Hacker News Reader

Comment by chrisfosterelli | Hacker News Reader

I know Apex Legends balancing is pretty aggregate stat based and with each release they monitor aggregate stats like what % of X character is on the winning team over all games or what % of X character killed Y character as well as different team compositions, along with similar stats for individual weapons, to try to balance the overall engagement statistics as much as possible.

View on HN

Dota has an interesting approach. They collect tons of statistics, but ultimately the game design comes down to one person, and they complement the stats with a lot of interviews with top ranked players to get their subjective perspective.

The end result is the game has gotten better balanced continuously with time. There were a handful of patches that were a regression, but usually are on top of it pretty fast.

One of the more pernicious problems they've faced is the team that holds the base on the bottom half of the map has consistently had an advantage, as much as 5%. This appears to be related to the perspective being 3d in the shape of top down, vs a literal 2d top down view. They've tried a few different ways to balance this, but what's ended up working best is trading it off vs first pick in the lineup drafting phase of the game start.

Zynga was infamous for a stats driven approach, but there it was all about tricking people into maxing out the micro transactions :(

I made a comment about trees, but the other issue is that the map is super asymmetric. The bottom half (radiant) has a generally safer jungle and a very difficult to contest small pull (always been this way. Camp has just changed over time). The top team (dire) has a worse jungle, but it is closer to Roshan. The problem is people don't know how to use Roshan well (either capturing it or using the aegis), so that advantage isn't realized unless the team is at the top levels of the game.

I think the perspective issue is there. But its not the entire 5%

> One of the more pernicious problems they've faced is the team that holds the base on the bottom half of the map has consistently had an advantage, as much as 5%. This appears to be related to the perspective being 3d in the shape of top down, vs a literal 2d top down view. They've tried a few different ways to balance this, but what's ended up working best is trading it off vs first pick in the lineup drafting phase of the game start.

The opposite to "the enemy's gate is down"

Exactly this - the ability to re-orient your view in a game (or in your mind in the case of the "enemy gate") can be a huge advantage. I personally play much better on bottom side in MOBA games and I have never understood why it can't be flipped. Even if the map is not 'symmetrical' the players could adapt to that change, rather than having to adapt to the change in side each game.

League of Legends: Wild Rift on iOS does this. You’re always the team on the “bottom half” of the map.

Yes! Although I didn't like the game (mobile is not usually my jam), it further solidifies my theory that it can work in desktop MOBAs.

valve collects statistics, yes. Are they used? Who knows, there's hardly any transparency in the balancing process. I will say that Icefrog has (supposedly) been in charge since WC3, back when there was no conceivable way they could have collected meaningful statistics..

I also haven't seen definitive sources about consultations with top players. I've heard about bouncing ideas off pros in the distant past, but not with any regularity, and not recently.

"Better balanced continuously with time" is pretty subjective..

Did they try rotating the camera 180 degrees?

I can understand how this might be confusing. Like a mirror track in Mario Kart, it would make the map feel totally different.

Its really difficult. The exact placement of individual trees in the game matters a lot and it affects line of sight, which is a key component to Dota. If you flipped the view, not only would it just be weird, but what is visible can change I think based on the camera position. A tree that blocks a path in one camera position doesn't block it in another. A single tree's position sounds trivial, but navigating them is crucial at mid and high levels of play.

This is really perceptive and on point. I think something a lot of casual observers miss about Dota is it's a game of momentum with very small edges that accumulate. Compared to the other MOBAs it's the game that punishes you for a single error far more harshly.

It certainly is confusing. I remember that someone once made a video with the flipped map and most of the comments were about how weird it appeared: https://www.reddit.com/r/DotA2/comments/5rmbd3/

However, I wonder this is something that people would get used to, if they played it for a while. As far as I know, it was never seriously tried anywhere.

So, I suspect why they haven't gone with an approach like that is they want to preserve the layout for spectators. Dota is the biggest purse in all of esports, and they take the spectatorship quite seriously as a result.

as i recall, spectators receive game data, not a video stream. you can spectate in VR.

so during real tournaments they could just broadcast from a spectator in the “standard” view.

This is really interesting - could you share a source on how Valve/Icefrog make balancing decisions?

Zero-K does it too for the different factory matchups :

https://cdn.discordapp.com/attachments/278908415756206080/86...

(I assume this is for ranked 1vs1 rather than casual lobsterpot though ?)

Reply on news.ycombinator.com