Charlie_Guthmann's Quick takes

Is there currently an effective altruism merch/apparel store? If not do people think there is demand? I'd be happy to run it or help someone set it up. (quick search shows previous attempts that are now closed - if anyone knows why that would be cool too)

I'm biking and Amtraking to Berkeley to join the plzdontkillme house july 1st. I'm gonna try to interview people along the way about ai/technology/practical philosophy. 

You can follow me on insta https://www.instagram.com/charlie.guthmann/ or youtube  https://www.youtube.com/channel/UCmTkQjHjs2cVgca3eC5vHcQ  
Bear with me as I'm learning how to use social media and do content creation.  

Here is the video I made a month ago (with the help of Diego, who is the linked channel), if you want to get a sense of what I'm trying to do. Advice welcome! 

OpenAI is claiming a legitimate breakthrough on a "prominent open question". I don't know exactly how to phrase this, but I'm well past the point where i'm interested in arguing whether or not we are on the precipice of some insane stuff. Yes there are still reasonable disputes about exact timelines, foom, economic impacts, diffusion curves, etc. I'm not saying the world is gonna end in 2029. But if you are legitimately arguing anything approximating stochastic parrots, or that this is just a hype train, I'm going to immediately assume you have motivated reasoning or need to think harder. 

Just was watching Dwarkesh/David Reich podcast, fascinating stuff. Looking back at how I was taught taxonomy and anthropological history I find it frustrating. Note that I don't know much about (evolutionary) biology or genetics or the frontier of what genetic-history research so this is my layman attempt to explain why it's generally been puzzling for me how i have had this explained by other people who probably don't understand either, not trying to propose that I understand something david reich doesn't. 

My main gripe is that we are taught evolutionary history mostly from the lens of evolutionary trees. But evolutionary history probably looks like a graph/stochastic process/markov chain, and only at very specific underlying parameters/ level of abstraction is well modeled by a tree. The reason we use trees is because that is the most sensible simple abstraction in some ways, if you are thinking about, "how did we get here?". But it's not a great way to think about "what happened/was happening".  I had chatgpt try to make the difference below (don't look too into the details, it did some hallucinating, just the general vibe). 

Taking plausible parameters here to me would be thinking mixture isn't extremely likely over short time spans because of distance/etc but quite likely and almost certain over hundreds/thousands of years. So what did the near east look like genetically 60k years ago? It could easily look like below. 
 

It seems totally possible that for long periods of hominid history genetics were well modeled by pretty smooth stochastic graphs with generally corresponding smooth genetics across geography (obviously with tons of exceptions or less true when you zoom in, e.g. bell beaker/corded culture), and yet when you look at our specific lineage it doesn't quite look like that (due to extinction, gene selection, or some other reason). I don't have a clear enough vision to say much more, but I think there are some interesting implications about what we mean when we say, this group or that group went extinct. 

Today was a pretty bad day for American democracy IMO. The guy below me got downvoted and yea his comment wasn't the greatest but I directionally agree with him.

Pardons are out of control: Biden starts the day pardoning people he thinks might be caught in the political crossfire (Fauci, Milley, others) and more of his family members. Then Trump follows it up by pardoning close to all the Jan 6 defendants. The ship has sailed on whatever "constraints" pardons supposedly had, although you could argue Trump already made that true 4 years ago. 

Ever More Executive Orders: Trump signed ~25 executive orders (and even more "executive actions" - don't worry about the difference unless you like betting markets). This included withdrawing from WHO and ending birthright citizenship, though the latter is unlikely to stick since it's probably unconstitutional. I haven't had time to wade through all the EOs but like the pardons, this seems to be a cancerous growth of executive encroachment on the other branches with no clear end in sight. 

Pre$idential $hitcoins: To be fair, the $TRUMP coin happened a few days ago, with $MELANIA following more recently. I'm not sure if people remember this, but it was a genuine scandal when Trump didn't release his tax returns in 2016. Now, 8 years later, he is at best using his office to scam American citizens. Less charitably he has created a streamlined bribery pipeline. It was a blip in the news cycle. 

benchmarking model behavior seems increasingly hard to keep track off. 

