Replying toPost-AGI Economics As If Nothing Ever Happens

Post-AGI Economics As If Nothing Ever Happens

but adults are prohibited to try to change child’s sexual preferences for their advantage.

Is this true? I've never heard of any laws about not trying to change a child's sexual preferences. Or really any laws against trying to change people's preferences for your benefit.

I suppose statutory rape laws could be construed a something like this, but it really seems a stretch.

Replying toStrong, bipartisan leadership for resistance to Trump.

Trinley Goldenberg1mo

Strong, bipartisan leadership for resistance to Trump.

I don't think that understanding people better than they understand themselves is empathy. That's cognitive modeling.

Empathy is being able to step into a worldview such that you can see what they see and feel what they feel, this requires some cognitive modeling but mostly seems to use different machinery.

Replying toChildhood and Education #16: Letting Kids Be Kids

Trinley Goldenberg2mo

Childhood and Education #16: Letting Kids Be Kids

Why are kids on their phones and tablets all the time? How could we stop this?
Easy, you let them have unstructured playtime, that’s it, that’s how you do it.

When I stayed with my cousin for the holidays, I noticed that my nephew used Minecraft as an unstructured play environment. He would basically just run around with his friends and do dumb shit and make up games etc.

Obviously this is worse in a lot of ways to in person unstructured play as you can't run around together and move and touch, but there's a lot of benefits compared to the logistics of having to get everyone together as well

I imagine that some of this replacement of offline unstructured play would be happening regardless of whether it was made harder to meet in person in the ways mentioned here, but that it being harder is driving kids towards online unstructured play even more.

Replying toCourtship Confusions Post-Slutcon

Trinley Goldenberg3mo

Courtship Confusions Post-Slutcon

Fwiw while wit matters in flirting, I've found it's much more about energy.

I don't know if this abstract description will make sense to you, but ime the way that flirting leads to sex is by continually ramping up sexual tension by a sort of plausible deniability of sexual interest -

one way to do this is you're talking about sexy things but not being overtly sexual with your energy, or having a sexual energy but it's more teasing or talking about non sexy things

In any case, if both people are into it, and don't let the tension escape through other outlets, the way to release it is sexual escalation. And as long as the escalation (E. G. Cuddling, kissing etc) is still less than the sexual energy calls for, the tension continues to build to sexual intimacy.

Replying toOut-paternalizing the government (getting oxygen for my baby)

Trinley Goldenberg3mo

Out-paternalizing the government (getting oxygen for my baby)

Note that getting an at home nebulizer is also quite easy, saved me many a trip to hospital.

Trinley Goldenberg3mo

This makes me think of eurisko/automated mathematicion, and wonder what minimal set of heuristic and concepts you can start with to get to higher math.

Replying toThe skills and physics of high-performance driving, Pt. 1

Trinley Goldenberg3mo

The skills and physics of high-performance driving, Pt. 1

I've heard that motor racing is a physically demanding sport, and the physical conditioning required is similar to that of other pro athletes. Any truth to that?

Replying toAI safety undervalues founders

Trinley Goldenberg3mo

AI safety undervalues founders

Aren't the central example of founders in AI Safety the people who founded Anthropic, OpenAI and arguably Deepmind?

This is consistent with founders being undervalued in AI safety relative to AI capabilities. My model of Elon for instance says that a big reason towards pivoting hard towards capabilities was that all the capabilities founders were receiving more status than the safety founders.

-2

Trinley Goldenberg3mo

It doesn't seem like either thing captures the gears of what's happening to an llm, but it's insofar as you're arguing it's impossible for an llm to be convinced, I think you should provide better gears or what specific aspect of "convincing" you don't think applies

I dint think that arguing with an llm will reliably enter into a sychophant mode, or that the thing that happens when you go through a series of back and forth argumentation with an llm can be fully explained by it entering into a mask/face

•••

Replying toRejecting "Goodness" Does Not Mean Hammering The Defect Button

Trinley Goldenberg3mo

Rejecting "Goodness" Does Not Mean Hammering The Defect Button

But more importantly it’s a whole memetic egregore, and you can’t really get rid of the bad parts without rejecting the whole egregore. And you can get the good parts without most of the egregore

Can you say more to about why you believe this second part? It's not at all straightforward to me how you pick and choose the good parts of an egregore to "get" them.

