Discussions Should Use Sources

People think you should debate or explain stuff yourself, not cite books or articles. But the truth doesn’t depend on what ideas are in my head or what I remember. So they aren’t using a truth-seeking approach.

The proper way to deal with complex topics is to look at what’s already been figured out. Don’t reinvent the wheel. Try to research and understand the state of the debate. If you find something missing you can make an original contribution, add something new, but mostly you don’t do that for well-developed topics.

E.g. take economics and political philosophy. That’s a big, hard topic. Even most of what Rand, Mises or Reisman said about it wasn’t new. Partly they organized existing ideas and figured out which ones are important and how they fit together. They did add in some new ideas too. Most people add a lot less than that. And it’s harder today after them. Anyway, the point of debate/discussion isn’t really to add new ideas to the field. If you have a new idea, write it down. Then when you debate someone, your idea is a book or article that you can cite. So it’s just like citing a Mises book, it’s pointing at the existing literature and trying to figure out what the discussion tree looks like, what is answered and unanswered, refuted or not, etc. If you have an original idea and write it down so it’s part of the literature, then the general project of evaluating the literature is going to include evaluating your idea. Nothing changes.

People object to citations in discussion to address some practical problems. They include:

• Citing stuff you don’t understand or haven’t read.
• Citing a million things to overwhelm someone.
• Citing a new thing every time one is refuted.
• Judging cites by the author’s name or what his conclusion is.

The main problems here are appeals to authority and argument quantity over quality. The proper way to use cites is only to cite your best material on each issue, and if it’s refuted you don’t cite second best then third best, you start reconsidering. If your evaluation of the best material (in your opinion) was incorrect, your evaluation of other material is also suspect. Your way of evaluating needs to be reconsidered.

Discussion should be kind of like a research project where you each help the other guy look through the literature for your side. If I talk with a socialist, he can tell me the key chapters in Marx that, as a Mises-advocate, are relevant to me and he thinks Mises didn’t answer. There’s a ton of socialist literature and a socialist is a good person to help guide me to the best stuff and also, simultaneously, to the key stuff to criticize (or cite criticism of) to change his mind.

This does depend on your goal in discussion. Are you trying to figure out what’s true? Are you acting the part of scholar, researcher, intellectual trying to reach some conclusions? Then don’t do a literature-independent discussion. Alternatively, you might be practicing talking about ideas and practicing debating. If you’re just trying to practice explaining stuff, not actually trying to reach a conclusion in the field, then using little or no literature can make sense.

People routinely mix these two things up. They debate like it’s unserious practice, won’t consider literature, but also think, at the same time, that they are reaching conclusions and this is a reasonable way to form their opinions. They think it’s a serious debate that can figure out the right economics while, at the same time, that they don’t want to read Mises, don’t know of any refutations of Mises by anyone, etc. You can’t figure out the truth of economics in that manner.

Rewriting published material doesn’t make sense in general. Books are carefully written and edited. What I say in a discussion is going to be lower quality (unless it’s about a position lacking good literature).

So for well-developed topics (like economics but not AGI), most of your comments should be about how the literature fits together, how it applies to particular cases that come up, stuff like that. Like if Mises wrote a general principle and a guy has a question that it answers, usually what Mises wrote is not a direct answer to that exact question. So I can write 3 sentences explaining how Mises’ principle relates to the question and then cite what Mises wrote. Those 3 sentences by me help customize the general purpose material to a particular case. Those kinds of sentences are generally missing from the literature, but they’re very important because people have particular questions and don’t always see how the one-size-fits-many general statements in the literature answer their questions. Even if they’re good at that, maybe they figure it out 9 out of 10 times, but it’s still a big deal if I can help relate the literature to the remaining cases.

And there’s a lot of literature, so a socialist might not know which Mises book to look in to get an answer to a particular question, and maybe I can find the answer and find some key quotes a lot faster than him because I’m more familiar with Mises’ writing. So that’s something useful I can contribute to discussion, it’s a way I can be helpful. And similarly, he can help point me at socialist literature that addresses some specific questions I have because he knows where to find that better than I do.

