Crossposted on The Memeticist and LessWrong
I discovered the rationalist community two years ago. I had just dropped out of college, was looking for a software engineering job in London, and was in the process of getting what would become The Case for Dropping Out of College published.
During my research for the piece, I had stumbled on Against Tulip Subsidies on Slate Star Codex. I found the post brilliant, and the couple of other posts I then read seemed equally insightful. The site also listed meetups, and so I decided to go to one which was organized by the 'London Rationalish' group.
The people I found there were welcoming, but I was suffering of mild impostor syndrome. At that time, I had only read a handful of SSC posts, and if I recall correctly, I wasn't even aware of the existence of LessWrong.
Also, the fact that many there had advanced degrees didn't help.
Nevertheless, I greatly enjoyed the discussions. I had been interested in EA for at least two years before that and consumed some rationalist-adjacent content such as Sam Harris' podcast, —where I first encountered Yudkowsky. I've also had a long-running interest in psychology, physics, and economics. So, I did end up feeling quite at home culturally.
At the first meetup I attended, I told people my story. I explained why I had dropped out of college, and the conclusions I had come to about the education system in general. I was a little surprised when I was met with near-universal agreement.
I found it quite validating and reassuring. I felt alienated at the time. I felt betrayed by society for keeping up with such a system. I felt doubt, because what are the odds that I had understood the problem, while all the people in charge were all oblivious? I was quite confident in my object-level model, but the argument from modesty still made me wonder whether there was something out there that would prove me wrong.
So meeting a group of young highly educated adults who all more or less agreed with me was comforting. But it was also a bit disconcerting. There was something bizarre about finding a group of people who all disavowed the institutions of higher education, while also having spent so much time within them.
Not that there would be anything irrational about it. From an individual perspective, it makes sense to go to college. College may be largely about a zero-sum competition for credentials, but little is achieved by single individuals forgoing getting a college degree. Education is in a bad Nash equilibrium, and only governmental action or mass coordination can get us out of it.
I broadly agree with this position (unlike what the slightly misleading title of my Quillette piece might suggest). But there still is something about this that doesn't sit right with me. An attitude that I have now encountered a few times after having explained that I dropped out of college because it was expensive and I wasn't learning much is something like, "but haven't you heard of Caplan's work? You have to go to college, or you won't get the signaling."
There is some truth to this, but there is something seriously going wrong when the very people who believe that education is in a bad Nash equilibrium become de facto enforcers of the status quo. Shouldn't EA and rationality be about overcoming this kind of situation? Why have there not been more attempts to bust smart nerds out of the hands of Moloch? To help them escape the immoral maze that is the current higher education system?
What makes this more depressing is that many of the most prominent EAs are working in higher education, and young EAs are generally advised to gain career capital in the form of university credentials.
At the same time, organizations like 80k and GiveWell generally advise against trying to improve education. In their problem profiles page, 80k presents about 40 problem areas. Not one of them is about education.
But how can it be that EAs see so much wrong with the education system, while also spending so much time in university—and then not see improving education as a cause with a great potential to do good? Shouldn't education be an area with a significant amount of leverage?
How many people have to recognize that higher education is seriously dysfunctional before they should try to coordinate an exit? At what point does a fatalistic attitude vis-a-vis improvements of the education system become a self-fulfilling prophecy?
How many people need to recognize this prisoner's dilemma before they stop defecting?
The Case For Education
"Education is the process through which the body of knowledge and culture that we have been accumulating for thousands of years passes from our generation to the next one." -Francois Chollet
It's easy to become cynical about higher education.
When I hear praise for the promises and virtues of education I often roll my eyes. Not because I disagree that education has the potential to change the world for the better, but because the institutions that people are defending by making vague idealistic statements about education are doing a terrible job at providing an education that actually has a positive impact.
So let me now make the case for education.
Education is key to civilizational sanity, sensemaking, and survival. Education is key to The Secret of our Success.
