Epistemic status: Completely schizo galaxy-brained theory
Lightcone[1] operates on a "generalist" philosophy. Most of our full-time staff have the title "generalist", and in any given year they work on a wide variety of tasks — from software development on the LessWrong codebase to fixing an overflowing toilet at Lighthaven, our 30,000 sq. ft. campus.
One of our core rules is that you should not delegate a task you don't know how to perform yourself. This is a very intense rule and has lots of implications about how we operate, so I've spent a lot of time watching people learn things they didn't previously know how to do.
My overall observation (and why we have the rule) is that smart people can learn almost anything. Across a wide range of tasks, most of the variance in performance is explained by general intelligence (foremost) and conscientiousness (secondmost), not expertise. Of course, if you compare yourself to someone who's done a task thousands of times you'll lag behind for a while — but people plateau surprisingly quickly. Having worked with experts across many industries, and having dabbled in the literature around skill transfer and training, there seems to be little difference within an industry between someone four years in and someone twenty years in, once you control for intelligence and conscientiousness.
But sometimes someone on my team does actually truly struggle to get better at a task, even if they are smart. Or I notice that if I were to try to get them to do something, they would have no idea how to even get started unless they spent at the very least multiple months, if not multiple years, acquiring the foundations necessary to do so.
So the question becomes: What determines whether someone is capable of relatively quickly acquiring expert-level performance in domains ranging from preparing a legal defense, to preparing an architectural plan, to physically renovating a bathroom, to programming a conference schedule app?
And my current, schizo galaxy-brained theory is that there are exactly 4 skills:
- Design skills: The ability to make good frontend design decisions, writing and explaining yourself well, designing a room, writing a good legal defense, knowing how to architect a complicated software system
- Technical skills: Follow and perform mathematical proofs, know how to program, make Fermi estimates, make solid analytic arguments, read and understand a paper in STEM, follow economic arguments, make a business plan, perform structural calculations for your architectural plans
- Management skills: Know how to hire people, know how to give employees feedback, generally manage people, navigate difficult organizational politics
- Physical skills: Be expert level at any sport, have the physical dexterity to renovate a room by yourself, know how to dance
If you are good at any task in any of those categories, you can become expert-level within 6 months at any other task in the same category.
Now why these exact 4 skills?
IDK, it kind of fits the data I've observed. But here is roughly how I came to believe what I believe:
First: across all tasks, performance correlates highly with general intelligence, and this dominates everything else. But clearly there's non-trivial variance left after controlling for it.
Then, there's an obvious divide between STEM and the humanities. Ask someone with a legal, history, or non-analytic-philosophy background to learn programming and mostly they bounce off or expect a multi-year training journey. Ask someone with a STEM degree to learn programming and it goes pretty well even if they've never programmed before.
Similarly, when I talk to people with a legal or humanities background and ask them about complicated frontend design decisions, they usually give surprisingly good input! They will pretty quickly jump into the fray of trying to model the user, figure out what a good product or information ontology, and have a sense of style about its presentation.
So that's it. There are exactly two skills. "Technical skills" and "Design Skills".
Then I tried to manage people. That... didn't go so well. Not only that, when I tried to get people on my team to manage other people, they also sucked at it!
So I learned that if I want to predict who will be good at management, I need to pay attention to whether they've managed other people before, and expect many months of practice until they are decent at it. Maybe it's a completely new cognitive domain, maybe it's just a domain where skill transfer is very hard and feedback loops are very slow and so it just takes everyone a while to learn the basic lessons, but nevertheless, if I want to predict performance at Lightcone, I gotta model people's management skills separately.
And then I tried to renovate a hotel.
And while the people on my team really ended up surprisingly good at a very wide range of tasks associated with construction and construction management, it also became clear that no one on my team would be able to perform the actual labor that our general contractors were able to perform. And also that they would totally smoke us in any sports competition. And that if I wanted to get someone on my team involved in the daily construction work, I sure expect that they would need many months of getting into shape and developing the right kind of physical skills.
So 4 skills it is.