The dominant egregores are selected, so if you want to beat them, you need to out-compete them, which means you can't just select on "goodness" as a criteria, you have to have a cohesive package that outcompetes the existing egregore.

-2

riffing off of this tweet:

i remember a conversation with a guy who was like 'i think christians are better than buddhists because they tend to be married and have happy children' 'oh, so you think christianity is better because it better upholds christian values?' this is a common issue...

whatever criteria you evaluate a religion on, it reveals your default religion

if you evaluate on beliefs and their correspondence with 'reality', your religion is scientific materialism

if you evaluate on how happy it makes people, your religion is individualistic humanism

there's nothing wrong with simply finding the practices that best instantiate your current religion, but i think there's another way

evaluating religion on whether it can show you things... (read more)

if i had to pinpoint what I think is the biggest issue with the rationality and ea community, i'd say that ea thinks wisdom is compassion, and rationality thinks wisdom is intelligence

but in fact, wisdom is a secret third thing which should steer intelligence and compassion

Personal responsibility and systemic failure are different levels of abstraction.

If you're within the system and doing horrible things while saying, "🤷 It's just my incentives, bro," you're essentially allowing the egregore to control you, letting it shove its hand up your ass and pilot you like a puppet.

At the same time, if you ignore systemic problems, you're giving the egregore power by pretending it doesn't exist—even though it’s puppeting everyone. By doing so, you're failing to claim your own power, which lies in recognizing your ability to work towards systemic change.

Both truths coexist:

There are those perpetuating evil by surrendering their personal responsibility to an evil egregore.
There are those perpetuating evil by letting the egregore run rampant and denying its existence.

The solution requires addressing both levels of abstraction.

A lot of people are looking at the implications of o1's training process as a future scaling paradigm, but it seems to me that this implementation of applying inference time compute to just in time fine tune the model for hard questions is equally promising and may have equally impressive results if it scales with compute, and has equal potential in terms of low hanging fruit to be picked to improve it.

Don't sleep on test time training as a potential future scaling paradigm.

Enjoyed this video by Veritasium with data showing how Politics is the Mind Killer

Want to help me out?

Vote on the book cover for my new book!

It'll be up for a couple of days. The contest website only gives me a few days before I have to pick finalists.

https://form.jotform.com/243066790448060

I desperately want people to stop using "I asked Claude or ChatGPT" as a stand-in for "I got an objective third party to review"

LLMs are not objective. They are trained on the internet which has specific sets of cultural, religious, ideological biases, and then further trained via RL to be biased in a way that a specific for-profit entity wanted them to be.

•••

It seems like the obvious thing to do with a model like o1 trained on reasoning through problems would be to train it to write code that helps it solve reasoning problems.

Perhaps the idea was to not give it this crutch so it could learn those reasoning skills without the help of code.

But it seems like from the examples that while its great at high level reasoning and figuring out where it went wrong, it still struggles with basic things like counting, which, if it had the instinct to write code in those areas which it's likely to get tripped up, would be easily solved.

In the early 2000s, we all thought the next productivity system would save us. If we could just follow Tim Ferriss's system and achieve a four-hour workweek, or adopt David Allen's "Getting Things Done" (GTD) methodology, everything would be better. We believed the grind would end.

In retrospect, this was our generation's first attempt at addressing the growing sacredness deficit disorder that was, and still is, ravaging our souls. It was a good distraction for a time—a psyop that convinced us that with the perfect productivity system, we could design the perfect lifestyle and achieve perfection.