Education is the scaffolding on which our society, culture and civilization are built and maintained.
For the first two or three decades of people's lives, they are taken on a guided tour of what humanity has learned and achieved. They are taught the laws that govern our universe, how these laws gave rise to life on earth, and ultimately led to the ascent of our species. They are instructed in the history of the ideas that shaped the trajectory of human civilization, the mistakes made along the way, and the principles of rational thought that have allowed the best ideas to win out. They are introduced to the institutions that form the bedrock of society and have allowed the current generation to live in relative prosperity.
And as they learn about the world, and their place in it, they are also shown the construction sites and inadequacies of our budding collective project so they will be able to contribute to maintaining and improving on what previous generations have built. And so, just as 20,000 years ago, the information needed for the survival and thriving of a hunter-gatherer tribe was transmitted from generation to generation, today the ideas, information and skills necessary for civilizational sanity and the survival of our species continue to be transmitted.
Let me be more specific. There are two areas for which education is needed in the modern world.
The first is economic growth. Education increases human capital and allows people to be more productive. The success of an economy depends on the ability of its participants to create value for others, and it is the acquisition of knowledge and skills, as well as the discovery of new ideas, that allow this.
Thus many economists believe that in order to come out of the slump in productivity growth that has afflicted advanced economies in the past few decades, education is key.
Where these same economists are often led astray is in thinking that the solution lies in sending more people to college. The problem being that college has largely turned into a zero-sum competition for credentials, rather than being a place where people learn valuable skills and information that would allow them to succeed in the modern economy. Only a minority of college graduates get to use much of what they learned in college in their professional life.
Critics of this view often appeal to the second area where education is useful—the public good of an educated citizenry.
This is an important point often overlooked by detractors of the education system.
In theory, a zero-sum competition for credentials can be worth it even if it doesn't lead to the acquisition of skills useful on the labor market if it leads to the positive externality of creating citizens who will productively contribute to resolving problem areas that are left unsolved (or even caused) by free markets. Whether it is by voting, the choice of a career, public service, charitable giving... people have great power to contribute to human flourishing.
And so, an education that is not only job training is crucial for a well-functioning society.
But once again, looking through this lens, the current institutions of higher education perform quite poorly. College grads know little about public affairs, are taught even less about how they might have a positive impact on the world, and even the most prestigious universities—priding themselves on educating the next generation of leaders—seem more set on creating mid-level managers than an elite that can take on the responsibility of contributing to human progress.
No wonder trillion-dollar bills are regularly left on the sidewalk.
So if education is so important, why do EA orgs like 80k and GiveWell recommend against working to improve it?
Isn't what I have discussed above exactly what EA is all about?
I think the simple answer is that EA and the rationalist community are in fact working to fix education. It's just that there is an implicit understanding in EA and rationalist orgs that the institutions that are conventionally thought to be responsible for providing an education are too broken to be fixed.
After all, you don't get involved in EA or LW by eschewing learning. Quite the contrary. There are few communities that value their education more. One of the most up-voted posts on LessWrong is titled The Best Textbooks on Every Subject. One of the most up-voted posts on the EA Forum is a long post describing the academic literature on the importance of economic growth.
EA orgs don't think that improving education is an important cause because they are the ones providing the education necessary for people to understand how they can positively contribute to the thriving of our species.
In fact, many EA and rationalist organizations are effectively trying to compensate for some failure of the higher education system.
LW was created to teach people how to think more rationally, and to draw attention to an issue—AI alignment—that has been neglected in academic circles.
MIRI is doing research in AI alignment that isn't being done in universities.
RAISE and AI safety camp were also created to support people trying to contribute to technical AI safety.
CFAR pursues LW's mission to help people think more rationally and helps them to think strategically about how to achieve what they aim to.
80k is effectively "career services".
The EA Hotel allows people to self-study, do independent research, and it gives people the Slack they need.
Charity entrepreneurship helps EAs create high-impact charities.
