Thanks to #8322, the languages in the filter list are now sorted top-to-bottom left-to-right, making them much easier to scan through to get to your language! It's a big improvement. :)

It's still pretty confusing though, because the list is sorted by the language's name in English, even though the languages are listed named in their own languages. So, for example, Suomi (Finnish) comes just before Français (French).
I feel that the list should be sorted alphabetically by what is visible, rather than by the hidden labels in English. I think having the languages sorted by the letter they start with in their own language is not a problem because if you speak those languages you know what they are called in their languages. So a Greek speaker would look under E anyway.
What you are looking for is phonetic sorting and I don't think it's easy. Normal string sorting algorithms won't put 日本語 under "n". Right now languages are sorted not by their English name, but by their ISO code. It's close, but a little more neutral.
I guess what's useful to know then is, would a Finnish speaker know to look at that part of the list, instead of under S for Suomi? I don't know how other comparable lists sort their languages, so perhaps some other people who speak other first languages will add helpful comments! I might be flapping unnecessarily. :)
I checked the list order on Twitter and it looks like this. The languages are in the order of their first letter in their own language, so Finnish is under S for Suomi. This makes a lot more sense to me than sorting by an ISO code order that isn't even visible to the user or memorised by most people who use the internet.

It seems the sorting is taking place here (filterable_languages.sort). If we drop the sort function, and make sure the languages stay sorted phonetically in this file, the languages should be ordered nicely, right? The only issue with this is that the result isn't sorted programmatically, and will be required to be kept in order in that file.