Thursday, 5 October, 2023 in Portland, Oregon.
I had said that I was going to take a break today as an act of compassion. I have a habit of going hard for things and then feeling terrible if I fall off the balance beam. But again, I'm enjoying this writing and feeling a compulsion to write, so I'm just going with it.
We are about to hit a stupendously unseasonable stretch of weather: the highs of the next few days, from today through Sunday, range from 78-83f / 25.5c to 28.33c, which coming from London is practically an unbearable and rare summer's day. I mean, I just got an email from a health provider that "flu season is here" now.
I've had a great time with my open chats-over-a-pseudo-coffee the last few days and met super interesting people. You can grab a chat, too.
Two things that caught my attention today.
A while ago when the automated art generation engines wave had another crest of popular interest (this was, I think, around the time that automated style translation became in vogue through phone apps, run in the cloud, and the attendant warnings about "don't send 20 pictures of faces to someone you don't know, idiot, even if it's fun") part of the concern was about how stylized automated image generation from text prompts was going to destroy the market for commissioned art. Commissioned art is totally a thing that happens, although it's not something I have direct contact with. Its results seep through into my feed bubble/bubbles, sometimes a furry commission pierces through in what I would like to think is a semi-permeable bubble membrane, a sort of active barrier.
Anyway. It struck me at the time that there didn't appear to be a worker/cooperatively owned commission platform, sort of like "do an Etsy for Commissioned Art, but without the incipient enshittification". I am excited to see that there is such a platform! It's 1, and it's in closed beta now.
A crucial thing about this problem, I thought, was discovery: there are lots of artists who do commissions but again, with my complete lack of research which, alarmingly, now feels like David Brooks' m.o. of extrapolating an entire country's worth of significant trends from meeting two people, or making an ill-advised joke about a $78 airport meal. So. Centralization for branding, so customers know there's one place to go to commission art. To reference that meme again, though:
we've had centralization, yes, but what about co-operatively worker-owned centralization?
You'll note that this type of co-operatively, worker-owned centralization would be the kind of thing that the blockchain and distributed ledger cult members would have jumped on and gotten a bunch of attention for, and you would be right! They would claim transparency and automation and ownership all built in, and you don't need me to point out that you can get all of those things without having to resort to something quite so complex. Yes, it revolves around trust, but trust also is dependent upon the amount of time and effort one is able to devote in order to verify that trust, otherwise that trust is delegated to third parties or other methods, like "someone influential told me blockchain and distributed ledgers are totally legit", in which case, listen to better more trustworthy people (yes it's a bootstrap problem all the way down.) Transparency and trust are human decisions and as such subject to how humans work together, which might also mean drama and politics! (Actually, it will always mean politics: the personal is political because we're never, ever, dealing with non-human politics until we've got non-humans with which we communicate, and have dialogue and negotiation).
Caught my attention because: this is perhaps another data point you could stick in a story about another wave of co-operatively owned and managed platforms.
What's going on here? In the conversation thread of Katie Dreke's from the other day, about the cosy early 2010s internet, John V Willshire made this comment, that caught my attention:
I’ve wondered for years whether the big, centralised social networks were stabilisers (training wheels) for the social web. Just as AoL taught people about search boxes and hyperlinks, Facebook taught people about profiles, connections, privacy settings and so on. Now people can take those skills to the communities they actually want to be part of.2
Training wheels! Definitely caught my attention. Centralization as training wheels, because centralization, when it's successful, provides
Centralization is of course one of the first steps in this plan:
I'd argue then that in enshittification we finally have solved answer to the ???? in the Step one, Step two, ????, PROFIT!!! business plan.
Training wheels because they are a common experience. Sort of base-level awareness and competence, like Willshire points out.
Centralization might be a precondition for more viable and sustainable specialization: create a big enough market for a thing such that there is a big enough market for a smaller thing. As ever, though, you don't want shitty centralization. This is what regulators and legislation are supposed to recognize as bad, if a society prioritizes dealing with that badness, or even defining it as bad in the first place.
Ugh. I hate this, because the headline is written as a shitty NYT OpEd, but hear me out:
(Actually, this follows on from the above thoughts about centralization)
There is this thing called context collapse which was recognized by sociologists Alice Marwick and danah boyd. Quinn Norton has written about context collapse before, and goddamn hilariously3 one good article about context collapse and its consequences is Norton's article about how The New York Times fired her doppelganger4.
I'll quote Norton's brief description of context collapse from her article:
context collapse, where people create online culture meant for one in-group, but exposed to any number of out-groups without its original context by social-media platforms, where it can be recontextualized easily and accidentally4
Here's why Twitter Imploding Could Be Good, Actually: I noticed that I've spent a little bit more time on Bluesky lately, alongside Mastodon. I've also spent a little bit on LinkedIn recently, thanks to some radical acceptance that it's necessary for "professional development".
Twitter collapsed context into one giant feed of largely undifferentiated text; it also wasn't the only social network to do so. Like social networks before it, it also attempted to re-create context by using circles, which seems to be the default product name for groups of people.
Collapsing context can be good. I think it provides the opportunity to be exposed to more things. That's what I think is one aspect people are grasping towards with the public square: a cacophony of voices where there's the opportunity to hear something different.
