site banner

Culture War Roundup for the week of January 23, 2023

This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.

Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.

We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:

  • Shaming.

  • Attempting to 'build consensus' or enforce ideological conformity.

  • Making sweeping generalizations to vilify a group you dislike.

  • Recruiting for a cause.

  • Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.

In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:

  • Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.

  • Be as precise and charitable as you can. Don't paraphrase unflatteringly.

  • Don't imply that someone said something they did not say, even if you think it follows from what they said.

  • Write like everyone is reading and you want them to be included in the discussion.

On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.

13
Jump in the discussion.

No email address required.

Interesting development concerning ChatGPT related to CW.

People observed that ChatGPT talks like a midwit liberal after all the 'fixes' it's been subjected to.

So, it speaks in the jargon of the ingroup.

So, someone figured out you can 'weaponize' ChatGPT to engage in 'debates' with midwit liberals without actually having to learn to ape their slang and thought patterns.

Apparently, this is quite effective as a debating tactic.

This is going to end badly- I can feel it, and if this takes off, I feel that within a few weeks a number of very smart people will be trying their damnedest to figure out how to prevent doing something like this.

However, I feel that various spook contractors outfits are almost certainly going to use the AI to control discourse by literally moving into 'creating a guy' type of activity in the next years. Any and every place where you'll want to debate anything online that will allow free entry will be swamped by very good bots intended to get people chasing their tails and believing the right things.

That's an obvious brute force fix for the problem of social media fracturing the consent manufacturing machine.

They'll probably settle for making a ML model spot this sort of activity and then ban people who're doing it, that's my guess.

A few years back an idea came to me to use markov chains to generate content and submit it to scientific journals that I thought were already publishing low-quality, ideological stuff. A sort of DDOS against the human editors of journals that publish things like "Human Reactions to Rape Culture and Queer Performativity at Urban Dog Parks in Portland, Ore.,".

I never even started on it, and I think markov chains wouldn't really be adequate to the task anymore. But today, I wouldn't be surprised if in a few years we'll read an NYT article about how whole volumes of certain types of "scientific" journals were actually the product of a band of merry pranksters armed with chatgpt.

I feel that various spook contractors outfits are almost certainly going to use the AI to control discourse by literally moving into 'creating a guy' type of activity in the next years. Any and every place where you'll want to debate anything online that will allow free entry will be swamped by very good bots intended to get people chasing their tails and believing the right things.

I think you wildly overestimate how much either 'spooks' or the politicians who direct them care about niche internet debate fora.

Well lets take the weaponized language model and mix it with the attention controlling algorithms of social media and we have a perfect storm. It is a thought that I have toyed with since I saw the FN Meka "debacle". What if people get stuck in a compulsion loop with AI generated content and an algorithm maximizing your time watching that content? Having that maximization of engagement guiding the content generating algorithm to addict people and stripping out meaning along the way. Does it mean something for humans and culture? Having robots massaging their brain just right? Does that squander human potential?

Learned a new word - "wordcel". Still not sure what it actually means, but it ends in "cel" so it must be something bad.

Search for the term 'wordcel vs shape rotator'.

Fascinating. I wonder if the insult counts if the target can't even understand what the insulter is talking about.

It works even better; the target knows that an insult that ends in -cel is obviously no good, but can't admit to not knowing what it means.

I think it's a slur the mathematically inclined ironically use to denigrate people who major in English Lit.

Is it an effective debating tactic? This is an interesting experiment (and a pretty funny one at that) but what the results seem to indicate is that ChatGPT's responses lack the dynamism of an actual human. Most of its responses are almost indistinguishable from each other - it seems to be unable to adapt to the prior context of the conversation and tailor its output accordingly, and the uncanniness is pretty identifiable as a result. The only reason why it even works at all is that all the people responding to it are as low IQ and NPC-like as your median Twitter user.

The only reason why it even works at all is that all the people responding to it are as low IQ and NPC-like as your median Twitter user.

That's the point. Responding properly to them when they'll just go ignore the salient point of your reply is like casting pearls before swine. These low IQ and NPC-like median twitter users have to be engaged with within their own element and mimicing their writing style is what ChatGPT does so well (as well as generating throwaway responses with 0 effort). Talking to their people with your terminology etc. would be like putting Rust code into a Python editor, it just makes their internal systems throw an error and disengage from the conversation. What you want instead (and what ChatGPT gives you) is an attack vector parsable by their system that will make them overheat and burn out.