I think there are a bunch of separate layers that are being analyzed, and it's increasingly complicated the degree to which they are separate. 

e.g.

level 1 -> pre-trained only 
level 2 -> post-trained
level 3 -> with ____ amount of inference (pro, high, low, thinking, etc.)
level 4 -> with agentic scaffolding (Claude code, swe agent, codex)
level 5 -> context engineering setups inside of your agentic repo (ACE, GEPA, ai scientists)
level 6 -> The built digital environment (arguably could be included partially in level 4, stuff like api's being crafted to be better for llms, workflows being re written to accomplish the same goal in a more verifiable way, ui's that are more readable by llms). 

In some sense you can boil all of this down to cost per run, like ARC, but you will miss important differences in model behavior on a fixed frontier. 
 

https://jeremyberman.substack.com/p/how-i-got-the-highest-score-on-arc-agi-again 

if you read J. Berman substack, you will see he uses existing llms to get his scores with an evolutionary scaffolding (hard to place this as being level 4/5). While I'm decently bitter lesson pilled, It seems plausible we will see proto-agis popping up that are heavily scaffolded before raw models reach that level (though also plausibly we might just see the really useful, generalizable scaffolds get consumed by the model soon thereafter). The behavior of j bermans system is going to be different than the first raw model to hit that score with no scaffolding and pose different threats at the same level of intelligence.

Haven't seen anyone mention RAND as a possible best charity for AI stuff and I guess I'd like to throw their hat in the ring or at least invite people to tell me why I'm wrong. My core claims are approximately:

• Influencing the US (federal) government is probably one of the most scalable cost-effective routes for AI safety.
• Think tanks are one of the most cost-effective ways to influence the US government.
• The prestige of the think tank matters for getting into the room/influencing change.
• Rand is among the most prestigious think tank doing AI safety work.
• It's also probably the most value-aligned, given Jason Matheny is in charge.
• You can earmark donations to the catastrophic risks/emerging risks departments
  
  I'll add I have no idea if they need/have asked for marginal funding.
  
   

Relating to the Bernie news today, I don't think capitalism is compatible with believing in the singularity and some combo of humanism/utilitarianism. I have been a neolib for a long time before now fwiw. If you ran the zoo would you give the biggest banana to the sexiest gorilla?

argument about anti-realism just reinforces my view that effective altruism needs to break apart into sub movements that clearly state their goals/ontologies.  (I'm pro ea) but it increasingly doesn't make sense to me to call this "effective altruism" and then be vaguely morally agnostic while mostly just being an applied utilitarian group. Even among the utilitarians there is tons of minutiae that actually significantly alters the value estimates of different things. 

I really do think we could solve most of this stuff by just making EA an umbrella org for ea minded orgs that have specific goals and ontologies (democratic negative utilitarians, dictatorial anti realist dog lovers). 

TBF I don't think it matters that much for above if you are/aren't anti realist rather it just reminds me that even here I have large worldview differences to a lot of people (which is totally fine) but I think it would be more clear to everyone if there were more de jure divisions. 

I'll start by saying I absolutely think it's a terrible idea to try to destroy humanity. I am 100% not saying we should do that. Ok, now that we have that out of the way. If you decide to commit your life to x-risk reduction because there are "trillions of units of potential value in the future", you are in a bit of sticky situation if someone credibly argues that the expected value of the future is lower if humans become grabby than if they don't. And that's ok! It's still probably one of the highest EV things you can do. 

And I'll say it again years later, https://forum.effectivealtruism.org/posts/KDjEogAqWNTdddF9g/long-termism-vs-existential-risk

this ^ post is not great. The entire thing basically presupposes that human society is positive, that aliens will not exist, that animals will not re-evolve if we die. I wouldn't bring this up if not for it being one most upvoted posts on the forum ever (top 5 if you don't include posts about ea drama). 