However, the edges started to fray when put into action. Location-independent digital nomads turned out to be just... (read more)

Good ways to monetarily profit from the increasing demand for power?

Trinley Goldenberg

There have been several good posts about how to profit from a world of AI takeoff.

However, when it comes to individual investment recommendations, most of them are for direct AI companies or builders of GPUs and compute infrastrucutre

Recently, it's become clear to me that power will be much more of a bottleneck than GPUs (and therefore even more valuable).

What's the best way to profit from this insight? Stock picks? Futures contracts? Something else?

Zuck and Musk point to energy as a quickly approaching deep learning bottleneck over and above compute.

This to me seems like it could slow takeoff substantially and effectively create a wall for a long time.

Best arguments against this?

Mechanism for feature learning in neural networks and backpropagation-free machine learning models

Trinley Goldenberg

Understanding how neural networks learn features, or relevant patterns in data, for prediction is necessary for their reliable use in technological and scientific applications. In this work, we presented a unifying mathematical mechanism, known as Average Gradient Outer Product (AGOP), that characterized feature learning in neural networks. We provided empirical evidence that AGOP captured features learned by various neural network architectures, including transformer-based language models, convolutional networks, multi-layer perceptrons, and recurrent neural networks. Moreover, we demonstrated that AGOP, which is backpropagation-free, enabled feature learning in machine learning models, such as kernel machines, that apriori could not identify task-specific features. Overall, we established a fundamental mechanism that captured feature learning in neural networks and enabled feature learning in general machine learning models.

Peter Thiel on Technological Stagnation and Out of Touch Rationalists

Trinley Goldenberg

Sharing mostly because I personally didn't realize how Thiel was viewing the Bay Area rationalists these days.

He explicitly calls out Eliezer's "Death with Dignity" post as ridiculous, calls out Bostrom as out of touch, and says that the rationalists are only interesting because they aren't saying anything, and are just an echo of prevailing feelings about technology.

I think it's worthwhile really trying to step into Thiels point of view and update on it.

Non-Coercive Perfectionism

Trinley Goldenberg

Perfectionism is a surprisingly common ailment among both procrastinators who don't get stuff done, and high performers who get a LOT done. With just a few tweaks you can go from the former to the latter. While perfectionism is quite complicated, a useful simplification is to break it down into two categories:

1. Perfectionism for others - done to gain praise, avoid punishment, or be accepted.

2. Perfectionism for self - done to satisfy taste, identity, or goals.

Quite often, both types are present in a person, but one is a shadow value. People will hide that they care what others think, and pretend it's only about their work. Others will hide that it's a... (read 785 more words →)

Would most people benefit from being less coercive to themselves?

Trinley Goldenberg

Non-coercive motivation is a motivation system that doesn't rely on forcing yourself, pressuring yourself, shaming or guilting (I explain more of what it means here).

Would most people benefit from being less coercive to themselves?

I sort of want to start answering straw versions of the question, because I think that they do answer real fears that people have when they hear about non-coercion and have the immediate reaction "You want me to just drop everything and do what I feel like!?"

Should you drop everything and focus on non-coercion?

So the first answer to the question as it is posed is "Of course not!" Goals are highly contextual. If you're the president of the United... (read 1181 more words →)

Why Productivity Systems Don't Stick

Trinley Goldenberg

Note: Experimental, trying to repost a Twitter thread as a Lesswrong post to see if people like it (so if you like it, or don't, tell me).

Let's talk a little bit about oscillating motivation, shadow values, non-coercion, and the problem of revolving productivity systems.

We've all had the feeling that new productivity system is working! Only to find that days, weeks, or months later it falls off, and we fall back into a period of being unproductive. Hell, I was stuck inside this revolving door for more than a decade.

