>>4992
You say this like there's nothing to be said about it. There's plenty to be said, though I feel the topic is only tangential to /h/.
It's a real pain in the ass that people uploading things don't do some basic tagging. There is absolutely zero practical use to uploading things to boorus if no one looking for that content can find it, and possibly no one may go out of their way to ever tag it. The only reason I can see for doing this lazy shit is your account might get good goy points for uploading.
Lack of tagging was an absolute bane upon download subscriptions and those who use booru tags for Hydrus. However, Hydev saw this former issue and created a workaround so that tag subscriptions that check for files with x tags every once in a while can still get files that were not initially tagged properly, but then tagged later. Whenever it runs a check for new files on a booru and finds one, it will then keep checking previous files with the tag(s) before that upload until it hits an arbitrary limit you can set (default is 100) while ignoring any urls it has already seen. This limit is also the same as the number of files it will grab at once, so if it maxes out, you may want to run a manual download of that tag with a larger search limit.
Even for those who use the booru tags for their personal Hydrus collection, lack of tagging is now becoming less of an issue with machine learning based auto-tagging, since ML can now recognize and generate images. The mistakes it makes with missing tags are far less than boorus and the mistakes it makes with mistagging seem to be roughly the same as boorus since it's trained off of them.
I personally do not like either, nor do I care for the Public Tag Repository community tagging as it seems pretty barebones. Boorus, the PTR, and AI tagging are going to use a lot of tags I don't like, like using the xgirls tags for futanari. Everything in my cartoon porno collection is manually tagged except for artist tags from subscription downloads, and I've tagged about 23K files so far. About 14K of these are /h/ material. I still have about 5k files to go, but I've been hovering around that number for over a month because since Hydrus makes file organization and searching so much more perfect, I've been more motivated to crawl boorus looking for artists and fetishes I like to add to my collection. I've already gone through about 140+ artists so far over the last half year. I currently have 27 to go before I move on to my list of, currently, 44 characters, then 25 fetish tags/promising tag combinations. I rarely add more artists while going through the artist list, but I plan on adding more as I discover them while going through the character and fetish lists. The most artists I've had in my backlog at once has been about 110 or so near when I started. I mainly check Gelbooru, Rule34.xxx, and AllTheFallen as my primary boorus. On occasion I'll check other boorus if I happen to come across the fact that an artist has more shit there. If I can hardly find anything for a good western loli artist, I may check Baraag.
Having these things in Hydrus makes it incredibly easy to find the files and topics I want, and even filter out what files I already posted in particular threads. Immensely more so than the old archaic folder sorting. Since I took over /h/
primarily to clear out the nigger porn, I've usually made at least one post a day in various threads. Once I finish my active collecting and tagging in Hydrus and return back to more passive collection of smut, I plan to pick up the pace of my dumps on /h/ and make a bunch of new threads. That's still probably 6 months away though. Not sure about my timeframe, as since I've started collecting my estimates towards finishing have been hovering around 3-4 months for several months now. I'm going to start including how many files I add a day and put in my calculations to get better estimates.
But I digress. Yes, lazy tagging sucks, but it's quickly becoming a non-issue thanks to machine learning. I imagine it won't be long before many boorus themselves adopt ML tagging, if they haven't already.