I stopped being vegetarian almost 2 years ago - one of the biggest reasons I'm not a vegetarian is that I stay up late pretty much every day and don't always feel like cooking or eating snacks so I will go to whatever is open near me. During university, nothing really stayed open after 10 anyway because Evanston is a lame place. So I would often eat at or before 10, and if I was eating out there were vegetarian options (stir fry with tofu, chipotle, etc.) still at this time.

Now I live in a predominantly eastern European and Mexican area of Chicago. There isn't much vegetarian food in this neighborhood in general, although there is some still. However, the vegetarian restaurants here seem to service a wealthier demographic than the non vegetarian food. It closes earlier, more expensive, etc. The cheap and late night options are fast food and taquerias,  which essentially have no quality vegetarian items. But since this stuff is open, it actually makes me lazier and I'll often eat at 11:00 PM because I can. However getting into this routine means I eat more meat. 

I'm pretty sure if there was a decent cheap vegetarian restaurant that stayed open till 2:00 am I would eat at least 1 less meat meal a week, probably 2-3. 

why aren't there any vegetarian late night options near me? probably the normal reasons - no one around here wants or can open one, or there isn't enough demand. 

In either case it got me wondering. If there is enough demand to recoup say 95% ish of cost for a late night falafel stand, would it be a cost effective intervention (over whatever other things ACE recommends) to fund that last 5%? I might think more about this unless it's super obvious to someone that this is orders of magnitude worse than other options. 

Grant-making as we currently do it seems pretty analogous to a command economy.  

Are there estimates for the different per animal unit suffering from animal consumption in different countries (holding animal constant)? 

Theoretical idea that could be implemented into Metaculus

tldr; add an option to submit models of how to forecast a question, and also voting on the models. 

To be more concrete, when someone submits a question, in addition to forecasting the question, you can submit a squiggle -- or just plain mathematical model -- of your best current guess of how to approach the problem. You define each subcomponent that is important in the final forecast and also how these subcomponents combine into the final forecast. Each subcomponent automatically becomes another forecasting question on the site that people can do the same to (if it is not already one). 

Then in addition to a normal forecast, as we do right now, people can also forecast the subcomponents of the models, as well as vote on the models. If a model already includes previously forecasted questions, they automatically populate in the model. 

The voting system on models could either just draw attention to the best models and encourage forecasting of the subcomponents, or even weight the models estimates into the overall forecast of the question. No idea if this would improve forecasting but it might make it more transparent and scalable. 

I wrote a bit more in this google doc if interested. 

edit: I think this might just be guesstimate with memoization

EA tends to be anti-revolution, for a variety of reasons. The recent trump appointments have had me wondering if people here have a "line" in their head. By line I mean something like, I need to drop everything and start protesting or do something fast. 

 Like I don't think appointing RFK jr. to health secretary is that line for me, but I also realize I don't have a clear "line" in my head. If trump appointed a nazi who credibly claimed they were going to commit mass scale war crimes to the sec defense, is that enough for the people here to drop their current work? 

I'm definitely generally on the side of engaging in reactionary politics being worthless, and further I don't feel like the US is about to fall apart or completely go off the rails, but it would be really interesting to see if we could teleport some EAs back in time right before the rise of hitler or pre chinese revolution etc. (while wiping their brains of the knowledge of what would come) and see if they would say stuff like "politics is the mind killer and I need to focus on xyz"

What do people think about posting urban planning/yimby/ local gov policy thoughts on the forum? 

I find that stuff really interesting and admittedly don't believe it is the most important stuff to work on but is "effective altruism" if you have an extremely local moral circle.

Has anyone thought through if EA should try to start a charter school/ done some sort of impact estimate? Specifically with the purpose of making a really good school, not specifically related to EA outreach.

I searched google and the forum for posts on this and couldn't find anything. 

