It's nearing the end of the semester, and I am desperately trying to avoid real work. Seems like the perfect opportunity to get my shit together and focus on my education write an essay about a video game and post it to Reddit. Today's topic: steamrolls.

I've repeatedly rewritten this to make it more concise and meaningful, so some of the older comments might not be as relevant. Sorry about that! I've bolded key points and put a TL;DR at the bottom.

Why Steamrolls Happen

First, we must understand how much impact an individual can have. An average solo winrate typically varies between 44% and 56%, but on the extremes can be anywhere from 38% (AFK every game) to over 65% (carrying almost every winnable game). That's insane if you think about it – we have 15 man teams, but a single player can change the win chance by over 25 percentage points.

How is this possible? To put it simply, this is a hard game to get good at. It's a very complex game with many moving parts. Mechanical skill is capped by RNG, so it comes down to vehicle knowledge, map knowledge, and strategy. Ten different tiers means ten different metas, the 30+ maps have dozens of important positions, and the roster has nearly 400 tanks. Inconsistent feedback means you often have to try a position a dozen times to find whether it works or not, and hundreds to understand why.

Maps are fought from a handful of powerful map positions, death is permanent, and momentum matters – once a flank is won, it gives your team a major advantage in winning the other flanks. Add in that the game has large casual audience and you have a situation where there can be a massive skill disparity between players in the same match.

All this adds up to our somewhat paradoxical conclusion: individual players are too impactful. Phrased differently, the average enemy team leaves open many, many opportunities to defeat them. The problem isn't that you aren't able to carry enough, it's that so many players on the other team are. Whether by skill or by luck, you only need a handful of players to make the right plays at the right times to win the game, because that's more coordination than the enemy team is likely to muster.

Fixing Steamrolls

We can't expect to "fix" the problem, but we can try to make it better. There are a few different strategies, which boil down to either reducing individual influence or making the teams more balanced. In other words, relative to the current state of the game, do you make it harder to carry or make it so players have to carry harder?

Making it harder to carry is a terrible idea. We could dumb down the maps – remove strong positions, so that a player can't as easily create advantageous engagements. This doesn't help: it makes it harder to initiate a steamroll, but also harder to stop one. Increasing RNG wouldn't help much, because the better positioned player is just as likely to get lucky as they are to get unlucky. This would reduce the viability of some risky strategies, but the most noticeable impact would be to make the game more frustrating. Human psychology is such that notice bad luck more than good luck, so if you want players to enjoy your game, this should be minimized.

Perhaps we directly target the good players, to limit their effectiveness? Funny enough, I think that's partly what arty accomplishes: calculated aggression is often the key to winning, and arty punishes aggression. This is of course a godawful idea; you should not punish your players for getting good at your game. Having dedicated players that get good at your game is something to revel, not revile. At best they'll game the system, and at worst they'll stop playing (and stop spending money).

Our solution therefore is to make carrying harder, by making the teams more balanced. Though maybe ideal, there isn't a good way to directly improve teamwork. On top of the large casual playerbase, many players can't make good plans, many players completely ignore the chat, and on servers like EU there may not even be a common language. Teamwork has to be an emergent property, not a forced one. Players will be compelled to play as a team if that's the only way to beat the enemy, but much of the playerbase isn't here for that type of experience.

Instead of making the best players worse, we could try making the worst players better: by offering resources for players to get better, we can reduce the performance gap between teams. The problem is, this only works when the players are trying to get better. There are plenty of decent resources already available that players don't use. Still, this is worth doing for other reasons.

Improving game balance is a no-brainer. If vehicles are roughly equally powerful, then one team getting more of a certain tank would have less of an effect.

A common suggestion is to make a skill-based matchmaker. This means ranking players by skill and having them only face players of similar skill. There are logistical concerns preventing this, like the large number of tiers, number of players per game, and lack of players in the queue. The bigger issue is that many of us don't want this. I want my reward for getting better at the game to be winning more, not to get matched against harder and harder opponents. Lots of games are designed this way for a reason. Some day, when bots are a high enough quality, you'll be able to separate the human players without changing this experience. For now, we have to look elsewhere.

What if, instead of changing the player compositions in a match, we changed their distribution? In other words, what if we tried to balance the teams so they are equally skilled? This is how we all picked teams for dodgeball in PE class – unless one of the team captains was a jackass, the teams would be pretty fair. Individuals can carry and the teams are balanced. You'd have to figure out a lot of details; for instance, maybe a unicum bottom-tier is balanced by two tomato top-tiers. This is doable, and I think it's the most viable solution we have available.

TL;DR: steamrolls are the result of players not fully understanding a very complex game; because the average team makes many mistakes, it takes a small number of smart decisions to win. We can't fix this, but we can reduce this by leveling the playing field: balance the tanks and balance the teams. Currently, trying to match equally skilled teams directly instead of tiering players into a skill-based matchmaker is the best bet.