STEAL

This is about kind of techie things, but the real center of this post is feelings. I can only speak for myself. I’m sharing them because I don’t think I’m particularly special, so I’m guessing bits of this will sound familiar to other people like me1. Some of this is still fairly inchoate and I’ll probably return to it somewhere.

When I was a kid on the Internet I felt intense indignation about copy protection. The way certain websites would override right-click, give you a pop up instead so you couldn’t easily save images. I’d insist on finding workarounds to save the things anyway, little pixel witch gifs, to folders from which they’d never be used. Who do you think you are to send bytes to my computer but then try to tell me what to do with them? A webpage felt like a document sent in the mail. Maybe you should be able to sue me if I sell some copy, some collage of it, but you can’t stop me from making the copy, the collage. DVD players that blocked screenshots were an indignity I couldn’t get around, and I fumed.

Now we live in a world where if you share a copy of an illustration hi-res enough to look good on people’s screens, it’ll be on an Aliexpress product as soon as it gets enough likes. Twitter bots looking for “I want this on a t-shirt” to swoop in and automate a dropshipped version.

I have to have a more complicated view about that feeble attempt to protect against image theft in that context. It’s why I never put the illustrations I was putting up on Instagram onto Pixelfed; sensible open APIs make it that much easier for some misguided individual2 to vacuum up everything you’ve put out there. I felt uncomfortable with that even as my discomfort seemed to contradict other of my values3. Nuance, maturing opinions…

And yet I have 25 lines of indignant JavaScript tucked into a Tampermonkey script because Library Ireland tried to stop me from copying a quote out from their site and How Dare They.

Writing about the web often takes as axiomatic that everyone wants attention online, and more attention is more better and we’re all supposed to Have A Platform and Use Our Platform. I have a social acquaintance with other people who have had their writing land on the front page of Hacker News, some of whom routinely have their writing there, and maybe it shouldn’t be surprising that having The Hacker News Community™️ tack a comment section you can’t moderate onto the thing you made isn’t something everyone always loves. This is a part of hypertext where the rubber hits the road, maybe. The freedom to link is the power to brigade.

The “immortal programming goddess” who checks for an incoming Hacker News referer header and redirects to a fart noise: heroic, iconic, we have no choice but to etc. etc.

The people who have less alloyed feelings about seeking attention are the ones trying to make a living selling something. I see people – artists selling prints and stickers, ceramicists, embroiderers – bemoan “the algorithm”, the platforms’ recommendation systems that mediate4 the presentation of their photo and video content. What is it that we are meant to contrast the presence of the all-powerful uncaring algorithm to? The past? Uncomfortable to note, but common to see: guides by and for these folks that advise doing little to none of the traditional marketing work that would have been necessary to their livelihoods before social media. Is the algorithm bad for “taking away” attention from them that they’d never have had in a pre-social media world?

No, this kind of exculpatory line of thought misses a lot of what’s really bad about the status quo by distracting you with a question of entitlement. Better to consider it together with gig work. These artists operate with supposed flexibility and autonomy, but then have to create reels based on rumored understandings of how Instagram will continue directing attention to their posts, their profiles, their sale announcements. These livelihoods entail having to pretend that your boss is not your boss, that you don’t have a boss, that you’re your own boss, and you post about being your own boss sometimes because #smallbusinesslife does okay numbers, but also what cadence do you need to be uploading on and do you have to use the app’s own camera and video editor for it to count? Someone says that they heard from someone who knows a Facebook employee that it works in such-and-such a way, but that was before the update six months ago that everyone remembers for tanking their engagement, their finances, so maybe it doesn’t work that way anymore.

So. If you’re not trying to maximize the attention you get and it comes for you anyway, that’s not all good. And if you’re trying to compete for attention with untold others on a platform where the ground is always shifting yet the house always wins, that’s not all good. Perhaps “more attention is better” is taken on faith if it holds true for the journalists who write about all this stuff, for the Substack commentariat class… but even for we the weirdo Posting 1%, I can’t say it feels true enough to generalize.

Artifact is a new app trying to be TikTok for text. I gave it a shot. I was curious as to whether, within the recommendation-system brief it gave itself, it would be more like Instagram reels – where I can sort of steer my way around, watch from searches – or TikTok – where any attempt to direct what I see other than careful Engagement Reactions seems useless. I could do a fuller write-up on it, but… more than anything, I was startled at the portion of the content that seemed to be documenting TikTok trends in blog posts.

You see people commenting on ChatGPT stuff with the worry that machine learning systems will begin to train on the generated output of machine learning systems. Maybe a good rule of thumb that we should have learned by now: never assume automating optimization is necessary for some weird thing to happen when there are already a bunch of people trying to optimize. The human TikTokkers creating and following trends and quota-hitting bloggers reifying them are enough to enact that imagined pattern5. Production turned into consumption turned into engagement, engagement analyzed, content promoted, encouraged, derived… the fact of the promotion and engagement becomes further material for derivation.

