How is it, that out of the entire internet there is no porn website that manages to execute as robust a tagging system as Rule34?

It would not have been my first guess, and I can still hardly believe it. For whatever reason, only Rule34 has been able to consistently tag content in depth and then allow users to effectively search and filter said tags. Why Rule34, why has no one else really been able to do this?

Yes, yes, I know there are other websites that use this same model like gelbooru/danbooru/sankaku blah blah blah. My point really is that I don't understand why this tagging approach was never executed so robustly for non 2D/CGI content.

What is holding back other 'real' porn websites from having a powerful tag based search/filter interface? And no, the tagging system on pornhub is a joke.

I think this is a sign you look at porn to much

Hot take: Many porn sites make finding the exact content you want more difficult on purpose in order to keep you on the site longer. We (and other booru sites) try to avoid this because being able to find what you want without fuckin' about too much increases your chances of coming back later.

Only real flaw in the tagging system here is that aliasing is busted - but considering that no other site even comes close, we can live with that.

I think most other places don't care about efficient tagging because they believe "porn is porn" - so long as you put something approximately correct in front of the average user, they'll jack off to it, so why bother putting a huge amount of effort into tagging?

On irl porn sites it's far less niche that it gets on sites like this one. It isn't necessary for something to have thousands of tags about hyper specific things. General categories would make way more sense.

There's far less demand for anything hyper specific for most irl pornsites at best people are looking for stuff involving a specific fetish or pornstar.

I get what you're going for, but e621 is more robust, open-source, and more consistent in how its users tag due to the high degree of self-policing present within the furry fandom. Booru sites are in a very privileged position where the man-hours put in by users can substitute for a generic search engine based on probability models or neural networks ("artificial intelligence" or "big data" in modern business-speak). Boorus are the equivalent of users making their own Google or YouTube by manually describing each webpage or video to the server. It would be lightning-fast for end-users and require far fewer resources for hosts, but would be able to access a fraction of the number of pages and be at the mercy of the community's enthusiasm for any particular medium.

Now that is an interesting take, exactly the kind of new perspective I was looking for. Dare I extrapolate, but wouldn't this be the same model for TikTok in that it has a clearly strategic lack of search ability? So you just have to keep scrolling and scrolling through the feed or through a porn website's search results?

I don't want to fully jump on this bandwagon unless you can provide some more proof of your theory... even if only anecdotal at least.

Mind you, it is called Erotic Art, and I am a cultured aficionado.

I think you are boiling this down to be too black and white. There is a huge in between from two dozen tags on a conventional porn website and hundreds on sites like Rule34. No need to jump to that "hyper specific" level when there is still a vast amount of specificity that isn't covered by conventional porn tags.

Really the matter is best described by the fact that they often call them "categories" instead of tags, because that would be an insult to tagging. Lets say someone wanted to use a porn website tag for "Russian Language", would that be hyper specific? There are already many tags by nationality, but oddly none exist for spoken language. By your approach would even expanding tags to spoken language be "hyper specific"?

But I do think that we have common ground that a tag like "wet socks" would be hyper specific... is that actually a thing? who knows here? Same goes for something niche like "third nipple" which again, is that a thing?

Does this mean you imagine a future where AI could go through every video in PornHub and generate tags for it nearly instantly?... I can imagine it would be pretty funny to see when it messes up too.

Almost, but not quite. What I'm saying is that is this future is already here but limited to platforms run by money-rich companies as opposed to time-rich communities, and that YouTube et al. do not share their "tags" with end users because of the high degree of error that is likely with current AI models. I suspect this (and this is pure speculation; I would only know if I signed a NDA) due to the mostly automated nature of tag generation due to the sheer impracticality of human reviewers manually categorizing every piece of user-generated content that may exist, I seriously doubt that the "tags" produced by recommendation AI models are similar enough to Booru-style tags to be useful for manual human database queries, for which Booru sites are instances of.