Expanding the Hunters

2K picked up Evolve in January 2013, and the sale had the side-effect of exposing the name to the public. It would be more than a year before it was formally announced and shown off, in the February 2014 issue of Game Informer. Behind the scenes, both before and after the sale, the studio was still making big changes.

"Initially, it was just four Hunters," Ashton said. "It was a cast of four, and they had this huge gigantic armory of weapons and equipment that the players could choose from. Once we started getting into that we found a lot of problems with it. For one, it made balancing infinitely more difficult. You had tons of variables with forty-some pieces of equipment, something like that. Players could choose anything they want. It just made an insane number of combinations."

"Not only that," Robb said, "but we found instances where players chose equipment so poorly there was no way they could win. So that started the ball rolling on iterating on that, and setting up limitations. Okay, you have to take a weapon, a piece of tracking gear, and a piece of healing gear, or whatever. That went into, okay, instead of all these pieces of equipment we just need to make load-outs: set load-outs that you can choose from. That got another step closer."

Another step closer, but far from solving the problem. Though this solution worked functionality, it opened up a design problem. They wanted their Hunters to look identifiable and special, but saddling each of the four with several load-outs severely compromised their signature looks.

The Hunter, Maggie, standing by with her pet Trapjaw

"The characters all needed to look cool no matter what, based on the different load-outs," Robb said. "That meant if you got the flame load-out, or the lightning load-out, the characters needed to look good with both of those so they had to be very vanilla. Looking at that wasn’t satisfying visually or from a character standpoint, these characters just aren’t interesting. So that’s when the idea was proposed that, hey, what if each load-out was a different character? That was a big change. That’s where we are now, where we’re able to take these load-outs and focus on the theme of the load-out and build a character around that. That allowed us to take the characters’ personalities and focus them a bit into these crazy, eccentric, interesting characters."

Carving Out a Monster

That's not to say the monsters were set in stone from the beginning. Evolve has been play-tested every day for almost four years. The studio has set up detailed telemetry data that it pours over constantly. Each press event, the Big Alpha, the beta, even the feedback during press time for review has been studied and analyzed. 

"We have a telemetry system and it monitors every round that’s in play and every time somebody takes damage, what characters were in the game, what perks they chose, where they attacked, where they moved, where they had conflicts with the monster, how quickly the monster staged up, everything is gathered on the servers and we’re able to look at all that data through a browser, change filters and monitor," Ashton said. "So we’re able to look at a character very easily and answer if that character is effective or not."

That can help the studio assess the meat and potatoes, the raw numbers. A stat that's too weak or too strong stands out, and the more data the better. Sometimes, though, Turtle Rock found themselves trying to answer a more nebulous question: what makes a character fun?

"In those instances where you try to bring in impartial guys who either have never played before or that aren’t in the studio, and those are always useful opinions," Robb said. "They’re not tainted by the everyday experience of playtesting. They have a completely different perspective. We usually learn a lot from those first-time players."

To the extent that some monsters have been unbalanced, the studio has been open about what they've found during the various public tests. Ashton noted that the Kraken was far too high at first. Simply from the win ratio, at a staggering 75%, they could tell something was off. They told themselves to be patient with the data. After all, Kraken was an unlockable character, so it didn't have any first-time players throwing off the curve. Players picking up Kraken were already better than the average, because it was filtering out inexperienced players. But if nothing was wrong with Kraken, Hunter teams should have gained more ground as players got more accustomed to it. When the final numbers came in, and it had only dropped to around a 70% win ratio, Ashton and Robb could tell it needed revision.

"It’s funny how little we changed," Ashton said. "We changed the radius of the vortex blast a little bit, we dropped the damage on the lightning strike just a little bit, like 100 points, and we reduced the radius on the lightning strike a little bit by I think one meter. I was kind of nervous, I didn’t want to go overboard. I didn’t want to make too big a change to Kraken, but I wanted to make sure it didn’t stay at 70%. So in the beta that’s one thing we looked at very closely. Fortunately in the beta the Kraken was I think at 52%, and we’re able to look at clear rankings, so we can look at high-level games and also low-level games and it was fairly consistent across both. So we were able to move him from being overpowered and made just a couple small changes, the right ones, by looking at the telemetry."

The beta period confirmed they had fixed Kraken, but it also opened the first access to another monster: Wraith. She opened to the same problems Kraken had, with a win ratio of 72%. Rather than depower her abilities, Turtle Rock traded the focus of her powers. The studio buffed two abilities and nerfed two others, alongside a lot of smaller changes. The approach was targeted more at rebalancing how Wraith works, rather than toning her down in a more traditional sense. This might seem like the studio is making tiny changes, but Ashton said their philosophy tends to think big when it comes to balance shifts.

The LocoTech dropship, used to transport Hunters to fight the monsters

"Rather than incrementally making balance changes--that might take forever if we’re really far off on the numbers," he said. "If the monster is overpowered one night and then underpowered the next night, that’s great, because we know that somewhere in between the two sets of numbers is where the balance is. If we were doing it incrementally, who knows, we might have to do that 100 times before we find the balance. When we more flip-flop between balance, overpowered, underpowered, we get a real quick sense of where the balance is between. So when we go live with Wraith, even though she was a 72% win ratio, she still was very close to where she needed to be."

As the game goes live, the two are no-doubt watching carefully to make sure Wraith, and the rest, are where they need to be too.