I don’t find music on TikTok or Instagram myself, but it’s common these days for songs to go viral there as background music. They say6 that music producers explicitly design around textures, interesting drops, moments that would be useful to be turned into video backing tracks. Some people talk about that like it’s horrifying that Art might be Degraded in this way. The idea that art forms are shaped by their social context is a little too foundational – foundational to, I dunno, four-fifths of what’s interesting about art – for me to take that seriously. Besides, I like texture7. But it’s a useful illustration of how limiting it is to approach this from an individual opt-in/opt-out mindset8: I don’t participate in getting my music from TikTok and Instagram, but the world of music that is available for me to listen to is different because we live in a TikTokified world. Because TikTok and Instagram have a recommendation model that uses choice of background music as a big factor in recommendations. Virality casts a long shadow.

I have never published anything under a FOSS license. Don’t get me wrong here – I’ve given money to open source projects, done some non-coding stuff, filed issues… I do believe in the thing, the heritage of the part of computers that came from acid-dropping academe rather than grindset. I can fairly say I don’t publish code because of my employer’s ridiculous policies9, but there’s something else as well. Automatic licensing, automatic permission, that’s what you need for industry scale, web scale, planet scale. But I like to get emails from other weirdos asking how things were done, to write back with cheerful comments. I like the inefficiency of it. I like how no one arrives in my inbox with Demands, only ever the beginning of a social interaction.

I have loved being linked to. The little bits of email correspondence. The conversations on Mastodon10. The acquaintanceships with other website proprietors. I have loved being listed in people’s handmade lists. Being aggregated and ranked is a different feeling. And there’s worse coming.

The search engines propose to summarize the contents of the web now (Bing, Google). This is somewhat like how they have been excerpting bits from pages for a long time, and contains both of the significant downsides of this existing practice: first, that it presents unvetted “information” with all the epistemic authority of the search engine’s brand, and second, through bypassing sending people to the original sources, it rips away from the original publishers whatever they had hoped to get from people visiting their pages. Ad revenue? Prestige? Clicks through to other material? All arrogated to the search engine.

These two effects would be bad enough to merit the commentariat’s angst. There’s a third and less concrete way it bothers me.

I have been fine with search engines linking to my posts, my pages. The web search feels virtuous in an era of Twitter jokes showing up on TikTok made into trends for blog posts aggregated into recommendation feeds…

The few things Google sends people to my site for seem reasonable enough. Google didn’t offer searchers anything about the cursed idea of dithering images with CSS before I did my write-up, which I know because I was searching extensively assuming someone must have done it already. That it is now an answer they give to similar queries: reasonable.

But I have never aspired to create neutral reference content. I only ever want to write things with enough voice that someone who’s on my frequency will enjoy it, and enough that someone who dislikes my style will wrinkle their nose and tab away. This sorting effect seems to me to be at the heart of what niche self-publishing ought to be able to offer; I’m not trying to maximize attention! I don’t mind when the search engines ignore me.

This site has been too much of an experiment for me to claim now that I have a grand theory about it, about similar sites – but certainly by now I have instincts, feelings…

…and making my writing available for automated summarization? So someone can sell ads by a depersonalized version of my stuff? The feeling is nausea.

I’m not a lawyer. One thing I understand to be true of copyright is that you are never meant to be able to use it to protect your ideas, but only your expression of your own ideas. Isn’t it exactly this expression that the large language models propose to be able to recreate, to sand off the things that make your version yours? It doesn’t have to be copyright infringement to be plagiarism – but then I suppose copyright never did much for anyone not shelling out for a lawyer.

I don’t derive the kind of gain from this site that justifies getting angry about “stolen” “traffic”. The professional artists and writers can point to the material effects on them, and that’s all real. People visiting my little blog posts more or less will make no material difference to my life. This line of reasoning – guesstimating impact – seems a bit hollow, though, as consolation goes. If there are other people out there who, like me, participate in the open web because we’re motivated by the small joys of connecting with similar weirdos, sharing with the younger weirdos coming up… well, feelings can cancel each other out. Does it feel good to be made gristle for a giant corporation’s sausage-making machine, even on your own independent website? Can I really tell people making a personal website is great with that as a caveat?

Where do we go when good web citizenship itself may be biting people?

Rutkowski has also added alt text in English when uploading his work online. These descriptions of the images are useful for people with visual impairments who use screen reader software, and they help search engines rank the images as well. This also makes them easy to scrape, and the AI model knows which images are relevant to prompts. 