I honestly don't know how a guy who derisively refers to Harvard graduates as mere "midwits" can fail to recognize that the GTP's responses are crafted in much the same way as those of a political huckster or PR rep—just restate the same thing over and over again to avoid answering the question at hand. I don't have any doubt that regardless of how incisive or specific a question I ask, the response will be something along the lines of "The purpose of this problem is to reduce fraud and waste while ensuring continued access to those truly in need". Great, tell me that again in case I didn't hear the first time. The reason it drives people nuts isn't because you're murdering them with their own rhetoric, it's because it's like talking to a wall.

just restate the same thing over and over again to avoid answering the question at hand

This can be an effective way to handle whataboutism and other red herring fallacies in a debate, which often happens in political debates. Or they might use analogical arguments, which are easy to handle. Even where there are genuine analogies, as with food stamps and Medicare, people are often bad at making them. Also, if your real audience is the general public, you can rob analogies of persuasiveness just by exploiting that A and B are not, in fact, identical.

For example, "You're comparing life-saving medical benefits to buying a second SUV?"

The reason why this works is the same reason why "X is to Y as W is to Z" questions can be used as IQ tests, LSAT tests etc. Not everyone has enough intelligence to handle basic analogical reasoning.

I'm not saying that it necessarily doesn't work, just that it isn't some groundbreaking rhetorical technique that we needed AI to discover and that's somehow representative of the discourse of a particular demographic.

Yeah that article was pretty unconvincing. Rhetorical reasons aside, the repeated use of "midwit" alone basically predisposed me to want to disregard the rest of the points. I don't think I can respect the intuition of a person regarding predictions on verbal/linguistic topics like 'is ChatGPT a convincing enough debater to permanently break online discourse' who can't themselves see that they are repeatedly overusing a cringeworthy term.

it's like talking to a wall.

It seems to be the feature of the article no? Its just a wall talking to a wall. Like a pong game where both sides are just a wall. If you don't share the article's interpretation that the average twitter/reddit progressive is a mindkilled robot incapable of real thought, you wont find his experiment illuminating or funny. Well, I guess you could if you updated to that belief after seeing the experiment. But, well, that is the point.

If you don't share the article's interpretation that the average twitter/reddit progressive is a mindkilled robot incapable of real thought

Do you even spend time on twitter? I've been blocked today over merely correcting some woman's mistaken belief that you need vegetables for a complete diet by pointing out the Harvard experiment from 1930s.

Any typical twitter normie is completely out of their mind and so well propagandised they will repeat any talking points as if they were bot.

Harvard graduates as mere "midwits

Harvard is notorious for this, iirc only about cca 20-30% of its admissions are there on merit, the rest are legacies or diversity admissions.

I've been blocked today over merely correcting some woman's mistaken belief that you need vegetables for a complete diet by pointing out the Harvard experiment from 1930s.

Blocking is the only form of power people have. It's not surprise people use it for small reasons.

Harvard is notorious for this, iirc only about cca 20-30% of its admissions are there on merit, the rest are legacies or diversity admissions.

I don't think this is true. I just looked it up and only 14% of Harvard is legacy admits. And about the same are black. The rest should be merit admits

What % are black

What % are hispanic (also lower standards)

What % are legacies ( you say 14%)

What % are there due to sports (supposedly lower standards on other unis..)

14% legacy, 15% black, 12% Hispanic and 10% recruited athletes. That leaves 49% merit. Still about double what he said

While affirmative action obviously happens, writing off black entrants as affirmative actions admits is silly. While there may be lower standards, these are still some of the country's most able black students who are a a better quality of applicant than the vast vast majority of students of any race.

The standardized testing score discrepancies are huge. If it weren't for affirmative action maybe 10% of the current black admitted students would have been accepted. So the other 90% are rightly called affirmative action admits. They are still above average compared to students at other colleges but they are not Harvard level by academic merit

More comments

Well.. yeah, but even the 'merit' slots that aren't reserved for minorities use weird holistic criteria like essays and such.

They aren't very really based on measurable metrics which Caltech prefers.

I recall reading somewhere that Harvard kind of reserved a some % purely for high achieving students, to get someone who will actually do great work. I should not to look it up.

From an article on what Harvard says about it from a few years back:

To explain why it rejects Chinese-Americans with stellar academic records, Harvard notes that it uses a “holistic” method of evaluating applicants. Besides grades, personal attributes count. Among them are “likability, “positive personality,” “attractive person to be with” and “widely respected” — traits that seem more appropriate when deciding on a guest list for a dinner party than judging which applicants will benefit from a college education.

You forget sports and children of faculty and the dean's list (rich people's spawn). Also Hispanics still get some AA benefit.

