A really good response to Cory Doctorow’s recent (frankly disappointing) suggestion that those of us who reject LLMs do so due to “purity culture.”
6 Likes
I enjoyed this - I didn’t read Cory’s article yet, but I find myself cringing at some of the quotes of his. I have to admit I don’t know his work super well, I just read Enshittification and I thought it had good points - but this weird defensive take on AI isn’t appealing haha. I’m interested to read more from tante, thanks for sharing!
Also, I really enjoyed your blog post Generative AI is Built on the Exploitation of the Global South btw, it’s been making the rounds with my friends too!
1 Like
The misappropriation of the term “purity culture” in this context is so repugnant that I’ve been contemplating whether to write a dedicated blogpost on that alone. I gather that Cory Doctorow’s post has already been met with a lot of backlash, but I don’t know if anyone else has really addressed that yet.
2 Likes
Agreed. I’d be interested in reading that post!
I’m also a little stunned by the false equivalency he’s established between scrapers and search engine bots. Scraping data from the entire internet to train LLMs is not the same thing as crawling the entire internet to build a search index! Not even a little bit! That probably deserves a response of its own as well.
I don’t read his (Doctorow’s) blog posts all that often, but this seems a little out of character for him, based on work of his that I’ve previously read. I’m not entirely surprised, though… So many people who use LLMs daily have a tendency to get weirdly, excessively defensive about it and lash out with all sorts of logical fallacies when challenged directly, or when they encounter people who are generally critical of LLMs.
Glad you enjoyed my post. It sucks that I had to write it … and it also sucks to know that writers I otherwise admire might read it and erroneously dismiss it as “purity culture.” Ugh.
1 Like
Thanks. I’ll keep considering it.
If this man is relegating the task of proofreading to LLMs and also thinks LLMs are no meaningfully different from search engines then that raises the question of why he thinks search engines are for proofreading.
I was a little surprised as well, given previous statements, but it also doesn’t seem all that far a jump for the kind of person who’d participate in something as absurd as Free Our Feeds, the third-party fundraiser for AT Protocol/Bluesky that was predicated on the idea that market competition is a blanket solution to the problems with social platforms. It’s libertarianism all the way down.
2 Likes
I don’t think it’s a strawman at all, there are absolutely those (“some of” Doctorow’s readers) who prioritize association over cause-and-effect in their thinking.
(And if you’re a prolific poster with lots of followers, you’ll encounter more opinions considered by others to be strawmen.)
When a person is concerned about the possibility of a real position being mistaken for a strawman, I think a great solution is to link to what you’re talking about.
In this case I don’t think it would be appropriate for someone with a large following to link to or QRT some individual person replying on Mastodon or elsewhere.
its slightly besides the point, and everyone here has very good commentary, but im honestly really amused that Torment Nexus has its own wikipedia page. I didnt realize it grew that far out of niche internet complaining. Love that!
1 Like
Interesting article! I’ve written out my thoughts as I read the article. I do disagree in parts but I hope I don’t come across as overly critical. I liked the article and also agreed with a lot!
I’ve thought a lot about boycotts, often critically. Often this stems from the inherent neoliberalism of quite using changes in consumer behavior as a vehicle for political action. How all too often modern day boycotts just offer alternative things to buy or have a predetermined end date. I’ve seen people push back on boycotts because of how they tend to happen in modern times, including using the phrase “boycotts are a privilege”, and I think the start of this article lists several examples of how that can be the case. If we look at historically successful boycotts, like the Montgomery Bus Boycott, we can see they coordinated an alternative through mutual aid (a very large carpool) so the boycott could be sustained indefinitely. It went on for about a year before succeeding!
To be clear, there are certainly still some effective boycotts today; the BDS website has a whole list of wins. But those kind of coordinated, sustainable and indefinite boycotts are hard to come by, and often do not work to provide alternatives for people who cannot feasibly use any alternative for any number of reasons.
Relatedly, there’s been several short term “boycotts” (or “economic blackouts”) recently that I think have been ineffective both as boycotts and as political stunts. Financially, people _over_compensated by buying more before or after than they would have during the blackout! Having an end date already meant there was no leverage for any demands, so these blackouts have not had any lasting impact. At best they get a bit of media coverage, MSM or otherwise, but tbh with the cadence of today’s news cycle I don’t think any political stunts in recent history have been very effective. To compare to the Montgomery bus boycotts again, Rosa Parks’ arrest was a political stunt (a term I’m using positively here) that was coordinated by the group Parks was a member of, who also went on to capitalize on the media attention to start the boycott. Parks actually wasn’t even the first arrest they organized but the second.
