Development

Few words on map rebalance. Shenzhen and the curious case of broken symmetry 

Update 2.7 for War Robots included balance tweaks for three maps. In this article we will go through the changes on Shenzhen to give you a better idea on what we’ve done — and why. 

Sit tight, we are dropping into the wonderful world of map design!

When we started working on War Robots we tried to avoid symmetrical maps. Firstly, we wanted in-game locations to seem like actual places (and there aren’t many real places that are perfectly mirrored). Secondly, asymmetry brings more variety: in theory, you should get an entirely new situation on, say, Springfield depending on where your team has randomly spawned.

But later on we stumbled upon an issue: unlike the maps, current War Robots’ ruleset is symmetrical by design. It’s not like one team is meant to attack and another is supposed to defend — not at all. They are both attacking and defending at the same time, maintaining the careful balance between the two. Both teams are chasing the same goals and have an access to the same tools. Hence the question: why would they have different starting conditions?

We believe that with our latest maps we struck a pretty good balance between their functional symmetry and the sense of place. For example, Dead City and Power Plant still have almost no deviation in win rates for teams starting on different sides of the map. But some weren’t that lucky.

What was wrong with Shenzhen?

Here is the bird’s eye view of the map.

yamaspring_18

In its pre-2.7 state Shenzhen had a major balance problem: from our statistics we knew for sure that one team (let’s call it Team 2) won about 18% more often than the other (Team 1). Why so? The map is mostly symmetrical. Shouldn’t that mean that opponents were put in the same situation at the beginning of the match?

Let’s see.

yamaspring_7_1-min

One side of the map mirrors the opposite side (aside from skyscrapers placement, which doesn’t play a huge role here). Symmetry axis goes straight through the map’s center. Everything checks out, right?

Actually, no. As you probably noticed, the devil is in spawn points’ placement. They were symmetrical — but not in the same way as the map geometry is. 

yamaspring_8-min

The map itself is symmetrical over the axis, like on the picture above.

yamaspring_9-min

The spawn points, however, were symmetrical over the central point, which is placed slightly above the central beacon. Now with this in mind, take a look at this picture again.

yamaspring_7_1-min

To put it in the simplest way, the map itself and spawn points on it were mirrored using different mirrors. That brought up many issues.

Let’s go through the most significant ones. 

Exhibit A

Both teams had to cover the similar distances in order to reach the closest beacon. However, southern Team 1 position had significantly bigger opening on the way to the beacon, putting players into a more vulnerable position.

yamaspring_10-min

Exhibit B

Team 2 used to start much closer to the central beacon due to point of symmetry being slightly shifted to the left. 

yamaspring_7_2-min

Exhibit C

If teams chose to go for the central point after capturing their closest side beacons Team 2 players again got an advantage. Central beacon is much closer to their nearest covers, so they spent significantly less time on the open field before reaching the point. 

yamaspring_12-min

We could bring some other examples, but these are enough for the conclusion: before 2.7 Team 1 players were at significant disadvantage when it came to map control. Even if they managed to push the opponents away, Team 2 had a good chance to take everything back with their superior defensive positions. 

So. How did we fix it?

We moved spawn points just so the teams start directly against each other. Notice that they are split into two groups, placed on the different sides of the road: this way players naturally realize whether they should go for an eastern or a western beacon. You are free to regroup if you want, but in public games you don’t usually get much communication, and this touch should set the proper flow for the match and prevent clumping on a single point. 

yamaspring_15

And since both teams contest for the same side beacons, first close combat skirmishes tend to happen much earlier in the match. Our playtests showed that it works quite well for the gameplay dynamic. However, many players tend to funnel towards the map center, entering a long game of tug-of-war. We might need to add more place for tactical consideration there, just so the fight for the central beacon wouldn’t be fully decided by raw firepower. But to that we will get back later — there are some other things we’d like to try before the end of the spring. You’ll know about these soon.

And for now, did you like the changes? Did you hate them? Why? Let us know and don’t let your guard down. Because in our next article, we’ll talk about Springfield. 

Categories: Development

Tagged as: , , ,