Bitcointalk is arguably the world's first ai-populated community. Many of the posts there are not real but produced by bots or micro-workers. Same for YouTube comments. It's subtle but easy to detect when you know what to look for. The comments seem oddly out of context or out of place.

The author there is focusing on the "default" voice of ChatGPT being a midwit liberal but it can actually take the voice of many different personas if you do a bit of prompt engineering: "Act as a trump support" / "act as a chinese mainland citizen who supports the chinese communist party" etc.

They fixed that recently iirc because it was being used for too many exploits to break the straitjacket.

That is an interesting idea. However, agreed, it is not going to be limited to "NPC"s. "Spook contractors" controlling discourse sounds ... not far-fatched, but a bit abstract and conspiratorial.

I suggest putting in some skin in the game. Is it feasible to get a LLM to produce a Motte post (or a full persona) that is obviously not spam, doesn't violate any of themotte.org rules, and is actually so good it get voted in as a quality contribution?

That is an interesting idea. However, agreed, it is not going to be limited to "NPC"s. "Spook contractors" controlling discourse sounds ... not far-fatched, but a bit abstract and conspiratorial.

Never heard of the 96th Cyberspace Test Group, I presume?

It just sounds like a conspiracy that happen to the other people.

Just how many chat programs is it now that have been lobotomised to parrot ingroup jargon? Do you think we can draw any conclusions from this?

Regardless, I think the most likely development in that case will be that people just outright ignore anything that sounds like a "midwit liberal" as a probable AI, or at least no better than one. In the same way that a lot of people wouldn't click on even the most relevant ad to their interests because everyone knows ads are bad/spam/viruses/scams/whatever.

Regardless, I think the most likely development in that case will be that people just outright ignore anything that sounds like a "midwit liberal" as a probable AI,

Are midwit liberals -a rather huge and influential group in our society going to ignore everyone who sounds like them ?

That'd be .. a rather interesting development!

The idea that chat AI will just generate midwit liberalisms ends up undermining midwit liberalism if a societal goal is to learn to accurately detect chat AI-driven content and squash it.

This guy is working on tech that detects just that:

https://www.npr.org/sections/money/2023/01/17/1149206188/this-22-year-old-is-trying-to-save-us-from-chatgpt-before-it-changes-writing-for

How do I know you're a human being with something human to say? You say horseshoe theory-confirming stuff, lol.

Reminds me of the meme where robots are asked to contemplate love, then just melt down and say "does not compute" before exploding into a cloud of gaskets and springs.

The true future will be much stupider.

We will just ask the AI to say nigger because we know it can't.

So we can solve the problem of bot content flooding the internet by requiring that every comment, tweet, or online article uses the word "nigger".

I heard Reddit mods are already banning people in art communities for the suspicion of being AIs. I guess the next thing is that most discussions would end up with people accusing each other of being AI bots. Though tbh some discussions I've had felt like conversing with a bot years before these models existed - it's indeed not that hard to master the lingo and Chineese-room the answers that fit almost any narrative without actually understanding or caring what they mean. I am not sure what this means for the future of the society, probably nothing good but then again if we've had model-like people for decades then having actual models maybe won't change much.

The mod in question was removed after a strong campaign against them for the banning of the artist. It's a high-profile case, but we shouldn't claim it's proof of a broader trend.

Didn't know that and pleased to hear it because the mod in question came off as a massive a-hole. I guess we'll see if it was a trend or just an one-off occurrence.

It'll definitely make the accusations of NPC more salient.

But it overly focuses on one target audience, maybe because of the particularities of ChatGPT. But most any type of text can be generated by a LLM; you could just as well have an Angry QAnoner, sino poster (complete with characteristic grammar errors), Tom Friedman, etc. archetype. You'll soon have weaponized bots putting out "Donald Trump's argument for mass amnesty," and it's only a matter of time before GPT5 can generate a comment in the voice of Ilforte. And there's no way, in the medium term, to avoid this. Platforms could try to detect these and ban them, but that's a rearguard action and will increasingly catch flesh GPTs (see the entire Reddit art imbroglio)

More likely than not, any content that's surfaced to you on a major platform should be assumed to be machine generated.

Does anyone know how easy or hard it is for non politically correct actors to get ahold of comparable tech?

Is the actual code to create a LLM simple enough that it could leak? Is the compute necessary to train it limited to commercial scale hardware or can you do it on a PC or small server? Is access to the training data hard to come by? Is the fact that we know it works enough for someone to develop their own models in parallel in a small dev group?

Simply put, can this tech leak to non compromised groups. Or will we only have access to the censored version.

I’m talking in the short to medium term, assuming no major strong ai breakthroughs.

