site banner
PaperclipPerfector  6mo ago (text post)   27881 thread views

Culture War Roundup for the week of April 13, 2026

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.

3
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

Question: What is 2 + 2

Model: Hmm, that’s 2 and then another 2, so 22.

AUTOMATIC VERIFIER: WRONG

——

Model: Hmm, that’s the sum of 2 and 2, so 4

AUTOMATIC VERIFIER: CORRECT.

The model is tweaked slightly to make the second output more likely, and that output is potentially added to the training set. Repeat for arbitrarily complex mathematics and other problems as long as the solution can be verified, even if it isn’t known in advance. In this way you can generate potentially infinite amounts of data, albeit limited to certain domains. However, problem solving ability has so far extended quite well to other domains even when trained in this manner.

Question: What is 2 + 2

Model: Hmm, that’s 2 and then another 2, so 22.

AUTOMATIC VERIFIER: WRONG

——

Model: Hmm, that’s the sum of 2 and 2, so 4

AUTOMATIC VERIFIER: CORRECT.

The model is tweaked slightly to make the second output more likely, and that output is potentially added to the training set. Repeat for arbitrarily complex mathematics and other problems as long as the solution can be verified, even if it isn’t known in advance. In this way you can generate potentially infinite amounts of data, albeit limited to certain domains. However, problem solving ability has so far extended quite well to other domains even when trained in this manner.

Generally speaking, how does this "automatic verifier" work? Obviously I am not an expert but it seems like this automatic verifier would require human level intelligence.

In this toy case it's just literally a calculator (a snippet of python code). The problem is 2+2, the calculator just does 2+2 and checks if the answer is the same as the LLM output. (The LLM is trained to format the final answer in a particular manner and wrap it with special tokens, so the verifier doesn't have to be able to interpret natural language.)

You can get surprisingly far with this. If it's a calculus question, you can use an automatic differentiator to check it. Likewise for factorisation questions, metric conversion questions, algebraic manipulation of formulae, etc. you put a little work into programming the automatic verifier and you can get an infinite number of problems.

If you're a big company, you might have human domain experts doing some of this work too. If you're a smaller company you have a big LLM do verification for the smaller ones.

Then you have leetcode and programming problems, and again you can verify these automatically. Does the program compile? Is the program output what was requested? Is it faster than the previous solution?

Like I said, this only works for maths, programming, and other domains where you can verify the answer with a computer relatively cheaply, but contra the model of multiple intelligence factors, heavy training on maths and programming seems to improve general intelligence and reasoning quite well.

Like I said, this only works for maths, programming, and other domains where you can verify the answer with a computer relatively cheaply,

This is what the armies of Kenyans are for. I'm actually surprised progressive libs don't use "muh mechanical Turk sweatshop" as an anti AI talking point.

Also I think in some ways what they use the thumbs up/down for? I saw people saying the sycophantic behavior of the 4o era was people love being glazed and thumbs that up a LOT so it crept in.

In this toy case it's just literally a calculator (a snippet of python code). The problem is 2+2, the calculator just does 2+2 and checks if the answer is the same as the LLM output. (The LLM is trained to format the final answer in a particular manner and wrap it with special tokens, so the verifier doesn't have to be able to interpret natural language.)

You can get surprisingly far with this. If it's a calculus question, you can use an automatic differentiator to check it. Likewise for factorisation questions, metric conversion questions, algebraic manipulation of formulae, etc. you put a little work into programming the automatic verifier and you can get an infinite number of problems.

If you're a big company, you might have human domain experts doing some of this work too. If you're a smaller company you have a big LLM do verification for the smaller ones.

Then you have leetcode and programming problems, and again you can verify these automatically. Does the program compile? Is the program output what was requested? Is it faster than the previous solution?

Like I said, this only works for maths, programming, and other domains where you can verify the answer with a computer relatively cheaply, but contra the model of multiple intelligence factors, heavy training on maths and programming seems to improve general intelligence and reasoning quite well.

Thank you for the explanation. My instinct is that even with this type of training, LLMs will still be missing something essential, but I will give it some thought.

My instinct is that even with this type of training, LLMs will still be missing something essential

Your instinct is probably correct IMO. This form of synthetic data generation is just another tool in the box, it's not the key to everything.

I will say that we've got far further than I ever expected us to get using these methods. I'm instinctively a Gary Marcus-style fan of embodiment and unsupervised learning, it seemed clear to me pre-LLM that models wouldn't be able to be anything resembling intelligent without a body and the ability to interact with the real world and 'test' their understanding in real time. When LLMs came in, I felt I had to admit that I'd been wrong. It seems clear to me that we have managed to get to something I would call 'intelligence' (even if it's spiky and fails in some cases where humans would not fail) through these means. So I no longer trust my instincts as much.

This kind of semi-supervised exploration seems like a good compromise for now. I am also very interested in LLMs that can combine next-token video generation and text generation, because video generation requires understanding a bunch of stuff about the real world in order to produce consistent results, but that's a way off.

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats