The current forum search is not the best, but a search for "forum categories" would've gotten you this explanation: https://forums.zotero.org/discussion/comment/420237/#Comment_420237

Categories, metadata, tags — those all just complicate things and require additional moderation time, because people frequently don't actually know whether what they're reporting is a bug, a feature that already exists, something that works differently from what they expect, etc. Just having categories was a mess, and it created more stress and work for everyone.

Beyond that, there've been 82,487 threads in these forums in the last 16.5 years, or 5,000 a year. Including comments, there've been 444,904 posts — that's 74 a day, every day, for the last 16.5 years. The idea that these forums "discourage extended discussion" or that they're designed to encourage people to go elsewhere is absurd. The core developers and long-time contributors are here literally every day discussing features and suggestions directly with users. From How Zotero Support Works:

Beyond technical support, the forums allow the entire community — Zotero developers and users alike — to help shape Zotero's future. Many of Zotero's features began with discussions in the forums, and many changes are the direct result of feedback from users there, so we strongly encourage you to get involved.

Generally, given that this is your first post here, I'd encourage you to actually participate in the forums before passing judgment on them. There are various technical things we plan to improve (Markdown support, images, featured answers, etc.), and we're happy to discuss other features, but the forums really do work pretty well.

What made the Zotero forums so different that made that the case?

People would post in one category -- then they'd realize their post was in the wrong category and they posted again in a different one. It's often not possible to know what category something should be in:
You want to do X in Zotero. Is the fact that you cannot
- a bug (because it should just work and doesn't for you)
- a documentation question (because this is already possible, you just need to know how)
- a feature request (because it really isn't possible)?
Again, that meant that relevant feature request debates ended up as bug requests and vice versa. I can't count the number of times we told people to just not worry about the category they posted under -- so yes, getting rid of categories was definitely the right thing to do, you'll have to trust those of use who have been there on that one.

This seems like the crux of the problem to me. It means that you can't discover popular past threads that may have generated lots of great discussion, the ones where the really useful tips, tricks, workflow advice, plugin discussion, etc. are.

I think that's a reasonable point, but (and this isn't defensiveness: for the CSL project where I have a direct say on infrastructure, we're running a discourse forum) I don't think technology is an easy solution there. There are a fair number of threads that do extend over many years (see e.g. the one for the ODF-scan plugin I co-maintain) but I'm not actually sure that's been that much more helpful in getting people to find the relevant solutions or workflows on page 20 of 25 (say) of that thread.

A lot of workflows used to be spread via blogs & social media (and you can still find a fair amount of that, but of course blogs have declined in popularity and social media... oh well). Is there actually any complex, long-running software where you think that works? The FLOSS software I probably use most outside Zotero are Firefox and R/R-Studio, both supported by significantly larger organizations, but I wouldn't know how to find specific workflow recommendations beyond asking experts and google for those, either.

Some of the things that dstillman mentions as feature plans are major annoyances. Decent support for images and markdown would save a *lot* of time and the search is indeed mostly useless for anything older than 2 weeks (I use google site search whenever I'm looking for something further back -- that works reasonably well).

I think those things would definitely help streamline conversations and make them easier for both newcomers (the screenshots, esp.) and frequent writers (oh, please give me md), and would also make discovery a bit easier -- but even that latter point is a two-sided sword: we do already have a significant (inadvertent) thread-jacking issue here and dstillman spends a fair amount of time splitting off unrelated reports/requests/discussions from existing threads. Better search might ironically make that problem worse.

To me, the only improvement that is almost necessary is to improve the search/query system. I am frequently frustrated by trying to query a phrase. Do I place the two or three words within quotes? That doesn't seem to help. I'd like to use simple Boolean AND, OR, NOT terms. If I had a month to donate, I'd volunteer to develop a simple thesaurus of synonyms (forum posters use different words that have the same meaning). Ignoring language, people frequently ask about something that has a "preferred" Zotero term but the questioner tried to find posts by searching for a different word -- and sometimes words have more than one meaning and there would be great benefit in getting a disambiguation screen (were you searching for this, that, or something else. It might also be helpful to allow for time restrictions -- maybe the default search could be limited to the past n years and dependent upon the recency of the Zotero version. (Is it useful or is it confusing to be presented with posts concerning problems from the days before Zotero was removed from browser-based?)

I do think some extra forum structure might provide more scope for discussing broader topics like "best practices for collections". Because I think most people probably just make up how they organize things as they go along, and are not sure if they are doing things in the most efficient way (which may differ for different applications). If they've inadvertently chosen an inefficient strategy up front, they can be up to their neck in it by the time they realize that things are not working for them (which can be hard to come back from). For example, should one just create as many new collections as one dreams up ... with the result that one eventually ends up with 100's of sub-collections, many layers deep ? Or should one actively try to minimize the number of collections ? Can a protocol for archiving collections that are no longer being used help (for example if one has started a new sub-collection for every paper one writes) ? Should one always place items in a 'top-level'/parent collection first, and only then copy an item to a sub-collection (which means sub-collections can be more safely moved/deleted later without worrying that you might be losing something that only exists in that sub-collection, and of course your main library) ? What are the pros and cons of setting View\Show Items From Subcollections ? Should one devise as many likely useful tag names as possible very early on, so that there is less need for difficult 'catching up' later, ie finding existing items that should have a newly-created tag ? What are useful ways to use saved searches ? Must metadata be carefully checked on every item as it comes automatically into Zotero (difficult when many papers are being saved at once) ... or might some people decide that just cleaning up wrong metadata they see in their reference lists in papers they are writing is less onerous ? What are peoples' workflows for integration with external software like Obsidian, Notion etc (although personally I am sceptical that many of the current 'zettelkasten'/note apps will still exist in 5 years time). etc etc etc

Thankfully I've avoided most of the sort of problems listed above, partly by accident, and partly by Zotero coming for me decades after boxes of handwritten index cards, followed by some reference storage system on a Vax mainframe that I don't remember, then Reference Manager (DOS), then Endnote. ;)

Ironically, the fact that the developers are so good at responding quickly to problems here means that other users are perhaps less likely to chime in with suggestions. So if there was a forum section that was not obviously for dealing with problems then there might be wider discussions from more people.

I also wonder if a plugins forum is necessary. Plugins are both a huge strength of Zotero as well as a common source of problems. While the source of some of the latter might not be obvious to the person who posts a problem symptom, such posts could easily be moved to a plugins forum when that becomes obvious. Likewise feature requests that can actually already be met by an existing plugin could be moved if not in the right place.

The current zotero forums lack a few features that would really, really help (like markdown).

Please change this running system. Maybe conservatively. But please do.

What also would be good is if the search results for forums do not show every single discussion post to a search, but only the threads that contain the search results, ideally with a count of how many times the search term exists in the thread.
This would allow the user to more quickly see which threads matter, how many relevant threads there are, but also to more quickly be able to move between different potentially (ir-)relevant threads.