I think monied prediction markets are negative EV. The original reasons the CFTC were not allowing binary event contracts on most things are/were actually good reasons. It's quite clear that our elected officials can get away with insider trading (and probably to a certain extent market manipulation). My intuition is that in the current admin I expect this behavior to increasingly not be punished and maybe actively encouraged. Importantly, insider trading on the existing financial instruments doesn't really work. My take here is just that the marginal value of a piece of information is pretty low for traditionally financial markets, so while it might allow a high-tech research shop to beat the market it doesn't even cover slippage/the lack of complex modeling done when in the hands of a dumb paper congress person. This is not the case for most/all binary event markets, where a single marginal piece of information can credibly flip probabilities from <1 -> >99. 

Insider trading, which I expect to be much more likely to happen at scale with these markets, is not that bad. It degrades the integrity of the prediction markets themselves and might lower volume but probably not that relevant for EA. However market manipulation could get really really bad in terms of EV (I think significant market manipulation is currently quite unlikely at the current levels of liquidity on these markets). In particular, the less likely something is to happen, the more incentive a prediction market provides for manipulating the market. And like in insider trading, I think market manipulation is going to be much more accessible and effective for generating alpha vs traditional markets. 

low stakes example: Trump's speech writer puts random phrase ( Lock x up, I hate x country, Sperm)

medium stakes: Every election market is an assassination market, and a bounty for either candidate to drop out

High stakes: "will x leader/country bomb  y"

Of course with the medium and high stakes examples, we are far from the liquidity on these markets to make the incentives provided to be worth seriously considering. But what if we 100x? what if e.g. mamdani could make 50 million dollars by dropping out instead of 500k? What if trump could make 20m by delaying a ceasefire a few days?

To some extent we will just have to see how the enforcement of this stuff plays out as that will shape the incentives. It might seem obvious that the stuff above will be easily sorted out by the relevant federal bodies. Again, my intuition is that (1) I don't actually expect this to be a huge priority right now and (2) the regulatory burden to properly regulate this stuff is ridiculously high (both in terms of just tracking everything that is happening and actually trying to apply the law to decide if various things constitute as "security fraud"). 

It's already hard for the SEC to enforce behavior the NYSE or CME, and that's with mostly big institutional players who cooperate and record all their communications, etc. I'd have to imagine especially in the short term the vast majority of questionable or illegal behavior on prediction markets will slip through the cracks. The space of potential "securities fraud" is just so ridiculously big and confusing when you have 1000s of random binary event contracts. 

Then you put that with the encouragement of gambling and incentivizing the spread of fake news, I just think we should be very very skeptical of going head first into this.

And the alternatives are pretty good! Non monied prediction markets like manifold and metaculus are huge improvements over previous epistemics, and to be fair the monied prediction markets are quite literally crowding them out. It's not really fair to compare polymarkets accuracy to manifold, when a bunch of people who are on poly would probably use manifold if they made poly illegal (and that being said I'm pretty sure manifold doesn't stack up to poorly but don't have up to date stats). And some of the most important binary event contracts were already being traded on CME because they didn't suffer from the issues I listed above. 

Given the public good nature of prediction markets, the government should be quite willing to front $10s-100s of millions to improve their quality. And there are lots of ways to improve the markets without directly providing personal monetary incentives. This could mean improving the site ui/ux, creating a public leaderboard with some sort of recognization/award, improving the question statements/resolution, or even providing prizes that can be donated to a charity of the winners choice. 
 

WSJ Op-Ed: ‘Effective Altruism’ Is Neither - I didn't see any discussion of this low effort but scathing review of EA. I wonder if people feel we should write a response? 

The article itself is bad-faith, bordering on the stupidest critique I have read on EA thus far, but I feel like WSJ op-ed is pretty important, and if SBF signed off on a response they would most likely publish it. 

Re the  "EAs should not should" debate about whether we can use the word "should" which pops up occasionally, most recently on the "university groups need fixing". 

My take is that you can use "should/ought" as long as your target audience has sufficiently grappled with meta-ethics and both parties are clear about what ethical system you are using.

"Should" (to an anti-realist) is shorthand for (the best action under X moral framework). I don't mind it being used in this context (though I agree with ozzies previous shortform on this that it seems unnecessarily binary), but it's problematic using this word around people you don't know or non-philosophy heads. It's completely absurd to tell an 18-year-old or anyone else who doesn't know what utilitarianism and virtue ethics are that they "should" do anything, and if they believe you, then you tricked them into that view (unless you are a moral realist, which I think is also absurd). 

