Arts & Letters #12
OpenAI, schizo posts, legacies, genetic engineering, and a bit of spice
No comment, and I have to admit, I struggle to read everything @thezvi writes, but…recent developments at OpenAI feel genuinely more important than the presidential election and
is absolutely on top of it.
Every field has some real scientists doing excellent work. But twenty-nine thirtieths of their peers are cargo cultists, outright fraudsters, or at best just aren’t all that great at their jobs. Much has been made of “impostor syndrome” in academic research. The fundamental cause is that there aren’t enough non-impostors to fill all the positions.
The effect is similar to a distributed denial of service (DDoS) attack. This is when an attacker shuts down an internet server by sending a massive flood of bogus communication requests, overloading the machine’s bandwidth until it is unable to respond to legitimate requests. Academia’s minority of serious researchers are also flooded with a massive, hostile flood of bogus communication requests. If most of their ostensible peers can’t discover new knowledge, then how will they find peers to work with? If most publications aren’t worth the time it takes to read, then who will bother to read the minority that are actually good?
Intellectual progress is held back substantially by these problems. If the academics do not raise their standards—and it seems like the opposite is more likely—then over time the real work will move gradually, and then suddenly, into different channels that are able to filter out the noise. This process is already underway.
Bravo to
, I’ve found the meme “academic DDOS” very useful and intuitive; people instantly intuit what it means, it “clicks” for them. Other than that, it’s just a good summary of the decline of academics and the issues following the replication crisis.In actual usage, though, I’ve found it deeply depressing. I’ll start reading through an interesting substack and I’ll see links to academic papers and then there’s just a wall of 40 academic papers and…I just tune out. Which sucks, because I remember being really excited about articles like that at one point, I loved them, and now I’m just like “You haven’t deeply read 40 academic papers, that’s literally 3-6 months of work at minimum. This is just a wall of abstracts, a few of which are good, most are garbage.”
And let me be clear how I filter, for anyone who thinks I’m overly dismissive. I will usually accept a paper if it simply provides me with the underlying data and code. We’re almost 20 years from Ioannidis’ replication paper and sharing your code and data should be the bare minimum. We’re all using R, Python, or Stata, very few papers use more than 1 GB of data, I’m more than competent to read, and it takes you what, 5 minutes to upload a csv file and a Python file? If they’re not going to do the easiest thing to address replication concerns, just fully sharing their data and code, then I don’t trust their output. And so far that’s been pretty effective.
A scientist in the thick of research doesn’t spend a lot of time wondering why he’s doing what he’s doing. His mind is occupied with the immediate, practical concerns of what to do next, when to do it, how to do it, how to pay for it, who to collaborate with, whom to cite, where it can be done, and where he can be hired to do it. The purpose of the enterprise rarely enters his mind at all. It is a non-question, which can yield no practical benefit to his research program. Yet every once in a while, in those quiet moments as he’s drifting off to sleep or staring out the window over the campus green, the dread question of Why emerges to trouble his conscience.
…
Follow that chain of scientific genealogy far enough back, say to the 18th century or before, and matters become rather different. In those days there was no such thing as a ‘scientist’, but rather his more noble precursor, the natural philosopher. The natural philosopher had no difficulty whatsoever answering the question of Why. His enterprise was a religious one: to use his rational mind to read the Book of Nature, and thereby draw closer to the Mind of God that had written His billions of names across the infinite facets of His Creation. Research was an act of devotion.
…
Oh, of course it’s nonsense to say that God is literally dead. One cannot actually kill the Almighty.
He is merely dead in our hearts.
And so, our hearts have died.
I’ve sat with
’s essay for awhile, it’s a monstrous schizo post…book-thing.Of all the lunacy, I’m surprised this part near the front has stuck with me but it really has and the guts is that I’m surprised how much I WANT thought to be divine.
I get manic when I code, when I’m deep into something, and I’ve noticed this in others. People like to think, they want new ideas, they want to dive into a new subject learn about it. I don’t know about normies but the kind of people I hang out with don’t dismiss things which disprove their worldview, they HUNGER for it, that moment when some singular element in your conception of the world changes, that change which causes you to update another element, which updates another element, and then everything changes and your whole perception is different and your neurons are alive with lightning.
There’s a manic, junkie element to deep thought. Man, I want that to be more than just a high.
One missing ingredient however is the level of will. Like Richard Hanania discussed in ‘Why is Everything Liberal?’, in a representative democracy it is a matter of cardinal utility, aka ‘who cares the most’. The Gulen movement was deeply committed to infiltrating the state apparatus and prepared to play a long game, and even though the movement itself lost power, they still managed to dismantle the secular Deep State so that Erdogan, a fellow Islamist, could fill the void.
