> If you need the info again, typing some related words into the address bar will usually surface the results from your history.
This just isn't useful for the kinds of information retrieval I find myself doing, and I truly don't understand why people think that ought to be enough for everyone. Perhaps other people don't collect and combine large amounts of information into new artefacts as part of their regular online activity?
When I'm retaining state such as tabs, it's not so I can find the same information by searching for it.
It's so that I can go through information I have already found through searching and following links, which has passed my filter as the useful things I'll need for the next phase in processing, to do something useful with.
The important part isn't that the information is findable again by searching. The important part is that I've already half-catalogued what's useful once, and it's that cataloguing which is what I need to build on a second or third time.
All the "next phases" in processing typically take from a few days to a few months to complete.
To give a concrete example:
Recently I had to do some research on QUIC protocol and HTTP/3 for a job. In my searches I found various pages of particular interest that I should at some point extract information from, and add them as references to a report I'm writing. But there was no point adding them at the time to my report, because until I had found more documents and pages to read, I wouldn't know which ones were the best ones to include as references for someone else to use, and which ones to drop.
Also, that list of 100 represented things I would need to systematically check through again in my code implementation and report before dismissing them. By systematically, I don't mean in linear order, but I do mean complete.
So I ended up with about 100 tabs, expecting to go through them systematically later, and turn them into about 30 references for my report eventually. As the task would be completed in about 5 weeks (in parallel with a number of other jobs), those 100 tabs were context for that task for about 5 weeks.
If I deleted them "to reduce clutter", the fact I could find the information again by searching for it was pretty much useless. That would be starting my task all over again.
And if I bookmarked them, that would loses track of the scroll position, which meant I'd lose which page sections, headings, etc were the relevant item I was keeping to use, or decide not to use, in the final code and report.
Really I had only two options: Keep a lot of tabs open for this one task for a few weeks. Or write down a lot of notes as I went, intending to delete some of the notes eventually. I've tried both, and find keeping the tabs to take much less time and overhead compared with taking notes in the first round.
Now imagine I'm doing several research-like tasks like that overlapping in time, each of them taking a few weeks or a few months each. The overlap means that number is multiplied by the amount of concurrency, so it reaches several hundred or a thousand easily.
I'd love to use something better, but tabs are the thing I've found most effective so far. In the past I've taken more notes, and I found that led to a lot of time taking notes. It's like the difference between a more efficient or less efficient batch sorting algorithm for information discovery, indexing, merging and eventually selecting out the most relevant things.
People love to suggest alternatives. I'd love one because tabs do have many problems. But I've yet to find a good alternative which is actually helpful. Bookmarks are not because of lost context and poor UX for going through them (large thumbnails with retained scroll position and/or highlighted current section heading and/or tags/notes would be great). Deleting and searching again is useless. Taking notes at every stage is ok but labour intensive by comparison. I tried Org mode but it has the same problem as taking lots of notes. What else is there for organising work-in-progress information like this?