EzraPound said:
Still, Valve are brilliant.
No they're not, they just test a lot. That shit from Portal about testing is actually a whole pile of company in-jokes about how much they depend on testing.
Basically, this is the Valve process:
[ol]
[li]Find a product that already works[/li]
[li]Test it[/li]
[li]Decide it's stagnant and unoriginal, change shit[/li]
[li]Test it, find out that nobody likes the changes[/li]
[li]Go back to the old formula, alter basic things arbitrarily[/li]
[li]Test it, find out the alterations stink[/li]
[li]Scrap the alterations, put in more of the stuff observed to be enjoyable[/li]
[li]Test it, find out someone objects to a minor art asset[/li]
[li]Overhaul the art style completely[/li]
[li]Test it, find out nobody likes the overhauled art[/li]
[li]Put the old art back in[/li]
[li]Test it[/li]
[li]Eventually decide the product was fine as it was originally and just do a build of it with more polish[/li]
[li]Test it about a hundred more times[/li]
[li]Release it[/li]
[/ol]
It's a strange, strange atmosphere. There is no creative center whatsoever, there is only this hive mind service to the product. There is no creative individuality, yet the whole process depends on the employees, who are under no deadlines, to fall back on independent thought and creativity to address creative decisions that would ordinarily come from a team lead. The music guy makes the music he thinks will work because XYZ, here's his reasoning for it, now let's test it. The art guy designs sheets upon sheets of monsters that he thinks will work because ABC, here's his reasoning for it, now let's test it. And every step of the way, they need to achieve confirmation first from everybody else on the team, and then from their testers.
The practical upshot is that there will always be a good product turned out from this, and every once in a while their mad experiments yield something innovative or cool or something a tester says breaks through and a dream game gets made. But, you have to sit there and ask yourself whether certain shit really did need to be tested or if they could have hacked out a few months in the process with a tiny bit of common sense in place of the testing.
Examples? Portal 2. Originally had nothing to do with the original Portal, or with using the portal gun. In fact, it was all about spraying walls with the gel, and there was one that made you able to walk on walls. You think I'm kidding, but they did it and they tested it, and everyone went "this is nice, but where's my portal gun?"
Alien Swarm? They put the game in a diorama, re-designed the aliens based on finger and toenails, then made most of them huge and tried to get away from the "space marine" aesthetic. Turns out that nearly all of those decisions, each one from a different individual contributing to this project, were complete bunk, and they went back to mostly the original format--tiny bugs and space marines included.
I almost guarantee you that with DotA 2 they tried shit like a first-person camera and a different art style before they tested it, found out it sucked, and went back to the Warcraft-inspired art style and strategy cam.
Don't get me wrong, I've got a healthy respect for Valve, but "brilliant" doesn't feel like the word for it.
