> when I see people using something in a way it wasn’t intended (...), my question is: how is this UI failing them?

This is a legitimate question, but not the only one in that situation. Another view, more in line with the unix philosophy, is that "using something in the intended way" is a meaningless concept. A tool must work always, no matter how it is used and abused.

Thus, instead of asking "how is this UI failing them?", you could be more pragmatic and ask "how can I optimize the UI so that this usage works correctly?"

As a non-programmer, this seems like an odd philosophy. If I use the claw side of a hammer to strike a nail, I’d expect that it is t going to work well. It seems like there’s a limit on “make the tool work regardless of the use case.”

A better comparison would be: Hammers are designed for striking nails, but some customers have figured out they work well for cracking walnuts, so as a hammer manufacturer we should avoid using toxic ingredients such as lead.

I think you misunderstand that some people treat tabs as bookmarks. Let's be honest - bookmark management is tedious in all browsers, so I would rather have tabs open with stuff that I want to browse later.

This is pretty funny because you're saying it's working as intended, but also saying the problem is that bookmark management is tedious. So fixing bookmark management would fix the tab problem.

I think many of us are also aware that saving a bookmark, whether in the browser or to something like pinboard.in, is an invitation to lose it in a back hole. Whereas if it's at least theoretically in "working memory" we want to keep it as a tab. But AFAIK there isn't a really good tool--and certainly not native to a browser as far as I can tell--that lets me quickly and easily rationalize all my tabs into coherent working sets. e.g. Project X, Project Y, email/calendar/etc., to read, and so forth.

TreeStyleTabs in Firefox (which is a great tool regardless of bookmarking), keeps track of parent/child relationships, you can drag & drop into subsets to create the relationship you're talking about, and has a "Bookmark this tree" feature, which would create a folder with subfolders with the top level named as you choose.

(I use the heck out of this, both for the extra vertical space that comes from hiding the top tab bar, and for the conceptual organization that it provides).

> I think the problem is that people generally don’t know when they’ll return to a tab, so they leave it open, and that creating a bookmark is asking too much commitment for something potentially ephemeral.

Safari's reading list is my solution to this. CMD+SHIFT+D adds the current tab to the reading list which is visible on a new tab page or in a side panel. Syncs across devices too. You just right click > delete to remove the item from the list when done.

As someone who doesn't use Safari this just sounds like a secondary Bookmark list.

Kind of but I think the implementation matters. I use this and bookmarks in very different ways. I can also add to it from the iOS share sheet so add links I see on the web/social media that I want to read later.

I think that's related. I think also for a lot of web pages, they have ephemeral state that would not be captured by a bookmark, so they get left open and accumulate. One can never know sometimes which page will succeed as a bookmark or just fail.