All of that is to say, it feels like society has had a collective amnesia on how to organize effective consumer boycotts, and it bothers me a lot. Which is to say, I related strongly to tante starting the article with examples of when people may not be able to feasibly align their consumer behaviors with their ideals. Ultimately I preach that people should stick to what they know they can do indefinitely, and not beat themselves up over consuming anything that they couldn’t sustainably cut out or replace. After all, there is no ethical consumption under capitalism so we’ve all just got to try our best. ;)
Doctorow himself actually regularly points out the issue with relying on consumer boycotts for political action: if we’re voting with our wallets, those with the biggest wallets (the capitalists) always win. In fact, I think later in the article when he mentions “purity culture” as leaning into neoliberalism, he’s actually quite correct to point that out! Consumers hate AI, and that has been ineffective at stopping AI because the collective wallets of OpenAI, Nvidia, Microsoft, and so on simply make our wallets irrelevant. To suggest otherwise is to fall into the neoliberal belief that the free market, untouched by government regulation, will self regulate via consumer behaviors.
With that in mind, I am a bit wary of boycotts in general, look out for greenwashing, and don’t kid myself about any economic revolutions happening from a consumer boycott. In a way, it does all feel a bit performative. But if it makes you feel a bit better about your spending habits, and like you’re aligning your actions and values (prefiguration FTW!), then that’s still worth it! And of course, coordinated, targeted, and sustainable boycotts, like what BDS is doing, are actually effective! The rest of this paragraph was just about personal boycotts, like eating vegan or otherwise avoiding specific products or companies. And none of this should feel like justification to completely disregard the ethical consequences of your consumer behaviors.
Alright, enough yapping about boycotts in general though. As for LLM use, I myself am by no means a purist. I’m staunchly against AI, but I have used it myself. I’m really into self hosting, so I’ve even setup ollama and openwebui. I’ve used ChatGPT as well, and even briefly had the premium option back when it first came out. In my honest opinion, Doctorow using a (presumed small) local LLM to do spelling and grammar checking is about as milguetoast as it gets, and his defensive paragraphs do come off as fighting a non-existent strawman. I’ve been aware he’s not been fully anti-AI, as he regularly goes into a whole spiel about centaurs and reverse centaurs, and how the problem, in his mind, is business people mandating the use of AI and the increase of expectations to the degree the human becomes an accountability sink more than anything else. And honestly, I’ve been fine with that rhetoric because imo it’s a pretty valid and reasonable way to attack AI, even if it does concede on the point of valid uses for generative AI existing (which in sure there will be post bubble, but my list is likely shorter than Doctorow’s).
Doctorow’s 6th anniversary post suggesting the issue people have with LLMs is that the tech elite are bad people is very odd to me. Why use that to defend your use of a local LLM, built by the open source community and using open weight models? Tante’s list on all the other reasons people are critical of LLMs are also on point. But it does make me think some of this is just due to how pluralistic articles tend to go. His high volume of posts is at least partially due to the fact he talks about the same topic many times from slightly different angles. He isn’t trying to make each article portray an exhaustive list of arguments on said topic, but rather finding a specific through line he can use to connect the topic to one of the causes he’s currently championing: in this case he used the “it’s made by bad people” strawman specifically to make the counter argument that the tech proletariat should size the tech from the tech oligarchs. I didn’t think the reader is supposed to interpret the list of one argument an exhaustive list of the arguments people have against LLMs. Still shitty of him to attack those critical of LLMs as purity testers, of course.
Another minor quibble with tante’s article is the ship analogy. I agree that artifacts can, by their innate structure, be political and not “reclaimable” like Doctorow suggests. However the ship analogy doesn’t illustrate this point well because it relies on the incorrect conflation of power and logistical management (a conflation that’s existed so long it’s basically the premise of Engels’ “On Authority”!). A horizontal power structure can still have someone with the responsibility of making logistical decisions, with proper accountability to the polity, without giving them institutionalized power over the others.
I’m glad tante brought up the issue with open source and it’s lack of consideration for responsible use. I think it’s worth contrasting with the values behind mutual aid: both are about giving to ones community, but mutual aid includes ideas of codependence and communal obligations. The examples tante gave here, particularly of the women victimized by image generation tools, are very illustrative of the point that “liberating” technology does not liberate people. Later on tante also brings up why scraping for indexing should be treated differently from scraping for training data. I saw a comment arguing that tante was essentially taking JSTORs side against Aaron Swartz, which I think is a take you could only arrive at by seeing liberation as “I can do what I want without regard for my community”, like tante says doctorow is doing. I think this is really important: I’m not a fan of copyright, a clearly capitalistic invention that only hinders our collective technological progress. But I do believe in consent, and that making content publicly accessible does not give implicit permission for it to be used in training data, but making it indexable is implicitly allowed by making it publicly accessible in the first place. (Side note: that comment is amusing to me because it basically creates the argument that simply not using LLMs is performative and ineffective, which made me think the conclusion would be “so you should be destroying LLMs” but instead it was the same pro-AI arguments Doctorow was making).
Freedom is not a zero-sum game but a lot of the freedoms that wealthy people in the right (which I am one of) enjoy stem from other people’s lack thereof.