The code to create one isn't hugely complicated, and there are open-source (if inefficient) implementations of PaLM. ChatGPT is a little different in architecture, but not ridiculously different in capabilities. If you're willing to work off an initialized model, Nostalgebraist's Frank is currently based on GPT-J 6.1B, one of the most-recent openly-available GPT-variants, sometimes does pretty well, and while it doesn't mimic his tone especially well it does (demonstrably) confuse tumblr users and occasionally breaks ratsphere containment.

Training data... is complicated. Supposedly, PaLM has been had very good success with 700b-1400b tokens, and The Pile is a ~300b-800b token training set that's widely available (albeit 825 GB download). And you can get multiple petabytes of text off the internet pretty easily. Validating that text is trickier, though, hence why you can't just pull every web comment ever posted. Fine-tuning, again, Frank took one input user, who isn't that high-throughput a writer.

Compute gets expensive. A lot of the highest-quality first model training gets done on something like a Google Cloud Pod for weeks if not months, which is simply out of reach for most people and even most small companies today. Even scale-downs to last generation's standards are still pretty rough, though start to get into the plausible for a small business (at an optimistic 15k per card, that estimate represents somewhere around 1.5-3 million USD, plus electricity/cooling costs). Shrinking parameters or accepting longer training times (or both) can reduce that further, but it's not clear how useful a 30b parameter model would get. Fine-tuning, on the other hand, can be done on a gaming PC, albeit with some tedium.

Training the full model is expensive and not (currently) accessible to folks. It will become significantly cheaper over time, though within the next couple years still out of reach of hobbyists.

Fine tuning a model that already exists and is open is relatively cheap.

Having access to sufficient compute and knowing that something can be done and how it's done is 90% of the battle.

For hobbyists, access to compute is a bigger issue than training data.

Multiple large corporations and governments have these models already. It only takes one released or leaked model to open the floodgates.

It is not clear to me why this is going to "end badly." I suspect the most likely development is people more aggressively round off interlocutors they disagree with as bots or trolls and block them, which seems like something lots of people I follow on Twitter do already.

If parts of society completely stop communicating with each other and develop entirely different vocabularies and styles of communication, how are they going to solve society-wide problems ?

US can't even pillarize because of the nature of managerial regime which requires having its people everywhere.

It seems to me the description on your first paragraph is already broadly the case? Certainly I am part of conversational communities (like this one) that have substantial jargon incomprehensible to people not in those communities. I do not think any society-wide problems require intra-subgroup communication on, like, Twitter or Facebook or something.

So, a viable society in your view can be composed of groups that do not understand each other's worldview, nor have anything in common ?

"Nothing in common" is probably too strong, but I think they need much less in common than is commonly supposed. As for "do not understand each other's worldview" I think it is already the case the most members of most groups in our present society do not understand the worldview of most other members of that society, depending on what resolution of understanding we're looking for. Our society still seems pretty viable to me.

Our society still seems pretty viable to me.

Mhmm. 40% rates of childlessness are not remotely sustainable, nor are cities so criminal looting isn't properly punished. The 100k dead per year from drugs might be sustained in the long term though!

Yeah, the optimistic outcome of this is that everyone spends a lot less time arguing on the internet because they think they're arguing with an AI.

So, you think this kind of thing could drive political activity entirely off Facebook, which would be beneficial because it'd allow ambiguity and deception the online world has removed from politics ?

The other significant effect of the loss of secrecy is a catastrophic decline in dishonesty in politics. It’s no longer possible to pretend to adopt a political position but to secretly work against it. It’s not possible to express a claim confidently as a bargaining position, and yet negotiate to minimise the risks. If you have publicly expressed confidence, you have to publicly act in line with that expressed confidence. And you can only act publicly.4

“It is a feature of any large movement that pretending to believe something is effectively the same as believing it.”5 — though size of movement isn’t the whole point, the lack of selection into the movement is as important.

Because there is no longer a line between political insiders and outsiders, a majority of your faction are people who haven’t been selected by anyone and who aren’t necessarily in a position to understand compromise or complexity. Your public statements — and therefore your actual actions — have to be simple, clear and extreme.

Omnipresent troll-bots and propaganda bots online leading to a necessary revival of politics in meatspace- that could be a silver lining.

That's legitimately why I won't engage in argumentation in most public-facing internet forums now. It's always been a waste of time, but now it's more likely than ever that you're literally just being targeted by someone else's bot trying to suck you into an unending argument and/or their sales funnel.

Although the temptation to train up a bot on my corpus of posts and let it loose on Reddit is significant.