[akarlsten]

The real reason people moved to the infinite ones is that the determinate progress bar is almost never accurate or representative, hence useless.

Like beyond truly "dumb" tasks like downloading a file it's basically a guessing game how long it will take anyway, right? Say you split the whole loading bar into percentages based on the number of subtasks, suddenly you end up with a progress bar stuck on 89% for 90% of the total loading time.

Obviously you could post-hoc measure things and adjust it so each task was roughly "worth" as much as the time it took, but people rarely did that back in the day and my boss would get mad at me for wasting time with it today. Hence, spinners.

[TeMPOraL]

> Say you split the whole loading bar into percentages based on the number of subtasks, suddenly you end up with a progress bar stuck on 89% for 90% of the total loading time.

Sure. But now as a user, I get to see the glimpse of what's going out under the hood. Combined with other information, such as a log of installation steps (if you provide it), or the sounds made by the spinning rust drive, those old-school determinate progress bars were "leaking" huge amount of information, giving users both greater confidence and ability to solve their own problems. In many cases, you could guess the reason why that scrollbar is stuck on 89% indefinitely, just by ear, and then fix it.

Conversely, spinners and indeterminate progress bars deny users agency, and disenfranchise them. And it's just one case of many, which adds up to the sad irony of UI/UX field - it works hard to dumb down or hide everything about how computers work, and justifies it by claiming it's too difficult for people to understand. But how can they understand, how can they build a good mental model of computing, when the software does its best to hide or scramble anything that would reveal how the machine works?