The discussion revolves around the loading slider and its inability to show actual progress. darkpixlz points out that the slider often reaches 90% before the page is fully loaded, making it seem like a less useful version of a traditional spinner.
Falco responds that showing actual progress is not a trivial feat, especially when dealing with dynamic resources of unknown size and variable internet speeds. The current implementation uses the average of the last few transitions to estimate the loading time.
piffy offers a different perspective, suggesting that loading bars are more useful as a psychological tool to soothe users rather than an accurate indicator of progress. They also share a video on the topic.
Heliosurge agrees with Falco, noting that accurately estimating loading times can be challenging, citing Microsoft’s past struggles with providing accurate estimates.
If I’m not even on the internet, is it really being loaded?
Especially on a really slow connection (and limiting it to edge in devtools) this effect is painfully visible where it’s 90% filled but the page isn’t loaded. This is mostly poor design, should it not show actual progress to be of some use and not just a new version of the spinner?
That was never the goal, and showing actual progress of a dynamic resource of unknown size over a variable throughput medium is not a trivial feat.
Currently, it parametrizes the animation based on the average of the last few transitions and will stop at 90% when this specific transition takes longer than the average.
Loading bars are useful primarily as a psychological soothing tool for the user more than an indicator of the precise amount of progress something has completed
A good video on the topic
Indeed it took forever for MS estimates to be close to accurate.
