chatter tags
Tumblr chatters into tags! So does Twitter. So does everybody. I love it dearly. In Tumblr it's the sotto voce, in Twitter it's metonymy, in the content-y spaces it's a prayer to the algorithm to help connect you to an audience. It's a whole different channel in which the memeplex runs.
The nature of tagspeak is pretty tightly enmeshed with the medium: all a tag does is index stuff. Try to cast your mind bakc tot he pre-internet imagination – from inside that mindset, a tag looks like a tool for accountants.
And that is worth getting excited about! Categorization is a communication channel, it's a mode of performative speech. Tags are exciting becase they are a communication medium, with two audiences: human and machine.
To a machine, a tag is an instruction: index this here. There is always, presumably, a giant list of posts using a tag somewhere.
PageRank made it pretty cheap to do this with any word, though. The utility of tags waxes and wanes as the ration of data to metadata changes, but in general it continues to be prtty cheap to do "tag lists" with any sufficiently rare word in any sufficiently well-behaved dataset; if it weren't, we wouldn't have search.
xkcd 1306
Alt: "The cycle seems to be 'we need these symbols to clarify what types of things we're referring to!' followed by 'wait, it turns out words already do that.'"
But to humans, a tag is like a part of speech with illocutionary content. Sure, we begin by communicating, list me in this list. But that's such a simple, extensible idea, attached to such a brief, memorable morpheme. It wants to behave like a part of speech. Could it be a modifier? an interjection? as aside? We've seen them all.
Below the surface, dreaming, is the listening machine. We don't really need to factor it out of the text medium to make sense of it as a part of language, because we know how performative speech works already. Your grammatical token weds the algorithm in the crucible of the social medium, and the content of the tag is the magic incantation spoken over the cup. When you drink, something happens: and now you have to desrcibe, or ignore, that too. Is it a room you've walked into? A mark placed on your hyperreal body? A graft splintered into the bones of some other hyperobject, its seams only visible to your myopic locus for just the time that the words hang in the air? Are you in a club? Are you part of a movement? Are you awoken?
I stress, language has always done all these things. Algorithmic language, in its native idiom, in the promise of its bloom, is a cybernetic enhancement. And the curse of a tag is that it's expressive in the algorithmic medium.
Our social dreaming doesn't let language lie still, however. This is why the meaning of a tag can very easily float free of its algorithmic underpinnig, given half a chance. Like an old-school telegram, making something a tag connotes that it's a higher-value signal – you're taking the trouble to transmit along a more constrained channel, invest greater compute resources, perhaps expecting momentous downstream consequences. The price of transmission is its own market signal – way more expensive whenever anything important happens, over time drifting in the winds of technological development, often caught and whirled about by the eddy currents of the social biome's reactivity.
What humans do with uncertainty is gamble. Really gambling is a superset of creativity, which really is just a hot-take way to say that what humans like to do in uncertain situations with limited levers is try variations over and over again. Give somebody a high-risk high-reward mouthpiece into which they can scream, and they will use it to try saying stuff to people.
The memeplex is alive, is what I'm saying. Nobody was ever going to let a tag mean just one thing for any length of time.
I'm all for it. Death to the prescriptivistss. To approach the boundaries of the possible, wade in among the collective soup of gamblers that are trying to find it.
So what have they found? What do people use tags for? I only know superfically. It's time to go read Algospeak, probably; I want to understand how to actually do good internet anthropology.
Whatever it is, it's dying. Adam Aleksic has more.
But let's take a moment for the accountants and the beleaguered software professionals trying to keep the lights on and the heinous mouthpiece reliably whispering its Megamyriad Words without cease into your conch radios. How the hell do you route this stuff? How do you make use of it? How do you support the misspellings, the deliberate misspellings, the eupehemism treadmills, the sheer burgeoning natural language of it all? Is it any fucking use, after all is said and done, to waste time and compute on the festering teeming mass of human speech?
Ellen Ullman, in Life in Code, discusses how software developers have fallen into a cultural tendency of contempt for the user. From the viewpoint of a developer designing a control system, the user is only really outside the system by a fiat boundary, one reinforced by social mores, but with the temptation to absorb them into the machine always looming. Once you let this little private act of phagocytosis proceed, a user is merely the least predictable thing in your control system, less like an independent actor and more like an agent.
The resulting phenomenology delights me.
We de-anthropomoprhize, and then re-anthropomorphize. The user is flaky, an unreliable part. The user must be cajoled, seduced; or coerced. The user may be occasionally need to be slapped like an uncooperative hay baler. The user is cattle, not a pet. The user-as-audience can be bought and sold. The user-as-buyer can be promised and delivered. The user must provide reliable value, which means that the user must be predictable, which means that the user must be as powerless as we can make it while still keping it here. All the benefit and profit of extending human capacity through the power of a social medium must be dampened and mediated before offering it to a user, because otherwise it's a way for the user to rootkit the whole thing. To a platform, the user's will is like the will of a cancer cell.
Tags, I think, are a little too powerful to ever have been a comfortable offering from software companies used to feeling contempt for their users. A parametrizable megaphone, a way to speak in concert; virtual rooms that overlap; indexed memories with memorable handles. Tags are expressive; tags are true cybernetics; tags meaningfully extend the language medium. Tags they were never used as intended, because their possibilities are generative. Tags will be taken away, not because they aren't useful, not even because they aren't useful to a platform, but because they aren't useful from the perspective of a platform.
Related (some stubs):
- systematic descriptivism is just science
- high-churn metadata
- a phenomenology of social media
- Life In Code review