I tried every planner, app, book, coach, journal, todo list, calendar, and retreat under the sun. But always, would find myself back in the... (read 825 more words →)

How to Write Like Kaj Sotala

Trinley Goldenberg

Since I started quarantine in late February, I've been trying to learn the skill of modelling implicit skills and knowledge. Over that time, I've interviewed nearly 20 people and learned one specific implict theory or skill that they use. Many of those people in the rationality and post-rationality communities.

Kaj Sotala is one of my favorite writers on Lesswrong. He writes clearly, deliberately, and precisely. His writing is deliberate, interesting, and approachable.

After nearly 4 hours of interviewing him, this is my model of how he writes. Note that while it is inspired by Kaj, it is not necessarily fully endorsed by him. There is always a translation process from his brain to his... (read 1352 more words →)

When Gears Go Wrong

Trinley Goldenberg

(This talk was given at the LessWrong curated talks event on July 19. mr-hire is responsible for the talk, David Lambert edited the transcript.

If you're a curated author and interested in giving a 5-min talk, which will then be transcribed and edited, sign up here.)

mr-hire: A question that I don’t think gets asked enough in the rationalist community is whether gears are a good investment. Typically, when it is asked, the answer is yes. Here's an example from John Wentworth talking about gears as a capital investment.

I'm not picking on John, and I mostly agree with him, but I think this is a question that should receive a little bit more scrutiny and... (read 1668 more words →)

Reconsolidation Through Questioning

Trinley Goldenberg

At this level, you're actively asking yourself questions about the correctness of the schema. You're not looking for any particular answers to these questions, or trying to get any result, you're simply holding the questions and seeing what comes up.

When coaching and teaching workshops, I find that questioning techniques are the most consistently successful at creating memory reconsolidation. They seem to strike the optimal balance of challenge and non-judgement.

They seem to work by actively directing our attention towards areas where the schemas may not match up with reality, without provoking any resistance by actively suggesting the schemas are wrong. For this reason, it's quite important that you don't actively... (read 369 more words →)

Reconsolidation Through Experience

Trinley Goldenberg

At this level, you're simply sitting with the schema. Oftentimes, this alone can produce a dramatic shift through memory reconsolidation simply as a matter of re-evaluating the old schema when you get a look at it. I also include non-memory reconsolidation techniques in this category that work through extinction of the old association.

Experiencing Evidence

Memory Collection

Memory collection involves finding and allowing yourself to experience every time you can remember when you felt or acted through a particular belief schema. For each piece of evidence, you allow yourself to think and feel through the schema in that particular memory, then move on to the next one. By simply allowing ourselves to see all the... (read 396 more words →)

LESSWRONG
LW

LESSWRONG
LW

Trinley Goldenberg

The 3 Books Technique for Learning a New Skilll

S-Curves for Trend Forecasting

How to Write Like Kaj Sotala

Why Productivity Systems Don't Stick

Trinley Goldenberg

Good ways to monetarily profit from the increasing demand for power?

Mechanism for feature learning in neural networks and backpropagation-free machine learning models

Peter Thiel on Technological Stagnation and Out of Touch Rationalists

Non-Coercive Perfectionism

Would most people benefit from being less coercive to themselves?

Why Productivity Systems Don't Stick

How to Write Like Kaj Sotala

Non-Coercive Motivation

Changing your Mind With Memory Reconsolidation

Trinley Goldenberg

The 3 Books Technique for Learning a New Skilll

S-Curves for Trend Forecasting

How to Write Like Kaj Sotala

Why Productivity Systems Don't Stick

Trinley Goldenberg

Good ways to monetarily profit from the increasing demand for power?

Mechanism for feature learning in neural networks and backpropagation-free machine learning models

Peter Thiel on Technological Stagnation and Out of Touch Rationalists

Non-Coercive Perfectionism

Would most people benefit from being less coercive to themselves?

Why Productivity Systems Don't Stick

How to Write Like Kaj Sotala

Non-Coercive Motivation

Changing your Mind With Memory Reconsolidation

Should you drop everything and focus on non-coercion?

Experiencing Evidence

Memory Collection