It also lubricated his path to power. At every step, he got support from people who were tired of the military unfairly controlling everything. Liberals who in other countries might have resisted his rise judged him the lesser of two evils, or even the underdog, and held off. Muslims, including the Gulenists, understood his ascendancy as their only route to fair political representation.
Interesting to compare these two on how Erdogan took over Turkey, especially the takeaways, from one guy trying to prevent it and another to duplicate it. But the main thing I think
added, and it’s worth going through more, is adding an entire section about the Gulen faction.Here’s the thing. It was kind of an open secret that Turkey was ruled by a secret cabal of military leaders. Like, very explicitly. And this lead, very naturally, to a lot of people becoming discontent with their informal rule and building their own networks. I don’t think Scott Alexander ignored this but focusing on Erdogan automatically pushes other groups into the background. Groups like the Gulens were very important, they’re described as a super-powerful Federalist Society who seized the judiciary and basically made Erdogan’s seizure of power possible.
But
’s call for right-wing unity feels…odd. That does not appear to be how Turkey changed. Instead, lots of groups were deeply displeased with the status quo and built their own powerbases. Erdogan didn’t form the Gulen faction, he didn’t march through the institutions. Instead of one, great united front, we’ve got multiple factions all trying to use Erdogan for their own political ends, along with assuredly various independent groups, like the Kurds, without much use for any of them. What’s described sounds very…decentralized.
The compounding effect of loving, super-rational instincts means that when humans are socially arranged appropriately they become more than the sum of their parts. When parents look after their own children rather than random people each day, the broader social fabric is enriched by the greater presence of this super-rational love and self-sacrifice. On this basis ‘unconditional love’ makes its appearance. This in turn compounds into otherwise unachievable social integrity and harmony.
Conversely, while it may be possible to achieve a greater number of seemingly morally desirable outcomes by abandoning your children and giving more money to charity, it must be acknowledged that by disregarding this central moral responsibility you are undermining your commitment to morality itself. Having abandoned your super-rational motivation for self-sacrifice, using your newfound freedom to achieve arbitrary and self-serving objectives is the natural next step.
I find this appeal to moral partiality extremely interesting. It seems to explain a number of situations where…good people take a variety of individually good moral actions and yet we end up in a worse situation than we started. A lot of homeless policy feels this way, everyone is doing their best but somehow the aggregate of all those individually good decisions is a culture of enablement that makes it even worse. A similar thing seemed to happen with Sam Bankman Fried, a culture that did a lot of good things somehow generated a really bad actor.
However, I think there are two shortcomings with
’s essay here:First, less importantly, it feels like
is critiquing Effective Altruism and utilitarianism without…challenging utilitarianism in its strongest form. Yes, many of the critiques he makes are accurate for many people but there are people who absolutely are true, sincere utilitarians.To use
’s analogy, there are people who would absolutely run into a burning building, save a $100 million painting rather than a baby, sell that painting, donate 100% of the proceeds to malaria bednets, and save 30,000-40,000 lives. And if you’re going to critique people for letting a baby burn to death, which, fair, then you also have to acknowledge those 40,000 lives saved. You can’t dodge that aspect, you can’t dodge the actual impact of utilitarianism.Yes, there is uncertainty on how many lives would be saved and there are complex systemic factors…but malaria bednets which clearly and obviously reduce malaria infections and clearly work are not a good example. No, being an effective altruist does not require you to donate 100% of your income to charity anymore than being a Christian requires abandoning all material possessions. And yes, many people who are charitable are also jerks but…just spend some time around Effective Altruists, this is not a realistic portrayal of them.
That’s why I reference the social-fabric argument above, because what I really want to see is something like that go head-to-head with utilitarianism. I want to see the straightforward critique, not a…strawman is unfair but…an argument that doesn’t take on the more serious impacts of Effective Altruism and broader utilitarianism is one that will be easily dismissed.
Secondly, I think the more important shortcoming is legacies are not binary; that we all have different circles of caring and obligation and what we need is guidance on balancing all those different obligations. You don’t only have your immediate family; you have your extended family, your church, nonprofit organizations you may have been involved in, friends or (if you’ve done very well) loyal employees or personal assistants. There are people and organizations at a level of caring and obligation between your immediate family and distant strangers. What’s needed is guidance on how to balance all these competing claims.
For example, if a married couple have tithed to their church their entire life, do they leave 10% to the church? If they’ve also been active patrons of, say, their local ballet company and they’d like to leave something to them, how do they manage that? Do they leave their family 90%, the church 8%, and the ballet company 2%? Or do they leave their family 88%, the church 10%, and the ballet company 2%? How do you balance these interests?
