Friday, May 24, 2013

Ranks and Tiers for Fun

I don't play Pokemon, so I never really learned much about the complex metagame that evolved around competitive Pokemon. In a normal fighting game, the characters are often ranked into tiers by the players - this player is top-tier, that player is bottom-tier, etc. This is normally based around how well they perform in the hands of the top-level players, but in general most of the top-tier characters will be better than most of the bottom-tier characters even in the hands of a relative newbie.

So you play top-tier characters more or less exclusively.

But Pokemon has a very different ranking system. I'm not 100% sure on all the details, but essentially the tiers are fundamentally similar to fighting game tiers... except that they are divided into leagues.

IE, both sides will play top-tier Pokemon (or lower). Or, alternately, perhaps it's a mid-tier and lower... or even a low-tier and lower. If you really love your shitty Pokemon, or have come up with an interesting combo using low-grade mons, you can dominate the lower-tier competition and not worry about the battles between the overpowered heavy hitters. It's sort of like boxing: you don't generally mix heavyweights and flyweights in one battle, but the flyweight battles are still very intense and interesting.

Unlike boxing, we divide up by global performance rather than by something like weight. This means that as the games iterate and evolve, mons can climb or fall. This keeps the whole arrangement flexible enough to support tiered play regardless of whether the balance shifts or not.

I really like this method. It gives us a way to have hundreds of options without making 85 of them useless to a top-flight player. More like the opposite: perhaps 85 of them will be useful to a top-flight player depending on which category he participates in. The others will be in the awkward state of being too strong for one tier, but too weak to be useful on the next tier up.

Now, the Pokemon ranking system focuses on rather simple conflicts rather than the more complex conflicts that are becoming the norm. It becomes very hard to come up with a ratings system when, for example, one character is point and two characters are support, and they can swap out.

Then you get something like the new "vs Capcom" tiers, where the characters have a front-line tier and a support tier. A character that is strong at one and weak at the other isn't exactly mid-tier... they are top-tier when used correctly, or bottom-tier when used badly.

This complexity adds a lot of metagame (can you force them to switch out their front line combatant for their support?) but it also adds a lot of complexity to the idea of weight classes. If you are playing a mid-tier conflict and you use a mid-tier support character, fine... but that character is god-tier front-line, so what happens if you switch to him mid fight? Similarly, if you simply ban all characters with any tier higher than the weight class, you'll end up with the mid-tier class dominated by people who are blandly balanced - mid-tier at everything. But in actuality, that means they are better than mid-tier, because they can serve reliably in all roles regardless of the situation. So they aren't mid-tier characters... but in a top-tier battle, they can't reliably hold up in any role...

There are a lot of really fun options you can play with how you implement this kind of thing. Rather than win/loss ratios, you could simply go by how often they are used. If a particular mid-ranked character is used in 30% of all mid-ranked matches, you know they are dominating the mid-rank and should be promoted even if their "stat" ranks are all mid-tier.

Another option is a point-buy system where each tier allows you to spend a certain number of points on characters, and the ranks are not about banning, but about deciding price.

Another option is an auto-nerf system. At the end of each month, each character is gently nerfed or buffed depending on whether it was over- or under-utilized. With a significant player base (> 1000, each using 3+ characters), this should actually settle pretty quickly.


MOREOVER, this can all be applied to user-generated content.

The big problem with user-generated content is that it is generally not seen by very many players. Most players are going to try out most Pokemon. Most players are going to try out most fighting game characters. But most players are not going to try out most quests created by random internet people. This leaves a very fractured ranking that tends to result in a few excessively popular pieces of content and millions that never get seen by anybody.

Rather than say that the good content is simply better, you could divvy up the content into weight classes. This super-popular content is top tier, the most polished and highly regarded. But you earn points by using content from the lower tiers - the amateur tier and the beginner tier. Of course, just by using them you push them towards the next tier up...

Hm. Maybe rather than giving "thumbs up" or "thumbs down", you rank it before you start on #/5 and then, afterwards, as "better/worse than expected"...


Sean N. said...

I think the idea of "weight classes" for something like a MOBA would be good, but let's say that instead of having predefined "champions" and "roles" in the game, you could gain points by doing well and winning matches, and as you went up (let's say there was a cap at 1000, 2500, 5000, and 10000 points) you would gain points and be able to buy more expensive combat skills and more expensive passive bonuses. Think of a MOBA where you create your own champion and as he or she gets more wins, he or she gets stronger, and plays harder opponents who are also ranked higher.

Craig Perko said...

The difficulty with that is the complexity of combined skills. Two 1000-point skills could synergise well and be equivalent to a 5000-point skill... or be pretty crummy, and only be as good as a 1500-point skill.

That's why predefined packs are easy to balance - the package never changes, and you just rate the final unit. Customizable units are harder to rate because they change.

Could still probably be done, though.