: I think 343 should use all that extra processing power to, for instance,
: remove taunt enemy corpse from warthog chaingunner AI.

: Or, in all seriousness, improve their situational awareness - don't shoot
: that crawler over there minding its own business - shoot that knight that
: has us in its sights!

: Our Spartan AI companions are like the keystone Cops.

: Who am I kidding? That's never going to change. Bungie and 343 both decided
: that the AI was good enough and spend as few resources as possible on it.

I've chatted with one of the main people responsible for enemy AI under Bungie, and when it came up he seemed perennially embarrassed that he couldn't get out all the kinks, and he always wanted it to be better and tried to make it better.

(Apparently part of the problem is that Warthogs can't strafe.
If I can extrapolate on that, I think what makes that such a big issue is that it requires the AI to plan ahead, to figure out how they're going to get from A to B, rather than just moving towards it until they get there. Planning is an incredibly hard skill to replicate with the best programmers and hardware; I can only imagine how much harder it is to get it to work on hardware that's middle-of-the-road at best with a team that's mostly dedicated to other things.)

: OTOH it feels like a manager thing and an agile thing - dev says "hey
: boss, taunt enemy corpse sucks when hogging on legendary - I'd like to fix
: it" - boss says, "put it on the backlog and I will prioritize it".

: dev says, "but it will only take two minutes to fix"

: boss replies, "that's not agile" ;-)

From what I've heard about other kinds of development, this is a relatively common occurrence. There are frequent problems on the managerial side, but there's also the occasional problem of what they have to work with. Sometimes a programmer becomes too attached to a feature or even an implementation of a feature (as a painter might with a particularly nice mountain that doesn't fit with the rest of the painting) and they might overstate its importance or understate how hard it would be to fix. For a big project the manager can't know how all the code as well as the rank-and-file programmers do, so there's this poker game they have to play in determining what to prioritize, along with weighing in what they do know about the code and what their manager wants.