Worse, because it’s money and
has chosen to focus explicitly on legacies and inheritance, it is so much easier and more important to quantify those relationships and obligations. A debate about volunteering 2% vs 3% of your time and energy at the local ballet is largely a meaningless abstraction but leaving 2% vs 3% of your estate is an easily quantifiable difference of $20,000-$50,000 or higher, depending on your wealth. While I don’t think @Kurtz, or many devout Christians, enjoy or approve of quantifying the value of relationships…this is the part of life where it is both easiest and most important to precisely quantify one’s obligations and relationships.
Those with a pro-surrogacy inclination may want to dive into an empirical argument about how a gay couple can be good parents or about how the outcomes for children born via surrogacy are not inferior. While the empirical arguments are worthwhile, there is a much more concise argument that severely undermines the case for opposition to surrogacy. If you are concerned about the well-being of a child who never meets his or her biological mother, then you need to seriously consider the alternative scenarios.
If surrogacy for gay couples were made illegal, the counterfactual is not that the surrogate mother keeps the baby. The correct counterfactual is that the child would have never been created in the first place. Whatever negative aspects you believe there are to never meeting one’s biological mother, it is quite difficult to imagine that never having been born is better.
Conversely, I think
should more broadly consider the potential consequences of…social externalities, for lack of a better term. Essentially, when evaluating consequences, we can’t only judge the direct impacts to participants but we also need to consider second and third order social and cultural consequences. And these get measured out at the societal level and therefore our counterfactuals need to involve entire societies, rather than just individuals.For example, it seems trivial that, while for any individual person, the changes of the Sexual Revolution might have been better, there were obvious consequences cultural and social consequences beyond the people directly involved, the consequences of which are still a point of live debate. We understand externalities from economics, a factory might buy raw natural resources, process them, and sell manufactured goods, all within a voluntary framework that’s profitable for everyone, and still be negative for society if the effects of pollution and deforestation on the local community are severe enough.
I don’t want to take this further, as I think
pretty narrowly tailored his original point and I want to respect that, but I think it is important to not only consider the counterfactual in the lives of the people directly involved but the counterfactual of the lives of all people in the society if the changes are made, at scale and over time.https://www.lesswrong.com/posts/yT22RcWrxZcXyGjsA/how-to-have-polygenically-screened-children
First, there is no company that publicly offers embryo selection for intelligence. I have spoken with a stealth mode startup that offers selection for disease and non-disease traits, including intelligence. If you're interested in selecting for non-disease triats, you can get in touch with them via Jonathan Anomaly, who knows some of the people working at the company.
Their current predictor correlates with measured IQ at about 0.4, which means they've likely compiled data from multiple sources to create it.
…
There is significant room for this benefit to improve in the near future. The million veterans project in the US has whole genome sequences and ASVAB test scores for (you guessed it), a million soldiers. If researchers were simply allowed to use this existing data to create an intelligence predictor, the gain from embryo selection would increase to 8.5-13 IQ points and the racial disparities in predictor quality would mostly disappear. The marginal cost of this would be virtually zero.
It’s been interesting to watch this scene develop since Gwern’s original monster post in 2016. Honestly, this last year or two have been weird, as it seems like it could go mainstream at any moment, provided you could get 30-40 parents willing to use the service. Maybe it’s for the best it’s staying low-key.
I do love the Less Wrong predilection for living cyberpunk plots in the most mundane manner possible. Like, if you want to genetically engineer your kids to be geniuses you need to talk with a guy in Austin who can hook you up with Johnny Anomaly in Quito who knows a stealth startup that can help you…for a price. You could run this with no alterations in Shadowrun.
Today’s Twitter contretemps involved one David Austin Walsh, a history postdoc at Yale. (For those who don’t know, a postdoc position is a sort of low-paid purgatory where people with PhDs get sent to keep doing research when they can’t immediately get jobs as professors.) Walsh is the author of a recently released book, Taking America Back: The Conservative Movement and the Far Right, but he’s best known as a vocal and highly opinionated commentator on Twitter. When my podcast co-host Brad DeLong declared in 2019 that it was time for neoliberals like himself to “pass the baton” to their colleagues further to the left, it was a tweet thread by Walsh that he cited. In recent years, Walsh has spent much of his time attacking Joe Biden from the left; these attacks have become more strident since the start of the Gaza war.
But that is not what got Twitter — or X, as it’s now officially known — in a tizzy today. Instead, it was Walsh’s declaration that his failure to get a tenure-track academic job is due, at least in part, to the fact that he is a White man:
…
But in addition to all that, I think academic advisors and other influential figures should work to make educated people’s expectations more vague. “I’m going to be a tenured professor in the field of 20th-century American history at a good university” is a hyper-specific expectation that’s very hard to meet. “I’m going to have a decent-paying job doing something intellectually interesting, while also perhaps writing a book on the side” is a very vague expectation that is eminently achievable for almost all American PhD holders. American strivers need to learn to go where the winds of life take them. Take it from me — it’s more fun that way.