But Twitter also, more than other large social networks explicitly collapses context by limiting you first to 140 characters. There's a plus side to that limitation of course in creativity, as well as the affordance in making it easier to, uh, share content into the gaping maw and engine of engagement. I don't necessarily think it's useful to think about how intentional that context collapse was, more than it was a side-effect of another goal, which I suspect was "engagement". Probably also due to the fact that anthropologists and sociologists aren't traditionally (ha) sought out in the design of social systems to consider potential harm because, honestly, what was the harm in blogging? This was just microblogging through and to your phone to your friends. Who knew what could happen? (This is also a false argument -- just because you had an intention at the beginning of something doesn't mean that intention and the choices are fixed for the next nearly-twenty-years).
But Norton is one of the clear examples where context collapse was not so good. Context collapse can be seen as a precursor or significant contributor to dogpiling, mobbing and abuse. It enables misquoting, and is one of the reasons why Mastodon's philosophy has been so against quote-tweeting or quoting in general: because of traumatic experiences in its constituent communities of having been mobbed, and the attendant fear of being mobbed and attacked intentionally by bad actors.
So why and how could Twitter Imploding be Good, Actually?
It could be good in the long run in that we're just not cut out for such large populations. We react quickly and emotionally, so it's easy to react to things that are taken out of context or twisted out of their intended meaning, that don't make sense in their original context.
Twitter's implosion is an example of perhaps a cycle of aggregation to disaggregation and splintering to (more mature?) smaller more specialized groups. Certainly something that's lost (at the moment) is the benefit of mass and reach.
That said, one of the reasons why mass and reach are harder with these networks is that some of them are intentionally not open. It's not like Mastodon or activitypub interoperate with each other, where you can reach both audiences easily. The need for profit and control lends to walled gardens and enshittification again: Twitter closes down its API and restricts bots, Reddit follows suit. Does anyone even remember when Instagram, Facebook and Twitter co-operated until it made more sense financially for them not to?
The implosion of Twitter was an opportunity for self-selection. For those who didn't want to be there anymore, like I've written about before, there was choice. While one of those networks came with a choice of more reach than the others (Mastodon) it had (has) a whole host of problems in terms of mass reach and usage (if it even chooses to set that as a goal), ranging from philosophy to the usual problems with open source projects and usability/user experience. Then came other attempts that were bootstrapped, like Bluesky, and other attempts that were less bootstrapped, like Threads. I make fun of post.news for it being the Aaron Sorkin Cosplay of text-based social networks, and I maintain that is true!
All of these places rapidly developed particular contexts. They have seams; everything in Twitter was significantly less seamless, or the seamlessness was easier to navigate. One particular seam that feels like it's always existed has been the difficulty of removing context from Tumblr posts: they're commonly encountered as screenshots, in part maybe because the whole point of Tumblr posts is that context is intrinsic to them; they're the sum of the conversation thread and, well, repost comments!
A point I've made before is that these contexts are software-defined, protocol-defined, and user experience-defined. Most importantly, I think is the user experience definition: they are different apps. They look different, they act differently (even if that difference has been minimized due to familiarity with "posting to a social network" -- again with the centralization training wheels from above!), they are different apps.
LinkedIn may try to be a social network. The content of posts to its feed have clearly gravitated toward from my point of view cringy influencer style posting, its own LinkedIn form and genre and poetry that's distinct enough to be recognizably parodied and the subject of satire. But it is a different place. You know when you're in LinkedIn. Bluesky is a different place. Tumblr is a different place.
Centralization meant amongst other things the ease of being in all places at once, or at least the easiest possibility of being in all places at once. Which inherently, I think, opens up the possibility for collapse.
Sure, there are things like hashtags that let you differentiate the melange of content (sigh), but those were and are everywhere, I think, post. Onboarding generally asks you for preferences or things you're interested in, at the very least because joining a social network where there's nothing or most things are things you're not interested in is a shitty unfulfilling experience. What, go to a meetup or a bar or whatever to meet people and there's nobody there? Or they're all doing something different? No thanks, I'll nope right out unless I'm the kind of person who derives energy from "making friends online" in an extraverted way.
It's not like I'm saying Twitter imploding was an unalloyed good. Like I said above, that's the needlessly provocative headline that isn't nuanced. Almost as if (or literally so) the context of this entire piece has been collapsed inaccurately, too!
It's more like: well, what did we gain? What's the downside of what we gained, and what might we attempt to recreate, and how?
Caught my attention because:
"The hype cycle for A.I. is different this time and won't collapse, because it's the tech hype cycle that has a non-zero chance of ascending to godhood" is a quote I made up in response to the position that the A.I. hype cycle will crash, just as it always has.
Little automations. Part of what slows me down in putting together this newsletter is the footnotes: grabbing the title and URL for a page and turning it into a Markdown link. It's something that I can do reasonable quickly thanks to my typing speed, but I finally finally got around to creating a Shortcut and assigning it to a keyboard, er, shortcut. Plus setting another shortcut to turn any URL into the latest archive.is link for that page.
... and that's it. How are you doing?
I'm doing better. I love the notes people are sending in, and like I've said before, even when they're just "hi!", and endeavor to reply to every one.
Things That Caught My Attention is a free newsletter, and if you like it and find it useful, please consider becoming a paid supporter, at pay-what-you want.
"I’ve wondered for years whether the big, centralised social networks were stabilisers (training wheels) for the social web...", John v. Willshire, 4 October, 2023, LinkedIn ↩
Quinn Norton: The New York Times Fired My Doppelgänger - The Atlantic, Quinn Norton, 27 February, 2018, The Atlantic (archive.is) ↩↩