In an ideal case I can only think of a few uses for LLMs in games. Like dynamic villager dialogue in an RPG. But this also takes the user out of a curated experience, so this is still peripheral or novelty gimmick stuff. I've yet to see a core concept proposed that actually requires an LLM.

Ran into that plenty with early feature phones when the hardware capacity was developing very quickly, and previous development was... as simple as possible.

Also any time I think about tying behavior to frame rate, it makes me think of taking older things where framerate is the basis of the calculation instead of time, and putting them onto newer hardware and HOLY COW THIS IS UNPLAYABLY FAST!

"Developers often make a lot of assumptions"

Testing has to run on a lot of assumptions, too. Especially on a live service application. "You submitted a change to feature A. Feature B is unrelated and you marked no change there, so I didn't look there because I can't check everything every time"

I admittedly got lucky. I wanted to work in games, and my first job out of college was a game tester. Turns out software testing fits my skills and mindset very well, so I've been adding to my skills and confidence for it since.

I can say that the worst crunches I ever went through were due to very clear mistakes in project or process management.
When that's the foundation, yeah, success is despite it, not because of it.

Today's spoiler is fun with double negatives

I haven't checked too deeply, but I've seen it proposed that the valuation was influenced by promises of AI massively improving efficiency, which... seems unrealistically optimistic to me.

I'm guessing we'll see EA rocket back to the top of the "worst companies" polls, soon.

To your last point, feels like something like The Sims may be at risk, which is not wise when it's a solid performer and the margins are getting real slim.

This almost feels like a vanity purchase to control the FIFA franchise.

I my head, Jim Powers was weirdly long tenured. From the Young Stallions to through peak WCW, at that spot in the card, I have trouble with that for some reason.

And yet I have little problem grasping the tenure of Gangrel.
I don't get it. Perception is weird.

Sometimes it's a pain, but sometimes good art thrives on restrictions and it'll get design thinking creatively how to incorporate them in an engaging way.
And then sometimes those solutions will be nightmares to implement, so you give up and go back to a simple & boring solution so it at least works

And as improbable as it is, it's not zero. I used that as an example because I have kept a game on, and active, for 3 days straight once. In college me and another guy traded off playing Final Fantasy Tactics over spring break. Playstation didn't turn off or reset for several days.

The probability of the user path is still a factor. Like, if you have to keep the game running constantly for 3 days straight to get a crash to happen... at a certain point it's "congratulations, you can push the system really hard"
That's where an accurate report and a producer's judgement come in.

But there were too many steps with too many dependencies, so the complexity of the process was a lot higher than he was estimating. It was REAL easy to mess up.

A personal example is when a manager set up a process where every individual step was perfectly simple, so he treated the whole process as simple, like the most complex step was the high water mark for process complexity.