Generally, it's more of a case of making the optimal build not so obvious in game. So that people can play the game without just clearly seeing or easily stumbling on the best build in the game.
Ideally there will be several builds that all perform similar to each other so thats either impossible to tell without a mathematical model which one is really the best, or that the best build is only a slight upgrade over other builds, instead of it being so much more powerful that everyone can instantly tell it is the best build.
When I played D:OS1, I didn't initially go looking for the optimal build. As a devout non-min-maxer and roleplayer, I just started playing the game, and after a few hours of playing was easily able to identify those abilities that were way superior to other abilities. So, I restarted my game and this time made sure to get those abilities for all my party members and develop my characters that way. Then, out of curiosity, I went and looked up "optimal" builds. And guess what? My builds ended up being very close to the optimal build that others had created and posted online.
By contrast, in the Pathfinder games, it is practically impossible to end up with optimal builds simply by playing the game. You can only come up with optimal builds by looking them up online. So, it depends on the complexity and depth of the game's character building and development systems. In the D:OS games, those systems were very simple and shallow, so it was easy to innately come up with one "best" build. In the Pathfinder games, by contrast, those systems were so very complex and deep such that it was impossible to come up with one ideal build for any character unless you looked up something that someone had spent a ton of time on crunching the numbers.