Cites also improve discussion by providing more targets to criticize. If I cite a Mises book, now you have plenty of details about my position that you can point out errors in. In literature-excluding discussions, people will bring up their ideas and never give you enough details about what they mean. They aren’t rigorous enough about explaining, piece by piece, how their claims work. They often change their position mid discussion. Literature is a fixed target that’s suitable for critical analysis.

And how do you get your own position, alone, on a complex topic like economics? You learn some parts of the field and, for other parts, you don’t investigate it beyond a summary level. You don’t have time to go into everything because it’s such a big field. Even professional economists specialize and can’t cover everything. For someone like me who has read a lot of economics but it’s not my specialty, it’s not even close – there are tons of issues where I believe it’s been covered by Austrian economics, and I could look it up if it came up, and I have some kinda summary info about it which makes sense to me and fits with other ideas and principles I have, but I haven’t carefully checked all the details. That’s how it’s gotta be. It takes many people working together and writing books and so on to develop all the complexity and detail that goes into a position in economics. The field’s standards are so high that it’s too much for one person to cover it all. You can understand the main principles as one person, you can think rationally, you can investigate areas you think may be problematic, you can investigate areas that discussion and debate partners bring up, but you can’t just go step by step through every last thing in the field, detail by detail, there’s been too much thinking about economics done. So to get a position I look for a body of work that I think gets stuff right. Ideally I find one I’m 99% happy with and my position can be “Austrian economics + X, Y and Z” and just make a few changes based on my own ideas (as long as the changes are isolated, that’s OK. If I want to change some major economic principle, it’d end up changing hundreds of conclusions, so it’d be a big issue.)

Less ideally (it’s more work), I might use the ideas of one school of thought for one big part of the field and another school of thought for another big part. That’s what I do in philosophy. I have Critical Rationalism for the majority of epistemology but not all of it, and I have Objectivism for some other parts of epistemology and for several other major areas of philosophy, and I also have David Deutsch for some other stuff like jumps to universality. To do this, one has to create more supplemental material explaining what’s used for what. It’s more complicated than just agreeing with one school of thought (even with some minor customizations). It’s still far less work than developing all the ideas from scratch.

Developing ideas from scratch is, in general, bad. It’s like rewriting software from scratch. You end up creating a bunch of new bugs. The existing stuff has been exposed to a lot of critical thinking. Many errors/bugs have been fixed. If you start over, you might think you’re fixing all the problems, like now you know what you’re doing and will get everything right, but what actually happens is making tons of mistakes including tons of mistakes that were already made and fixed in the past.

If the existing ideas are inadequate, in general you should help improve them instead of just ignoring them and trying to develop new ideas. This is especially true for complicated, established fields like economics or philosophy. It’s less true for a very new field like AGI, but even then you shouldn’t be e.g. reinventing algorithms, data structures, or programming languages – there’s lots of existing stuff that’s worth using (even an imperfect programming language is generally far better than trying to make a new one).

It’s kinda like existing human knowledge is a million points but has flaws, and if you help get it up to a million and 500 points, you improved things. But if you start over, you aren’t actually helping for the first 999,999 points, you’re still behind, so you have to do so much work before starting over is useful. Yeah maybe if you reinvent 100,000 points from scratch there will be a big chunk there someone could use and combine with some existing knowledge, but if that’s what’s going to happen you might as well do that yourself (develop in, from day 1, as an improvement on some existing knowledge – as something that can be added to some existing knowledge and/or some changes to some existing ideas with problems – rather than ignoring existing knowledge and leaving it to someone else to convert your work to be relevant to other ideas humanity has).

It’s hard enough to work with existing knowledge and improve it. Most attempts actually make it worse. It’s hard to understand how existing knowledge works, what the problems are, and how you can make changes without breaking things. It’s much worse, though, to just take the field itself and try to solve it yourself without all the help and guidance of existing knowledge. Then you’re trying to outcompete thousands or millions of people’s cooperative efforts by yourself.

Most people trying to build up intellectual systems from scratch don’t know much about the literature. It’s related to the cliche that you need to know the rules (e.g. of English) before breaking them. If you don’t understand what’s already known and what’s good about it, you aren’t in a position to do things differently and do a good job. But once you do understand the literature well, and get a good grasp on what’s already known, then you’re in a good position to improve it. A lot of why people want to start over, instead of adding to existing knowledge, is specifically to skip the step of learning much about what’s already known. They’ll never accomplish much.