One of the most useful aspects of Less Wrong is that you can read the Sequences and then start posting on the forum, engaging intelligent people in discussion and trying to apply what you learned to important topics. Far from Less Wrong being without exercises, this seems like one of the best kinds of exercises.

And you've got to be at least at philosophy 75% to even start reading a lot of the material here

I disagree. I came in with essentially 0% philosophy skills, I didn't even know what consequentialism or utilitarianism meant. And I was able to understand the sequences, and don't have a problem with the posts now.

By the way, there is one benefit of a systematically fed knowledge -- it prevents you from getting into selection spiral, where you want to learn about something, find the first piece of knowledge belonging to some small subset of what you want to learn, become impressed and learn the whole subset... mistakenly believing that you actually learned the whole set.

For example, one can learn programming in Java and believe that "programming in Java" equals "programming", and there is nothing useful to be learned outside. Or one can learn behaviorism and believe that behaviorism is the only and the complete answer to psychology. Or learn one economical theory, or learn one political view, or...

In such situations, the best part of the systematically fed knowledge is that it usually starts with a roadmap: it shows you a list of subsets. That's the best part... and probably from that point on you could again be better by learning those subsets on your own. (Supposing you are able to update about the existence and importance of the other subsets; many people aren't.)

For the things I believed to be important, I leaned more on my own than in school. But the school taught me a few things I didn't consider important... until I learned them. (And also many things I still consider unimportant.)

Why is this specific to autodidacticism? I went through school and college without ever learning anything about the valuable content here on LW.

If you need structure in your life (e.g., forced to wake up early 5x/week), or you need motivation (e.g., want to learn math but keep procrastinating), or something like that,

Or you want someone to give you feedback on your homework and answer your questions (and hence a chance to clear up your misconceptions), or you want hands-on experience with lab equipment, or you want to socialize with other people interested in the same things, or you want to be able to legally access papers behind a paywall for free, or...

It's low status to say one could benefit from external motivation and structure, but I suspect that for the vast majority of people and general, and also the majority of LWers, external motivation and structure are hugely valuable.

Tests are valuable to give you an idea of how you're doing with learning stuff. IIRC, some education research suggests that spending an hour taking a test is actually substantially better for your understanding and long-term recall than spending an hour studying.

I think if you wanted to make a serious "alternative technical college education" for yourself, one way to do it would be to work on a series of technical projects that interested and motivated you, then learn whatever was necessary to complete the projects well (see this post by Robin Hanson). Probably find some local hackerspace so you don't have to buy equipment, you get some social pressure/networking, and you can pick up some tacit knowledge from other hackers. If your projects were impressive, they could look good on your resume to the right employer (open-minded small business owner/startup founder?)

This strategy could be difficult because most people don't have the balls to say "I am going to complete project X in a field that I know nothing about and learn whatever I need to know in order to complete the project on the fly". You'd run a risk of getting intimidated and giving up, I think. A certain amount of confidence in your ability to understand anything given sufficient time, effort, and pestering of knowledgeable people (online or offline) would probably be necessary to make this work (relevant rationality quote). I still think going to college is a better idea for most people.

I've learned more about physics from school than I would have learned on my own, and I think your comment is pessimistic. University has advantages over the internet, even if your goal is simply to learn material.

What sealed my decision to spend my last two years of high school at the state Math and Science school instead of the same public high school I'd been attending was the realization that I was learning more from the internet than I was in classes, with the exception of a few chemistry concepts that I never thought of looking up.

When I graduated, I realized that even the Math and Science school didn't catch up with what I'd learned from the internet until the final semester. I'd still say it was worth it, and I decidedly did not optimize (mostly because I was still thinking of science more as "stuff scientists discovered" than "an insanely useful method that you should pay attention to outside of science fairs"), but college? Not so much. There were benefits, sure (I think the main one Crux left out that I completely failed to even try at is networking with professionals in a broader sense than work experience), but the most important things I learned while at college were things that came entirely from the internet or experiences on holidays. There were quite a few things I stubbornly tried not to believe (mostly things about human psychology) that were thoroughly demolished by the internet (especially when I finally got to LessWrong, but by then it was a bit late and I was already in the "No, I'm not doing this anymore" phase of college.).

In spite of all of this, I'm still frustrated with my timing; college really could have been an outstanding opportunity had I, say, read the sequences a year earlier. Attending college is an incredibly easy way to access resources like labs, equipment, experts and cheap labor (especially if you can turn a project into academic credit, which I totally could have at my college; I would have gotten credit for this terrible virtual series had I just finished the blasted paperwork. (*mumbles something about a RATIONAL! rewrite*)). For example, when it manifested that Senseg might have been overly optimistic with their predictions on getting their tactile technology on shelves early in 2013, I found myself frustrated that the technology is so simple that I could toss together a simple example over an afternoon if I had any of the resources I had since my junior year of high school, but I wound up not learning about this until the cost of going back was way too large for way too small a benefit and my resources had largely dried up.

So I absolutely agree: you can get a college education by only taking courses with Dr. Google. Other professional goals, skills, or structure-based needs might be satisfied by college, but there's no point in spending thousands of dollars just for the data in the courses.