[bmcahren]

I was once speaking with a group of AAA game developers at Steam Dev Days discussing how to balance random interactions with gameplay. I think everyone with me had worked on larger teams and discussed complex solutions to create balance. My solution was much simpler and was similar to the random walk example in the article - each event of being grabbed by a zombie was generated based on a random chance that only initialized after X seconds from the last time you were grabbed.

The group of AAA devs laughed it off as not being feasible but our game had market validation (10m+ downloads). It made the game fun and eliminated the "I can't believe I got randomly grabbed twice within 5 seconds" type of interactions that make you want to quit.

We also observed a lot of panicked players jumping, twisting and doing everything but hitting 'shove' the 'V' key - especially streamers. So we added in support for being able to jump repeatedly/shake to get out of a grab.

Random is too random when it comes to gameplay enjoyment for sure.

This is a very good write-up on solutions to the random placement (2D) and random interaction over time (1D) problems without resorting to less savory solutions like I made.

[matsemann]

Same with Tetris: Don't randomly select the next tetramino (piece). Then you can get the same multiple times in a row, or long streaks without ever seeing a certain tetramino. Instead, make a "bag" holding one of each tetramino, shuffle it, and use that as the next 7 pieces. Still random, but it's bounded, and as you say not "too random".

[livrem]

Just don't lie to players. In Tetris it is well-known that a bag is used and that is part of the game rules that everyone but the most casual players will know about. Other games, unfortunately, secretly filter randomness behind the scenes to make Gambler's fallacy real, pretending that random events are independent when they're not.

[lifthrasiir]

> The group of AAA devs laughed it off as not being feasible but our game had market validation (10m+ downloads).

I would be very surprised if they actually said "feasible", and wonder if you happened to paraphrase as such. Your solution is indeed great if you only need to ensure that grabbing shouldn't happen too frequently, and no different from a temporary invulnerability after the damage in the classic game design (i.e. a validated strategy), but also not much flexible from that point. I'm sure that they have many other things to worry about as well.

[aomix]

This topic reminds me of how dota does it. Which may be common but was the first time I had read about it. When you see “X% chance to happen” the true chance is much lower but increases with each miss. So on a shorter time frames where you’d be paying close attention to the frequency it’ll be correct. But you also are very unlikely to go a long time without it happening or have it happen many times in a row c

[swinglock]

Fallout developer Tim on randomness.

https://youtu.be/DqL9R5PqE20

[MichaelRo]

He, heh, I read your sentence as "I was once speaking with a group of AA game developers at Steam Dev Days".

Recovering alcoholics and game developers ... seems something that goes well together :)