A lot of other organizations also do research that is important but isn't being done within academia.
So, why not go all the way? For how much longer will young EAs and rationalists have to navigate the immoral maze that is the education system?
The current strategy that EA orgs seem to be pursuing is to imitate the success of neoliberalism. That is, gaining the respect of academic institutions, and thus increasing its influence among elite circles. But why should EA try to gain the respect of institutions that themselves don't deserve any?
Therefore, what I suggest is that EA starts aiming to Create a Full Alternative Stack. That is, replacing the current higher education institutions entirely. That would mean creating the space for learning and research that isn't subject to the traditional academic incentives. A space in which young people don't have to ask themselves what they have to do to get their degree, but one where their learning would be guided by the question of how they can become successful while contributing to human flourishing.
One of the reasons 80k lists in arguing against focusing on education is that the field is neither neglected nor tractable.
Before I go into how this issue might be tractable, and what concrete steps I recommend, I will explain why I think this issue is neglected.
It is true that hundreds of billions are spent every year on education in the US alone. However, the organizations spending this money show very little understanding of why the education system is broken, work from misguided assumptions (like the blank slate ideology), and demonstrate even less creativity in how they go about trying to fix it.
Fixing education will require a radical rethinking of how education is approached and what is taught. And the key to answering these questions lies in the Why of education. What is it for? This question is what EA is all about.
The reason this question is so important is that much of education is all about the motivation of the students. One of the main roles of professors is to provide students with incentives to learn. Homework, exams, grades, and the authoritative presence of a professor—to the extent that they enable learning—do so by helping the student overcome akrasia. The struggle that even if one actually wants to learn a subject, putting in the work necessary through sheer will power can be incredibly difficult.
I believe that by providing a student with the ultimate purpose of their education, some incredible gains in efficiencies are possible. To quote Neil Postman, there is a "fundamental simplicity [in] teaching and learning when both teacher and student share a reason for the enterprise."
No wonder EAs and rationalists tend to be lifelong learners. When your education might contribute to saving the world, a teacher is often not even necessary.
So, concretely, what do I suggest should be done?
One of the reasons I'm frustrated with how little effort has been put into fixing the education system by those who understand why it is broken is that I believe the solution lies to a large extent in the for-profit sector.
Alongside this post, I am publishing two others. In the first I describe my very positive experience self-studying ML at the EA Hotel. In the second, I describe my plan to start a for-profit project inspired by the EA Hotel, for people to study and do independent ML research.
This is because I believe that replacing the current institutions will require a two-pronged approach. First it will require the creation of learning communities where people can go and find peer support and accountability.
The second prong is the creation of comprehensive standardized exams to replace the signaling provided by a college degree (see What I Learned Dropping Out of High School).
I have encountered some skepticism about the potential of such exams to compete with college degrees, despite the fact that such exams have the potential to be up to two orders of magnitude cheaper than a college degree. But I find it easy to imagine how a Pinker degree of Psychology, a Marginal Revolution degree of economics, or a MIRI degree in AI Safety would successfully challenge the dominance of college degrees while also making a profit.
At the very least, creating such exams and degrees would demonstrate the dysfunctional nature of the higher education system in a more convincing way than Thiel fellowships.
The LW community has gotten a few things very right in the past decade. Seeing the potential of crypto was one, taking the threat of COVID-19 seriously earlier than most was another, and I believe that education will turn out to be a third. I just hope that, if I'm right, this community will take better advantage of this opportunity than it did with crypto.
Beyond that, existing EA orgs and initiatives such as AI safety camp, CFAR, and the EA Hotel can be given more funding and expanded. I think that, for example, a permanent AI safety camp would give young talented aspiring AI safety researchers the slack to self-study, do independent research, and prepare for a career in AI safety research. A MIRI AI safety degree (or whatever organization might create such a thing) would also provide those who fail to get a job in this field with something to show for their effort.
