but clearly such a scheme only works _if the tags can be relied on_

what's needed is _standardization_, coming up with a reliable system for concise description of all the subject matters that might conceivably turn up in a Mastodon post

that sounds scary but really it shouldn't be. library scientists have been at work on this matter for thousands of years now. coming up with a scheme for classifying posts by their subjects shouldn't be that hard

BUT...it requires thought and care and, most importantly of all, it requires the will to make decisions about infrastructure that are attuned to the common good

hardly any FOSS programmer, however, is good at making thoughtful, careful decisions about infrastructure that are attuned to the common good

unwilling to make hard decisions about design, they instead seek refuge in crowdsourcing and making everything infinitely configurable, hoping that maybe if enough people poke away that the amorphous slithering mass that they've created, it'll eventually acquire some sort of structure by some nebulous evolutionary process

deciding that _trends_, not a rigorous system, should decide how material is organized is a perfect example of this spinelessness

@kara I feel like this is something that can work (mostly!) for a library system, but really doesn't work for users tagging their own content.

Have you read Cory Doctorow's classic article Metacrap: https://people.well.com/user/doctorow/metacrap.htm

@gcupc why not? I would rather select tags from a standard list than trying to _invent my own, holy crap

@kara @gcupc
If tag frequencies were computed but ONLY for the sake of autosuggest in the tag completion field, in a distinct tag field, then I would be very happy with tags & tag frequency calculation.

@enkiv2 @gcupc ah, that's a good compromise.

@kara @enkiv2 @gcupc What you end up with then is write-only tagging leading to social cliques -- you can only look for tags you already know about, so you have to be part of a clique to know which tags to look for.

And you also end up with more tag-spamming as people add to their posts any tag someone might reasonably search for to find that type of post.

@byronalley @enkiv2 @gcupc oh, yes, I'm quite familiar with the cliques that closely guard the use of topic names. I have to fight them off every day when I use the library

@kara @enkiv2 @gcupc Mastodon is inherently public, so a lot of its value is in new people being able to join and find conversations.

Topic names are hard as the Metacrap article describes: is quantum computing physics, computer science, cryptography, math? Or its own topic?

Hashtags are harder and that's where it can get cliquey if there's no easy hashtag discovery, because then you have to be in the in-group to know which tags to search for.

@byronalley @enkiv2 @gcupc hence the need for standardization and conciseness

computer science isn't a real discipline so the answer to the first question is physics

@kara @enkiv2 @gcupc That's thoroughly unhelpful. What you're saying is that you don't recognize that categorization is a hard problem. Saying "I'm the infallible decider of categories" doesn't change the fact that you can't choose a single category for a topic, or tag for a post, that everyone agrees is the one true metadescription.

Libraries organizing books into single categories is a systemic flaw inherent in physical media, not a feature.

@byronalley @enkiv2 @gcupc no, in fact, I _do_ recognize categorization as a difficult problem--one that's been worked on by many many people much more knowledgeable than myself, people called "library scientists"

you're just making a bogus distinction excusing any policy applying to physical media as being magically inoperable or obsolete because the words being categorized aren't on paper

@kara @enkiv2 @gcupc I clearly didn't say any of that.

Sometimes limitations are actually features--eg. Birdsite's claim to fame was limiting how long posts could be, which apparently many people preferred. But when you see all the work that libraries had to do, to compensate for how hard it can be to find books on a topic, it's clear that it's not a beneficial limitation.

When Gmail moved from folders to tagging it was a big improvement.

@kara @enkiv2 @gcupc One alternative to tagging, for searchability, would be using machine learning to guess topics for messages.

The more ways there are to find something you're looking for, the easier it is to find. So even if *you* think computer science is a bogus discipline, and that quantum computing is clearly physics and nothing else, everyone who disagrees can also find it under math, crypto, comp sci, etc.

@kara
@byronalley @enkiv2
I'm not disputing that a trained librarian can correctly tag a toot (for humane values of correct that we agree to agree on). But can users? And if they can, will they?

@gcupc @kara @enkiv2 Tagging isn't really so much about content (a simple AI would do a better job than the average poster) but about conversation.

If you tag a toot with mastoart or Linux or Mexico, or with SomeRndCon2018, you're trying to be part of people searching and conversing around that topic.

