• he/him

I never wanted to make a difference

I just wanted to make animated gifs.


My scrobbles


jkap
@jkap
Anonymous User asked:

if I understand right, the only way to get your posts out there to folks who don't follow you is through tags? there's no search function that shows you posts with those words in the body itself? (I hope I worded that in an understandable way lmao)

yes, tags are the intended discovery mechanism. i know that, especially for people coming from twitter, this is a tricky mindset to get into.

we don't have full-text search for a few reasons:

  • it's hard
  • it removes control from posters about where their content is visible.

posting something and not tagging it makes it effectively follower-only. if you want people to see it, tag it. that's it


dog
@dog

I don’t think people realize just how high cohost ranks on Google. It’s a blessing and a curse, but legitimately posts here are much easier to find via outside search engines than posts on pretty much any other major social media site.


karobit
@karobit

having a somewhat unique username (go homerow error!) led me to make a google alert for it years ago, which is almost always silent except for when someone in indonesia uses it? However, the few times i've had a post be somewhat popular on cohost, google alerts is right there to tell me. google loves cohost.


You must log in to comment.

in reply to @jkap's post:

Being able to go something like, "Yankees suck" in a post, leaving it untagged, and not being bothered by weird sports fans is Very Nice. Compared to Twitter when I got a bunch of random tweets when making fun of the Knicks or Heat.

there's a certain heat death of tags that's constantly approaching as people make more and more tiny variations of tags, to the point that if you search for any keyword, chances are the top 10 results haven't been used more than once. It's a short-sighted way to do things, and theres no protection against some asshole, like me, polluting those results further.

The problem is that most people coming over from twitter aren't using tags, as this is not common knowledge. I want to say 90% of the artists I've followed off twitter to here dont tag their stuff.

The difference on Twitter being that there's actually an incentive to spam. Getting more eyes and reactions actually boosts things algorithmically. Here, there is no algorithm at play, just chronological timelines. If you spam those, it'll be pretty obvious and easy to moderate, and won't really result in any big payoff down the line.

i have thought about cohost full text search and i think client side search may actually be feasible:

  • implementing it should be easier than searching the whole post db by just caching the posts the client has already seen (so zero sql, thank god)
  • the control thing is basically ruled out with that as well because you can only see posts you saw before
  • you only see posts you saw before, which is a huge advantage if you're looking for a post you remember (but not for new posts tho but would still be hella useful)

also i have played with the thought of implementing that as a userscript (i really like userscript and i also really like forgetting things) but was too lazy to do it lol

also, forget the thing about zero sql, this whole thing has another advantage: you could do client-side sql queries over the posts if you want to do that without having to worry about someone having access to the prod db lol

(inspired by twitters search being garbage. it would be 100x better if you could inject use sql)

some pretty significant problems:

  • this falls apart pretty quickly if you have more than one machine you browse cohost on
  • you have to 'age out' or compact posts to avoid going over the localStorage limit (which is browser-specific)
  • anyone could easily recover deleted posts or see posts from someone that blocked them if they were in the cache
  • similarly, any edits wouldn't be reflected unless you happened to see the edited post

tbf on 1 and 2 i think itd still be better than no search at all. 3 and 4 could be fixed by just fetching the results of a search again to check that theyre not edited/deleted (and technically anyone can easily recover deleted posts already) but i think even with those limitations it'd still be pretty useful

if you're doing the check again then you might as well just implement it on the server, though, which is already authoritative for this. and that doesn't deal with the case where a post that doesn't match the queries is edited to contain the query text, because you don't know to fetch it in the first place

in reply to @vogon's post:

yeah unfortunately it basically only works if you (or someone else) linked your post off-site, because of the way google crawls the web, so it won't pick up every post -- and google is obviously steadily moving away from the platonic ideal of "search all the text on the web from one place" and has been for about 15 years

but! it's better than nothing.