I'm a bit nervous about publishing this. After all, I am making two bold claims here: first that EA should change its macro strategy (i.e. stop following the neoliberal playbook), and second that there is a lot of wealth to be created by disrupting an education system that has proven immune to large changes for more than a century.
And so I understand if people reading this might have reservations. However, even though creating a full alternative to college might sound radical, I think none of the projects and initiatives I propose above seem to be that drastic or have a big downside risk. At the same time, the potential upsides are enormous. Not only do I think that there is a chance to create a great deal of wealth with for-profit initiatives, but there might also be the potential to influence new educational institutions, and expose a new generation of talented youth to EA and rationalist ideas.
One last thought. I don't want young people facing the decision of pursuing a college education to think that they should boycott the education system at their own expense. I'm still not sure if my own decision of dropping out of college was the right one. However, I did notice that in most EA and rationalist circles, at least, my decision and my write-up explaining my decision seem to be a stronger signal of competence than a B.S. degree.
And so, if you think there is a good chance that the current educational system will be exposed for what it really has become within the next decade, then it might be in your long-term self-interest to quit now. The first ones leaving a sinking ship might be frowned upon at first, but once the gaping holes in the hull become apparent to everyone, they will be praised for their foresight.
My hope is that this will turn out to be a self-fulfilling prophecy.
Probably the most important lesson I have learned from the EA and LW community is that individual actions matter. That we shouldn't take the world as it is for granted. That changing things for the better is possible. Moloch Hasn't Won. So let's not act as if the war has already been lost.
Thank you to everyone at CEELAR (EA Hotel) for their useful input. I'm especially grateful for the feedback from Kris Gulati, David Kristofferson, Rhys Southan and Michele Campolo.
A key claim in this post seems to be that the vast majority if not all EAs see higher education as broken and that they only go to signal their ability. Is there sufficient evidence for this claim? Maybe the claim is true, but the only evidence you provide is anecdotal and based on a single meet up.
Is it possible, even if that claim is true, that a minority of people really benefit from the higher education system and have a great impact on the world through it? Perhaps top academics? I genuinely don't know the answer to this question, but I think it's another important one to ask. If the answer is "yes" then perhaps the existence of the current higher education system is justified, but in a smaller form.
Thanks for writing this :) I certainly agree that the education system isn't optimal and maybe only useful to a handful of people. However, I'd like to provide myself as a data point of someone who actually thinks they benefit from their education. I'm worried that people might sometimes come away with the feeling that they're doing something wrong and pointless when going to uni/only doing signalling when that's not true in some cases.
I'm a bit of an outlier in that I'm actually in my second bachelor's degree and I definitely don't want to claim that that's a good idea for everyone. The first one was from a not well known University and my current degree is at a prestigious university. After my first year at my current university I was offered a job at an EA organization and after a lot of deliberation turned it down. I'm not sure that was the right choice but I still think I got a lot of benefits from continuing my degree. Here are some examples for why:
Admittedly, all of these reasons mainly apply to my second degree. I'm a lot more willing to relent that my first degree was mostly a waste of time, although I'm still often surprised by how much stuff I learned was actually useful (mostly stats). I also think the case is much different if you're not interested in research, or interested in e.g. ML engineering.
Hi Chi! I keep thinking about this:
If you have a moment, I’d be very interested to understand what exactly this tutor did right and how. Maybe others (like me) can emulate what they did! :-D
Hm, I'm not sure how easily it's reproducible/what exactly he did. I had to write essays on the topic every week and he absolutely destroyed my first essays. I think reading their essay is an exceptionally good way to find out how much the person in question misunderstands and I'm not sure how easily you can recreate this in conversation.
I guess the other thing was a combination of deep subject-matter expertise + [being very good at normal good things EAs would also do] + a willingness to assume that when I said something that didn't seem to make sense, it indeed didn't make sense, and telling me so/giving me all the possible objections to my argument; and then just feeling comfortable talking for 20 minutes (basically lecturing). I think that worked because of the formal tutor-student setting we were in and because he evidently and very obviously knew a lot more about the topic than me. I think it's harder in natural settings to realize that that's the case and confidently act on it.