Man, this…I’m trying to stay away from politics and hot takes, but…the subtext appears to be that a 35 year-old postdoc from Princeton with a social media following and good references can’t get a tenure track position anywhere. And people are focusing on the racial element, he’s white, but I’m stuck on the Princeton thing. Really? A Princeton PhD can’t get a tenure track job at, like, Ohio State? Then, and I hate to phrase it this way, but what is the point of Princeton?
I feel like I’m missing something. It looks like 3 of the current US Supreme Court justices did their undergrads at Princeton (Alito, Kagan, Sotomayor), so pretty elite…but a history PhD should…I dunno, the comments on Noah Smith’s blog are suggesting becoming a middle school teacher? “Set his expectations lower”. Maybe start a Substack? Dude, the whole point of Princeton is to go hobnob with the other proto-Illuminati, right? Not to be struggling to find a job at 35.
At a fundamental level, is Princeton elite? If it is, why does a Princeton PhD in history struggle to get a job teaching and researching history? If not…why do we care about Princeton? Why should we assume its graduates are elite?
#9
https://www.themotte.org/post/1028/quality-contributions-report-for-may-2024
Some highlights I pulled from themotte’s quality contributions for May. While I don’t follow the motte, I have found these highlights very interesting and I keep coming back every month for the quality contribution roundup. I wish datasecretslox did something similar.
https://www.themotte.org/comment/208493?context=3#context
A really nice run down on the upcoming South African elections. The story appears to be that South Africa’s main party, Nelson Mandela’s ANC, is fracturing with two parties splitting off to their left: the EFF who…basically sound like violent communists and the MK, who are the party of Jacob Zuma who is kinda of the Donald Trump of SA without the right wing politics but with twice the zaniness.
https://www.themotte.org/comment/209912?context=3#context
A run down of a series of recent bipartisan bills where conservatives feel betrayed, such as the Pregnant Workers Fairness Act being used to extend abortion protection. This wouldn’t be less of an issue if Republicans weren’t assured this wouldn’t happen while the bill was being written and discussed. Without commenting on the actual facts of the matter, I do not in fact have in-depth knowledge of the debate over random bills drawn up over the past 10 years, I hadn’t considered how hard it is to make deals and compromises with a counterparty when that counterparty can’t actually enforce the deal. Subtleties of language that are clear to the legislative authors and may represent hard-won compromises actually get enforced by judges and executive officers who may not see or respect those subtleties. That’s a whole other level of complexity.
https://www.themotte.org/comment/213098?context=3#context
I think the main thing that boomers and even elder Millenials might be missing is that literally EVERYTHING in Zoomer culture is in a constant state of molochian hypercompetition/red queen races thanks to the influence of social media and algorithmic ranking of every aspect of their performance in life.
This is something I’m starting to wrap my head around. Like, social media is bad, m’kay, and so I generally avoid it but there is entire generation who not only grew up with social media but grew up with a way more addictive and harmful version of social media than I did. Like, Facebook was bad back in 2012 and we kinda knew but Facebook seems kinda quaint compared to the disfunction of Instagram and TikTok culture and I’m struggling to wrap my head around the fact that there are people in their 20s who grew up in this, who are inundated with ever-increasingly addictive technology and don’t know any other way.
https://www.themotte.org/comment/214773?context=3#context
Not actually a uniquely insightful comment but that line: At some point down the line, the (political) escalation for one side or the other will be unsurvivable to the system as a whole, and it will fail. Again, people counting on the system's survival should be made aware of this.
That last line is sitting with me. Not that I’m a government employee or anything but…I have a 401k. I will depend on that 401k for 20-30 of my life during retirement. And, to keep this as neutral as possible…the political events of the past 8 years are introducing elements of political risk that were not there in the 90s or 2000s.
Allow me to phrase it this way. Maxing out your 401k contribution for 30 years of your career and putting it in an index fund will guarantee you a good retirement. Looking 20-30 years into the future, there are certain political scenarios where that 401k looks less secure. Not a bad option but, like 95% instead of 100% guarantee.
But mostly I’m just wrapping my brain around two concepts. First, despite the fact that I don’t think of myself as terribly invested in the system…I kinda am. Second …I suspect there’s a big gap between what people say online and how they actually invest their money…but the situation in the US does seem legitimately quite chaotic and getting worse.
Appreciate and enjoyed the thoughtful commentary