 If your target audience does not know what the is-ought problem is, it's better to stick to output-based cost-benefit and not enter into this "cause agnostic" tier list type thing since inter-output rankings rely on arbitrary metaethical functions that aren't well-known by most or standardized for quick and reliable reference.

However among my friends, we use should all the time because we know what generally mean (our relatively shared utilitarian-ish meta-ethical worldview), and we feel comfortable clarifying this if it seems to be the crux of the debate. But at this point, should loses all of its emotional oomph and maybe it's just not worth the hassle to shorthand a 7-word sentence. 

Would be interesting to compare my likes on the ea forum with other people. I feel like what I up/downvote is way more honest than what I comment. If I could compare with someone the posts/comments where we had opposite reactions, i.e. they upvoted and I downvoted I feel like it could start some honest and interesting discussions. 

Assume there are two societies that passed the great filter and are now grabby. Society EA and society NOEA. 

Society EA you could say is quite similar to our own society. The majority of the dominant species is not concerned with passing the great filter and most individuals are inadvertently increasing the chance of the species extinction. However, a small contingent had become utilitarian rationalists and speced heavily into reducing x-risk. Since the group passed the great filter, you can assume this is in large part due to this contingent of EAs/guardian angels. 

Now society NOEA is a species that passed the filter, but they didn't have EA rationalists. The only way they were able to pass the filter was because as a species, they are overall quite careful and thoughtful. The whole species rather than a divergent few has enough of a security mindset that there was no special group that "saved" them.

Which species would we prefer to get more control of resources? 

The punchline is that the very fact that we "need" EA on earth might provide evidence that our values are worse than the species that didn't need EA to pass the filter.

I feel like " x-risk" is basically tautologically important and thus ceases to be a useful word in many cases. It's like the longtermist equivalent of a neartermist saying "it would be good to solve everything really bad about the current world".  

I just watched Weapons in theaters. The amount of hype and critical acclaim this movie got caused me an emotional response somewhere between rage and sorrow. I can definitely be a bit pedantic and nitpicky as a person. Still, I can also totally enjoy movies that make no sense and/or especially suspend disbelief for magic or world rule changes that the plot purposely introduces. And in fact, I did enjoy this movie somewhat; it was well shot, well-acted, and fun. 5-6/10. Entertaining but ultimately meaningless and without real themes or allegories because it was an idiot plot of the highest degree. This movie put some classic miscommunication-based rom-coms to shame in that regard. I don't 100% agree with everything in this post, but it explains the general outline of the plot holes in the movie for those who have seen it or don't care about spoilers. 

I'm somewhat ok with people taking the perspective of "look i know this movie makes no sense but i really enjoyed it and thought the campyness/nonsensicalness added to it" and/or "I like this movie in spite of the plotholes". That being said, the degree to which people seem to be able to enjoy vaguely serious-seeming movies despite massive, world-shattering plot holes doesn't sit right with me. It feels like a socially acceptable form of wireheading. The average person has a slider of how much they care about whether or not things make any sense in the current moment that they can swing up and down wildly at will, without even realizing it.  And yes I already knew this. I know what a blockbuster hollywood film looks like, I know people like ridiculous conspiracies, wasting all their time consuming brain rot, how the average voter chooses to vote, and yes i've met and engaged with a wide variety of people in my life. I still get triggered because it does matter to me that things make sense. I'm aware that it is a waste of time to loop about this and that it should update the reader more on my aspie traits than the state of the American public. 

It's also possible that most of the difference in ratings that viewers and critics gave the movie is due to them not agreeing with me about the plot holes being plot holes. This doesn't make me feel much better because that would imply I'm on a completely different plane of reality, which is also probably true. 

(This comment is probably out of scope for the forum. On the other hand, I generally think the forum would be better if people felt comfortable spewing more low effort garbage so i'm going to go for it but totally understand if people don't want stuff like this polluting the forum.)