I hear you. My last phone was an ancient Xperia from about eight years ago–but I finally needed to get a new one. I love most everything Sony is doing with the Xperia series, and I would have gotten another. But the update guarantee was so abysmal I just couldn’t do it. So now I have a Pixel which I think is worst in design in a lot of ways. But at least it will be updated. (And can run GrapheneOS).

Seconding Sway. I will admit I prefer autotiling (switching the split for new windows between horizontal/vertical automatically, rather than choosing which split you want), but overall Sway is so good in configuration that I still use it in spite of being a manual. The configuration takes time, but that’s common to pretty much any tiler.

The first thing that came to mind when I saw the question is perhaps a bit of a weird answer–but I really want to learn SELinux. It’s completely overkill for my Linux desktop and the few services I run on my network. The same with OpenLDAP, I want to play around with it even though I have no real need for it with my setup, I just haven’t gotten around to it yet.

On that note, I also feel like I want to learn Ansible, or some other configuration management tool. The thing is, I haven’t even played around with it (or any others) enough to really even get what the intended use case is. I’m looking for ways to manage policies and configurations across multiple machines in a common way, but it feels like the more common use case is deploying webapps. So while it’s on my list of things I want to learn I don’t even have sufficient background at the moment.

Then, finally, the other thing that came to mind was timeshift–or really BTRFS snapshots in general. It would be nice to have that additional feeling of safety while playing around with my systems.

I appreciate the author took the time to do this, but I’m a little skeptical of their results. I pay for Kagi, so decided to directly paste their queries into it just to see – while none of the ‘bad’ results would have completely flipped based on the author’s ranking system, the “very bad” would have been just “bad” or possibly “ok”. Adblock showed my uBlock as result number 5; youtube-dl was result number 10, and so on. Mind you, I am running it with uBlock origin and although I haven’t done any custom rankings it is quite possible that some of the other settings (geographic point of origin, for example) would have affected the results. I’m curious to try it with some of the other engines now (as well as a completely blank slate for Kagi) and see – because at least in that instance, I’m not entirely sure these are replicatable results.

Also: a bone to pick with the tire query – the answers the author were scoring higher didn’t answer the question “why do wider tires have better grip” (which is what the author searched), but instead queries which answered “what are the advantages of wider tires”. Not sure it would actually change the proportionate results, but assuming their bullet list is all they were looking for in the answer, it’s hard to criticize any results for not answering the slightly different question than what they posed.