are almost impossible and very rare. On the other hand, stall tactics can thrive in many other tiers.

You may note that each tier is named after a level of usage. That is actually how the tiers are determined! If a Pokemon is used enough, it gets promoted to the next tier. This is a core part of the healthy metagame philosophy they espouse.

Essentially, the idea is that a healthy metagame has a lot of viable options. There should be a large choice of possible builds instead of just one dominant strategy. A meta is considered "over-centralized" when there is a single dominant strategy. An over-centralized meta ends up requiring that every player either runs the best strategy or runs a specific counter to that strategy. This is kind of boring and unfun, so the goal is to eliminate it.

A strategy that counters a dominant strategy is called "anti-meta". Anti-meta strategies differ from just normal strategies in that they exploit usage rates to make tricks work that wouldn't normally work. The perfect example of this is Alice from our previous example. If the usage pool was diverse, she would suck. But because of Bob's dominance, she can be very effective even though she's only good at countering one specific build. This has been used to great effect in Magic tournaments, where decks that only focus on shutting down meta strategies occasionally do really well at tournaments.

Anti-meta strats are often hated by people running meta strats because it's always frustrating to run into someone specifically countering you. But of course, if an anti-meta strategy is good, it isn't long before it gets enrolled into the meta and just becomes another meta strategy, with its own anti-meta counters. Interesting, I've noticed that a lot of people consider any strategy they find frustrating "anti-meta", instead of just considering it a part of the metagame (the memoir of a stall player). I think this is part of the big drawback of metagames. Because strategies do differently against each other, once a metagame is well developed, a match may be determined before it even starts by team composition. This is incredibly frustrating for players, invalidating the actual playing of the game and reducing play to rock-paper-scissors.

This also leads into the fundamental defining idea behind a metagame: you are trying to play the ladder. When team building, your goal isn't to build a good team. It is to build a team that counters as much of the ladder as possible, weighted by liklihood of encountering. To give an example, my favourite Pokemon team I've made was a stall team for gen 7 ubers. I won't bore you the details, but stall struggles a lot with a strategy called hyper-offence but does well against balanced teams. There were things I could do to make my team work better against hyper-offence and I did almost none of them. Hyper-offence teams were rare, maybe 1 in 50 battles, and so weren't worth the effort compared to being more consistent against balanced teams.

A really interesting and funny consequence is that metagames are inherently decided by public opinion. Note that they're defined by use instead of what is actually good. If everyone decides one build is really cool, that build climbs into the metagame even if its effectiveness is bad. This means that key figures in the community actually have a lot of control over the shape of the metagame. For example, to demonstrate the idiocy of Showdown's pure usage based tiers, at several points people have led campaigns to use bad Pokemon a lot, often causing noticeable twists in the meta.

There are other problems with a purely usage based approach for balance. Some Pokemon are over-centralizing in one tier without ever seeing use in the next one up. Showdown has Borderline (BL) categories, indicating these mons. Tragically, this caused a bunch of mons that were banned from OU to not really have a home anywhere. Ubers is a hard hitting genre and mons that are too good for OU just don't cut it there.

And sometimes things are just too strong. Famously, Mega Rayquaza was so strong, they invented an entire tier just to contain it: Anything Goes (AG). AG did later acquire its own identity as the place for banned gimmick strategies and is a truly wild meta to play on.

Another interesting thing to note is that metas change with your place on the ladder. As different levels of player skill interact with different strategies differently, more skilled players will tend to use different strategies. What works on the low ladder may not work on the high ladder, adding some dimension to the game. This also further complicates balancing. Remember when I said balancing a game both for casuals and competitive players is hard? It's just as hard to balance a game for both the low end and the high end of the spectrum. You see this in, for example, Smash Bros, a game with a thriving competitive scene (who have strong opinions about balance changes) and is also mostly played by casuals against their friends on a couch (who have very different opinions about balance changes). Keeping characters approachable for casual players and balanced for those who are eking out every possible speck of performance is a hard task and it's open to interpretation how well Nintendo are managing.

This is also how I once destroyed a meta. While I was running bots on Showdown, I created a gimmick bot for gen 7 ubers and left it running for a few hours. It failed to climb because it lost against more skilled players and stayed in the low ladder area. Because it was a smallish ladder and the bot was playing 5 matches at once, every player in the low ladder was seeing it fairly often (1/3 to 1/2 matches). When I came back, the low ladder players were divided into the camps of "ragequitting on sight" and "specifically modified their teams to counter me", often in ways that made their teams worse in general.

Left to their own devices, metagames are in theory self stabilizing. This is considered a goal of Smogon's metagame guidelines. The benefits of a stable meta is that it can be played very casually. A metagame that shifts a lot is alienating to anyone who doesn't want to put in the work to closely follow the game by reading analysis or following very skilled players. If the metagame is stable, you can learn it slowly over the course of weeks or months and what you learn one week will remain useful the next.

In theory, the numbers should slowly stabilize as a good sense of what is available becomes known to everyone. This is harder in some games than others. Frequent balance changes and patches or bans destabilize metas. Influential players suggesting specific strategies shape things strongly. Trends sometimes come in waves: a given strategy becomes dominant, its counter becomes dominant, the counter counter becomes dominant, the original strategy comes back to fill the gap. In newer games, metagames frequently destabilize as new builds, tricks, and tactics are invented. As well, they may stabilize to a bad place, being over-centralized or unfun.

A common thing that drives me off competitive games is when the key meta strategies are not fun. I'm a stall player by nature. I love long drawn out very defensive battles. Many other players (understandably) hate stall for taking longer, making it harder to ladder. Accordingly, metagames often end up very punishing towards stall, which makes me less inclined to play them. There are other ways this shapes out too. The key strategies for impossible mode in XCOM (my favourite game) involve sneaking around the map and taking forever to do anything. I understand all these strategies in detail. I also despise doing them as they are boring and dumb. This is why I've never beaten it on impossible mode and tend to play on the lower difficulties: I find the less effective strategies more fun.

Metagames are inevitable for any kind of competitive multiplayer game. Some strategies will always be better than others, and that will always encourage counters to these strategies, and so on. Because of that, trying to prevent a metagame from forming is like fighting the ocean. Far better to master metagames than to drown in tem. Managing metagames is an art and it is best performed by those deeply involved in the game. Tragically, the metagame can only ever be balanced for some of the people some of the time. I do wonder what it would be like to have a game with actual different rules depending on your place on the ladder. Perhaps it would be very frustrating to play?

In conclusion, I should play games more casually. Anyone up for a gen 7 match?

Today's link of the day is pluralistic, the blog of science fiction author and activist Cory Doctorow. As well as writing incredible cyberpunk and sci fi novels and short stories, he regularly blogs about issues with tech, capitalism, and surveillance. Heartily recommend adding it to your RSS reader if you have one (and you should).