Pure free-text search is not useful: generalist search engines already do it well. This is about search that filters on metadata that generalist search engines don’t take into account.
I’m grouping several related problems to solve here. Their common point is the desire to find threads not based on a free-text search, but including other criteria (in addition to some free text) such as:
tags;
stuff I’ve interacted with (read, upvoted, posted, …);
The average user does not understand complex search. And even those who do have been spoiled by Google. The end result, IMHO, is that the primary search should be a freeform text box but let the back-end weight the search results appropriately - a search for “widget” might get top ranking for all posts with the “new-fangled widgets” Tag followed by “How can I fix my widget?” title followed by random text that happens to include “widget” in the body. I don’t think “Stuff I’ve interacted with” has much bearing on it - if I want the stuff I’ve interacted with, I can go to my Profile and look there (I do that frequently on SE). Obviously age & score are relevant too, all else being equal regarding keyword frequency, etc.
I think the only one of those that should be a candidate for MVP (and I’m not sure if it should be in) is tags, by which I mean: click on a tag and see a list of other things with that tag. That’s enough for people to start to understand a tag collection. Lots of other stuff is nice and should come later, but I don’t think we need it on day 1.
I don’t know how much is in the MVP, but for a fully operational search, true tag-based search is a must. Giving weight to tags has far too many false positives. It’s one of these misguided ideas where search always expands to find more stuff whereas the user’s problem is to narrow it down.
It’s not just about finding solutions to the problems I have — actually for that I usually use a search engine. But some topics have too many false positives and for that tag-based search is useful. On SE I typically use on-site search when I’m looking for a thread that I’ve already seen, or a potential duplicate target.
The more I think of it, the more I think (accurate!) tag-based search is necessary for tag curation: to figure out what questions are currently in a tag, and what questions should be in a tag, and whether the tag makes sense.
Oh, I agree that we need real search. This was tagged MVP so I assumed it was an MVP proposal; did you intend that, or is this a longer-term proposal? I definitely agree with the latter; it was only MVP that I was questioning.
This hasn’t seen any interaction for a while. As far as I can tell, there is consensus for the following to be in MVP:
a simple text-only on-site search
a way to show and search all questions within a tag
Is there something I haven’t seen? Someone wanting to object? If so, please do it now, as this topic will close in 36 hours, unless there has been an objection. I’ll then insert the consensus into the Wiki.