It’s enough to make me consider running back to cozy group chats, removing anything scrutable from public availability. Sending out printed content with stamps.

I said that Google sends people to my stuff sometimes. That’s been true, but I’m hoping it’ll stop. I set up a robots.txt file, and I’ll switch it to an allowlist approach. I added meta tags in the head11.

I don’t think it’ll ensure anything with regards to the model training; the pattern they’re establishing for images is setting up the datasets under the auspices of some “non-profit” “research” institution and gosh isn’t it all the most fair use thing you can imagine? Never mind what hypercapitalist nightmare the funders have in mind for it. They don’t think they need artists’ consent, so they don’t really think they need mine. Levers exist for me to communicate my discontent, opt out of search, but only because I have a role worth small respect as my own server admin… so there’s some tradition to let me indicate how not to overload my system.

I know we all know by now that the ad industry is low-sulfur Satan. But it is also true that my mother is a librarian and her mother is a librarian and I always kind of looked at Google and thought, you know, maybe being a Professional Computer Person could end up being a little like being a librarian too. Family joked about high-school age me – “Mayapedia!” – for my enthusiastic insistence on looking up anything that was wondered about in conversation. Search has been a force for good in my life, over and above the good of open publication. To remove my own stuff, my own little pen recommendations and book reviews and tiny Chinese videogame handheld info… it’s sad.

You, reader, when you came here – likely from Mastodon – you reached out with your computer, your phone, to my server, where I had assembled my nonsense ready to send back to you. I sent you gold picture frames and a starfield GIF12 and a moon phase emoji with JavaScript to match it to the great moon phase emoji in the sky. And my writing – overwrought, terminally online, smeared with syntactic remnants of whatever book I last read. I put it together for you. The stitching is visible, crooked seams.

Will the future hold clean summaries, phrasings less convoluted than mine? A chatbot delivering the hallucinated view from nowhere? Will you be able to fine-tune a Maya model 20% less annoying than the real one?

There are men in California ready to sell ads on its sides.

  1. At least a bit technical. Both consumer and creator. A life deeply involved with technology. A sense that that’s not been only for the good. ↩

  2. This is me trying not to say “some asshole” so please read “some misguided individual” with the appropriate tone thanks. ↩

  3. “But Maya, no one’s going to bother stealing your stuff” is maybe true13, but it doesn’t make the dynamic more comfortable. Compare how any streamer considering showing her face has to reckon with not only the current availability but the future availability of deepfake tech. The expected-value calculation – risk, impact – might come out in your favor, but boy does it not feel good to have to do the math. ↩

  4. I like Corey Doctorow a lot more than I like Benedict Evans14, but on this, the EFF activist who I’d hope would be more plugged-in has failed to understand the dynamics as well as the venture capitalist, which is, like, kind of a bad sign for all of us. ↩

  5. I would be titling this pattern The Content Centipede except that the content involved doesn’t have to be shit. (It is often pretty crap, but not always! Creatives can get creative, even in cringey bad-meme-shaped boxes.) ↩

  6. This was in coverage I’d read, so I went and looked for it to link here. Duck Duck Go often lets me down, and I’m still trying to degoogle, so out of habit, I went to Bing15. And they chatbotted me! In my research for my own goddamn thing about it! Agh! ↩

  7. This is my favorite drop. You should listen to it in context, too. I have been striving in vain to find ones as good for almost a decade. If you, dear reader, know of a drop of this kind, reach out. ↩

  8. Those remembering my earlier variation on this theme may rest assured that I will stop talking about it exactly as soon as I stop tripping over dumb-ass well-no-one’s-making-you-do-anything takes on the topic. ↩

  9. Here’s something I always wonder about: if they’ve explicitly stated that even a code example short enough to fit in a tweet is impermissible, is website JavaScript verboten? Would it matter if I minified it? Should it matter that I have never written them a line of JavaScript, of CSS? ↩

  10. This isn’t me omitting Twitter to be catty. People didn’t talk to me much on Twitter about my stuff back when I used it and had more followers there than Mastodon. You probably can’t extrapolate anything interesting from that. ↩

  11. I live in a swamp! I put up signs! I’m a [technically competent admin]! What do I have to do to get a little [moral rights]? ↩

  12. Alternate post title: if you can’t handle me at my deranged aesthetic choices then you don’t deserve me at my fun facts. ↩

  13. Manually, it’s happened before. ↩

  14. Reasonable to guess my readers have heard of the former and not the latter. Benedict Evans: notorious thin-skinned VC who is, at least, often considering interesting questions in his writing. I’ve annotated a couple of his posts before. ↩

  15. I know DDG is backed by Bing. I have not always known this. Habit is habit. ↩