When time is money, optimizing for time is a worthwhile endeavor. 6+ year dev cycles aren't something anyone wants. But realistically, large file sizes probably aren't going away with more dev time.

Pray tell, how do you think "they" should optimize game assets? People who have zero experience in game development love to talk about "optimization" like it's a slider in the options or some challenge that can be solved by throwing resources at it. Modern compression algorithms like Oodle are about as performant and artifact-free as mathematically possible. Everything beyond that is a trade-off between quality, size, and performance.

But are you thinking about it in abstract? Would you prefer a game to have occasional pop-in, stutter issues, and CPU spikes in exchange for a smaller size?

(edit) Before you say it, I know there are relatively low-fidelity games that run like ass. Those are usually caused by poorly written game logic or shader codes, which can be refactored, but won't have an effect on file sizes.