I don’t understand this sentence. Could anyone here help clarify it for me? Are they saying that they’re a right leaning individual, because that doesn’t line up with the rest of the text imo. If they replaced “right” with “global north” or similar it’d make sense to me.
In a way this framing shows more about Cory’s thinking that about that of the people he criticises: Cory is focused on markets and market dynamics and in that world it’s about purchasing.
This section is interesting to me, because at no point does tante use the word “boycott”, and these sentences seem to suggest the reason tante doesn’t discuss boycotts in this article is because tante doesn’t really care much about “markets and market dynamics” like Doctorow supposedly does. I think that’s why tante also just dismissed Doctorow’s legitimate criticism of using consumer behaviors as political action being based upon the lie of neoliberalism. Tante essentially dismissed it as mere identity politics, saying “This is narrativecly[sic] clever: Tell those stupid leftists that they are just neoliberals, the thing they hate! Awesome.” In a piece about “acting ethically in an imperfect world”, one where they specifically call for us to improve society through political action, I think completing ignoring criticisms about the effectiveness of consumer behaviors in enacting positive social change, as was referenced directly in the article tante is responding to, is quite the omission! I’ve brought the term up in my response here due to its relevancy to the discussion, not because I’m just such a fan of market dynamics. Trust me, I too want to be in a world where market dynamics are something I never have to think about. But alas, they are in fact useful to discuss to ensure our political action is effective.
Alright, we’re at the end of the article. I agree with the conclusion, and calling for us to have earnest discussions about how to restructure society, and criticizing Doctorow for essentially trying to disregard the opinions of those critical of LLMs.
1 Like
I started writing a reply to this but I found myself yapping enough to make a whole post about it, whoops.
3 Likes
Honestly? Doctorow’s always been a libertarian techbro-type person. I tried to like his writing because so many people praised it so highly, but…
So I’m not entirely surprised that he’d be pro-LLM or that he’d misunderstand ‘purity culture’.
1 Like
If this blogpost was about responding to some individual person replying on Mastodon then I think that renders the blogpost itself inappropriate. Substance aside, of course. For annoyances like that he can just take it to the private group chat like the rest of us.
You know, I’ve seen some folks characterize not-reading-bot-generated-text as a boycott, and while I’m sympathetic to that perspective, that’s probably not the word I would have reached for, myself. A boycott is a financial term for avoiding a market transaction. That’s not how I look at the choice to read or not read a blogpost. To me it’s more like choosing who to hang out with and whose conversations to join at a gathering – a social decision.
No. If he’s not talking about the specific subculture around Christian evangelical control over sex that crystalized in the 1990s, the emphasis on the pre/post-marital dividing line, and the sexual abuse it entails, then he’s not talking about purity culture.
I wondered about that as well, but I don’t personally know enough about ships to tangle with that one.
Here’s to that.
1 Like
I interpreted the article (tante’s) as also being against the use of LLMs, not just avoiding LLM-written text. I suppose you could argue Doctorow’s use, using free models locally, is also not a market action, but tbh I think the use of the term boycott is fine in that context.
Fair enough, I’m not well studied on the origins of the term and was mentally defining it based on how tante and Doctorow had been using it. I saw you mention earlier about considering writing about it’s misuse here and I’ll also look forward to it if it happens.
1 Like
Got around to reading this and absolutely loved it. I also read your post on the lack of AI “artist” role models, which reminded me a lot of the recent video essay by Adam Neely on suno, which touches on the same problem. I really like your writing 
1 Like
6 Likes
Thank you for this! I was vaguely aware of the term, but didn’t realize Libby Anne coined it. Gosh, been a long time since I read her work…
1 Like
Good write-up! I was extremely confused by the use of the term in the original blog post. I am thinking he was actually talking about (and getting the terms mixed up) political purity tests / litmus tests? But labelling this as being an “artifact of the neoliberal ideology” is still confusing and incorrect.
1 Like
I grew up in what many would call “purity culture” by its actual Christian definition–however–I do agree with the use of the term for certain sects (pun unintended) of the internet that spout rhetoric that’s strikingly similar in character. I’m not sure what it has to do with LLMs and AIs, but there are definitely some online who, while decidedly not Christian, still use Christian purity talking points, especially to shut down creatives they don’t like.
I agree with its extended use because it feels functionally the exact same to me: people obsessed with perceived sexual and moral purity who use it as a means of control over self expression.
3 Likes
I should clarify, Libby Anne didn’t coin it per se – as she acknowledges in the linked post, there were other uses prior to hers – but I think she deserves to be recognized for the role she played in helping it spread among bloggers who were exploring these same issues.
It doesn’t feel the same to me at all; these are distinctly different bodies of rhetoric and priorities. I have never in my life seen anyone styling themselves as an LLM critic argue that using an LLM is perfectly ethical and appropriate as long as you wait until marriage.
1 Like