What I mean by [normal good things EAs would also do]: Listening to my confused talking, paraphrasing what I was trying to say into the best steelman, making sure that that's what I meant before pointing out all the flaws.
Interesting, thank you! Assuming there are enough people who can do the “normal good things EAs would also do,” that leaves the problem that it’ll be expensive for enough people with the necessary difference in subject-matter expertise to devote time to tutoring.
I’m imagining a hierarchical system where the absolute experts on some topic (such as agent foundations or s-risks) set some time aside to tutor carefully junior researchers at their institute; those junior researchers tutor somewhat carefully selected amateur enthusiasts; and the the amateur enthusiasts tutor people who’ve signed up for (self-selected into) a local reading club on the topic. These tutors may need to be paid for this work to be able to invest the necessary time.
This is difficult if the field of research is new because then (1) there may be only a small number of experts with very little time to spare and no one else who comes close in expertise or (2) there may be not yet enough knowledge in the area to sustain three layers of tutors while still having a difference in expertise that allows for this mode of tutoring socially.
But whenever problem 2 occurs, the hierarchical scheme is just unnecessary. So only problem 1 in isolation remains unsolved.
Do you think that could work? Maybe this is something that’d be interesting for charity entrepreneurs to solve. :-)
What would also be interesting: (1) How much time do these tutors devote to each student per week? (2) Does one have to have exceptional didactic skills to become tutor or are these people only selected for their subject-matter expertise? (3) Was this particular tutor exceptional or are they all so good?
Maybe my whole idea is unrealistic because too few people could combine subject-matter expertise with didactic skill. Especially the skill of understanding a different, incomplete or inconsistent world model and then providing just the information that the person needs to improve it seems unusual.
Here are some resources about the case against the current education system for those not familiar with the arguments:
Bryan Caplan on 80k
The Case Against Education (LW)
The Case For Dropping Out of College (my best summary of the arguments)
I see you'd like to start a new organisation competing with universities, but I don't see you talking about any of the existing programmes or certifications that compete with universities. Have you looked into any of those? If so, how would your offer be different? And why did you write about your proposed new organisation on the EA Forum - what are you looking for?
I'm not sure what kinds of programmes or certifications you're thinking of, but as far as I know, if someone wants to learn maths, physics, economics, ML... or just almost any kind of academic subject, universities are literally the only option. There is no middle path between learning things completely on your own and jumping through all the higher education hoops. And even if a person manages to learn some subject up to a graduate level, there is no way to get it recognized.
The reason I posted this here is because I think there is an interesting contradiction in how EAs approach education for themselves (very important) vs at the institutional level (not worth improving). I do think that the goal of improving education should be taken more seriously by EAs. Though I might have underestimated the inferential distance between my understanding of the problems in the education system and that of the average EA.
Writing about my own proposed solutions wasn't really the main point of posting on the EA Forum. In this post I only mention it in one paragraph.
The main reason I crossposted The Best Educational Institutions in the World is because it is also a testimonial of the EA Hotel.
I think this take is interesting; I've given it an upvote.
I don't agree that the education system is as thoroughly broken as you say. For what it's worth, I've enjoyed my time studying CS at Cornell, and I'm starting my graduate studies there soon. Your mileage may vary - some schools do education well, others not so much. But I do think the institutions you're proposing can complement the existing education system by filling gaps in it.
Sorry, I downvoted this. I skimmed this, and it was hard to find the main points in the text. The claims that I did see written (such as EA already focuses implicitly on education) didn't feel convincing enough to delve in the details.
I appreciate you writing this though, and I'm interested in understanding the case for education and what might be the counter-arguments against the arguments against education. I think that for me it would help a ton if you would have made this much shorter and to the point, and make it very clear what claims you make - both from the start and at each point in the post.