It's an amplification that can be community-building, but also abusive if the platform can't limit its effects.

@gcupc @enkiv2 @byronalley please keep AI out of this

@byronalley @enkiv2 @gcupc only some will, which is better than none, and so long as the choices of tag are clearly delimited then there is only so much damage that can be done

otherwise you are making an excellent argument against enshrining trending tags

@kara
@byronalley @enkiv2
Let it be said that I completely oppose trending tags, though for different reasons.

@kara @enkiv2 @gcupc So you're actually arguing for a central authority to decide which tags people are allowed to use -- not by banning abusive tags (which would make sense) but by allowing only a preselected set of tags, eg. chosen by Gargron since he's the main one writing it all.

Hell, no.

It's not bad or lazy programming to leave a system open-ended to user self-organization; it's a feature that decentralizes power in important ways.

@enkiv2 @gcupc @byronalley then it's always gonna be a joke, sorry

without a clear organizing principle, or mode of governance beyond "what one programmer is willing to take responsibility for, which isn't much", Mastodon will always be a joke

@kara @enkiv2 @gcupc If your idea of a joke is user empowerment and an organizational system more effective than 19th C technology, then I hope that's what it remains.

Organizing principles -- yes we can do better. Make hashtag support much better.

Governance -- eventually Masto needs some kind of non-profit organization, absolutely. But in the beginning it's those willing to code who tend to decide what to code. So it needs to get bigger.

@byronalley "If your idea of a joke is user empowerment and an organizational system more effective than 19th C technology--"

please stick to playing with computer programming and leave organizational matters to others

@kara What's sad is that dichotomy. As it happens I studied organization from a variety of perspectives in university--organizational theory, sociological perspectives and computer science with software engineering. And it's both fascinating and unfortunate to see what each discipline, working on similar problems, could learn from the others.

It's never wise to entirely dismiss any discipline or area of knowledge.

@byronalley I agree. when computer science manages to achieve the stature of "area of knowledge", write me. right now it's about as much of a discipline as alchemy was (in Western Europe, anyway, the Muslims were a lot better at it)

@kara I'm not sure why you're hating on Comp Sci. It's a discipline that inherently changes as we learn more about things, but it's definitely a body of knowledge.

Have you studied any Comp Sci?

You could argue most distinctions between disciplines are arbitrary -- eg. Poli sci, econ, sociology & anthro. Or Math/physics/chemistry/engineering.

Comp sci studies interesting things with data & algorithms and info theory, with many applications.

@byronalley I have a CS degree from SDSU, in fact. got it the same time as I got my Classics degree (I'll show you the diploma sometime, I'm legitimately proud of half of it, anyway.) but the bulk of my education, and my masters degree, is in chemistry.

so far as I can tell C. S. is about one part stultified applied math, nine parts vocational schooling, and ninety parts running away from any notion that the subject ought to be regulated and enforced like a true engineering discipline

@kara I think that universities struggle with comp sci programs to handle the fact that a lot of people just want vocational schooling, and with the speed at which understanding of the field is changing.

And there's a movement to make software engineering a separate thing which is important.

But I also studied in an area studies program so comparatively comp sci seemed relatively well-defined.

@kara The problem is that users are, in general, lazy and won’t take the time to tag things properly. People already have some idea how #hashtags work, so they use those instead.

@USBloveDog fine! if only 10% of users use it, then that's better than no structure at all, and in time the community will learn (with the aid of gentle reminding) how to use the categorization system better

and think about what you're saying anyway

"users are lazy and won't tag things properly, so we should enshrine their lazy improper tagging in the form of trends for everyone to see"

@kara My other concern is whether, for example, alt.erotica.fanfic.harrypotter and alt.literature.harrypotter.fanfiction.erotic get treated as synonyms or as two distinct categories

@USBloveDog "writing"

like, we're talking about a few subject headings here, not minute subdivisions

@kara
The explanation I heard for why search only works on tags is that it's an attempt to make sea-lioning a little harder -- the idea being, if you use a tag, it's an invitation for discussion outside your network, while stuff that's untagged will only be visible on local/federated/home timelines, where somebody with an axe to grind has to put in much more effort.

@enkiv2 while I'm sympathetic to making systems harder to abuse, I think it's utterly foolish to make fundamental design decisions that hinder every user based on "but what if someone abuses the power?" paranoia