I think every single person who’s played Overwatch has at least experienced one moment where they thought they got play of the game, only to watch in horror as something else came up after the round – something potentially less significant, or far less impressive. Why is that? How exactly does that mechanic work? What determines if a specific play is really good?
At the moment, raw numbers it seems, though there are other variables considered too. Overwatch’s lead software engineer, Rowan Hamilton, shared some insight into the system over in an interview with Gamespot.
From a technical standpoint, it’s a really hard problem to have a computer figure out what is cool. They’re not very smart. They take some numbers in, they put some numbers out. It’s hard to figure out what is cool there.
We’ve added some stuff recently, such as determining how hard a shot was to hit based on how fast the target was moving, how far away the target was moving. So a snipe of someone half a screen away who was just chilling out and waiting to be headshot won’t be weighted as heavily as a Tracer zipping across, barely in sight that you manage to pick off. We’re constantly looking at different things we can add to that.
That sounds good, but there’s still a ton of room for improvement. In my own experience, play of the game has almost always boiled down to who dealt the most damage, and if they did it in the vicinity of an important objective.
Supports aren’t forgotten, but I seldom see top plays from the likes of Symmetra for example. Mercy pops up now and then, but only if she manages to resurrect lots of her allies. As for Lúcio and Zenyatta, I’ve only seen them feature if they happen to land a particularly good killstreak, which for them, is a lot harder because their primary roles are healing, not dealing damage.
Despite those hiccups, play of the game is definitely one of my favourite features in Overwatch. I’m sure with enough time, Blizzard will manage to tweak the system to choose better, more relevant plays.
Last Updated: May 30, 2016