Now, am I confident I have seen all skills there are in the world, such that no additional cluster will arise? Actually, yeah, kind of.
I have been walking through the world trying to keep track of what kind of career many of my acquaintances and colleagues go into for something like the last 2-3 years, and haven't really noticed any big holes. I have also been actively trying to think about careers that currently seem off limits to someone who has basic expertise in these 4 skill domains, and I have so far not been able to find something. My guess is if there is something I am missing it will be in something less career oriented.[2]
Need someone to build a script that automates filling out some business forms?
Give your econ masters student 3 months to learn programming and he can do it.
Need someone to drive your marketing push?
Give your interior designer 2 months to figure it out.
Need someone to head your internal legal department, double check the work of your lawyers, and prepare your legal defense in a high stakes trial?
Give your very smart frontend designer 3 months and they will go toe-to-toe with your lawyers.
Want to promote an engineer who has never managed anyone before to a manager?
Well, you better strap in for a year or more of pain while they acquire this completely new skill domain and traumatize all your new interns while doing so.
Want to get your backend engineer who is not good at writing, and is not good at interior design, to start taking more charge over your frontend?
Expect them to suck for at least a year until they can start competing with the smart designers on your team.
Want to get your quant finance guy who has never worked on a big codebase to start writing maintainable code and make nice clean Pull Requests?
Well tough luck, predict many months of telling them that yes, it is actually important that anyone can read your code and figure out how to modify these abstractions you've created.
Want to get your philosophy grad student dropout who has never done physical labor in his life to start managing construction projects and get their hands dirty?
Expect at least a year of getting into shape and used to the work, if they don't bounce off completely (though many subtasks of construction can be done with pretty little physical alacrity).
Give it a try yourself!
(Unhappy with any of my classifications? Fight me in the comments!)
Is there any externally validated or scientific basis for any of this?
Yes! It's not like, total consensus in the field of psychometrics, but task performance being extremely g-loaded across a wide variety of tasks is very well supported. People can really learn a very wide range of skills if they are smart.
And then within intelligence, math tilt and verbal tilt tend to be commonly used abstractions in psychometric testing that are predictive of success in careers in STEM or humanities.[3] Math fits nicely onto the technical domain. Verbal fits nicely onto the design domain.
A generalized "physical skill" factor is also well-supported. First, enough high profile athletes have switched from being world class in one sport to being world class in another sport such that there must be substantial skill transfer for these domains to explain that outlier success.[4] Second, somewhat unsurprisingly, if you measure people's sports skill you will find a strong "General Motor Ability" factor that explains performance across a wide range of motor skills.[5]
On management? IDK, that one I haven't seen much support for, but it sure matches my experience. There is an emotional intelligence literature, but that construct adds extremely little on top of just general intelligence. My guess is it's just a task that's very important and has terrible feedback loops, so everyone needs to fail for a while before they get good at it, but who knows.
Design. Technical. Management. Physical skills. Long ago, the four nations lived together in harmony. Then, everything changed when the Management Nation attacked. Only the True Generalist, master of all four elements, could stop them, but when the world needed him most, he vanished.
- ^
the organization I run, and which runs the website you're reading
- ^
If there is a missing cluster, I can imagine it being some more "relational" skillset around doing high-quality emotional labor, or maybe something genuinely associated with age and wisdom where certain skill domains are just really hard to perform well in without being at least 35+ and having the associated life experience. But I don't currently think such a cluster exists, and that four is really the right number.
- ^
- ^
Claude lists: "Bo Jackson (NFL All-Pro + MLB All-Star). Deion Sanders (NFL HOF + MLB). Rebecca Romero (Olympic silver in rowing, then Olympic gold in cycling four years later — different disciplines entirely). Clara Hughes (Olympic medals in both speed skating and cycling). Rugby → NFL is a well-trodden path (Jarryd Hayne, Christian Wade)"
- ^
This research is a bit more controversial than I expected, but I don't really understand the controversy. There are definitely some people in the field who insist on there not being a strong general motor ability factor. IMO this study also points in the direction of there being a general motor ability.
Discuss