Enjoy solving "impossible" problems? Group project!

. I am not going to sit around waiting for an opportunity to become stronger. I'm going to seek them out.

That's fine, but attempting to bench press your car is not the most effective way to increase your bench press. You don't try to lift an impossible weight, you select a possible weight that stresses your capabilities.

Also, I think the "I want to get stronger" ethos is taken in terms of incremental improvement, not in terms of "I want to be all powerful today".

Thinking about an impossible or merely very difficult problem, because you think that putting forth effort on it will make you stronger, is very different from what Eliezer is talking about.

Ask yourself this: if you spend time working on one of the problems from this thread, and in the process become stronger and learn something, and eventually give up to work on something else, will your reaction be more like "I have failed" or "at least I learned something while failing"? If the latter, Eliezer's post is not relevant to you, and your attempts are not in its spirit.

What I'm seeing is that a rational agreement software would require some kind of objective method for marking logical fallacies, which the logic checking AI would obviously be helpful with. Not sure why the rationalist agreement database would help with creating the logic checking AI, unless you mean it can act like a sort of "wizard" where you can go through your document with it one piece at a time and have a sort of "chat" with it about what the rationalist agreement database contains, fed to you in little carefully selected bits.

Every square is a rhombus

If the difference between a hasty generalization and a fact is that the fact is true, then to call something a hasty generalization we need to say something about its factualness.

Not all claims of the form "all Xs are Ys" are false, and neither is every conclusion of the form "all Xs are Ys" a product of bad reasoning. Suppose your software were to replace "All electrons are negatively charged" with "All rabbits are highly educated". How is the reviewer supposed to react? Is she supposed to conclude that the original statement is false? Why?

You are using the phrase "hasty generalization" in a highly non-standard way here. Philosophers classify hasty generalization as an informal fallacy. The "informal" means that the content matters, not just the formal structure of the argument. Also, a hasty generalization is an argument, not a single sentence. An example of a hasty generalization would be, "I was cut off by an Inuit driver on the way to work today. All Inuits must be terrible drivers." The fallacy is that the evidence being appealed to (being cut off by one Inuit driver) involves much too small a sample for a reliable generalization to an entire population. But just looking at the formal structure of the argument isn't going to tell you this.

There are formal fallacies, where the fallacy in the reasoning is subject-independent. An example would be "affirming the consequent" -- arguing that since B is true and A implies B, A must also be true. You could build the kind of software you envisage for formal fallacies, but you'd need another strategy for catching and dealing with informal fallacies.

I doubt that even one of these ideas will be pursued to the point the effort pays out (if I had to give you a probability, I'd say p=30%).

I understand placing a low prior on ideologues and pop social sciences in general. I don't believe Diamond should be considered either of those, though. I've read Guns, Germs, and Steel and most of Collapse, and I haven't really seen any attempts by him to sweep any problems under the rug. He didn't seem to be oversimplifying things, to me, when I read him.

Could you recommend a criticism of Diamond's material to me?

I don't necessarily disagree with the idea that people invent ideas, but generally I think that having help with something is better than trying to do it all yourself. I think having two different extremes interact and debate each other would probably yield more interesting truths than having one very smart moderate attempt to discover the truth on their own, provided that those extremes are being evaluated by a neutral and intelligent third party and those extremes are trying to court the third party to their point of view.

I think Taleb's quote is more about how all people attempt to predict the actions of all other people, and then act accordingly. Lots of behavior in the social sciences functions like a super extreme version of a Keynesian Beauty Contest where all participants are both judges and players, there is no limit to how meta or recursive you go except whatever limit is imposed on you by your cognitive limits, and you have access to information that isn't just incomplete, but is often actually wrong.

It's not physically impossible, like simulating a complicated simulation that simulates itself (and is thus bigger than itself which is basically a contradiction) but it's in a similar sort of vein and degree of difficulty. Very meta and recursive, very difficult to measure things or to verify if what you're doing is correct or if you even understand what you're doing.

The goal is vague because I don't know how to get started with it.

I'm not quite sure what you're saying with the rest of your comment. I understand that economics and foreign policy are basically two different areas. However, the policies of both fields interact quite a lot, and both disciplines use many of the same tools, such as games theory and statistical analysis. I would perhaps even argue that IR studies would be improved overall if they were widely conceived of as a sub discipline of economics. They also share many of the same problems.

For example, in both fields there are large difficulties with comparing the results of economic and foreign policies and comparing them to the results that other policies counterfactually would have had, because countries are radically different in one time period as compared to another, and because policies themselves are more or less appropriate for some countries than others. Figuring out how to apply the lessons of one time and place to another is more or less what I was envisioning when I said that I wanted to make the social sciences more empirical.

There are also problems with measuring variables in both fields. In science, it's relatively easy to determine what the output amount of energy from a system is, or the velocity of a specific object at a specific time. But in economics and IR, we have lots of trouble even understanding exactly what the inputs and outputs are or would be, let alone understanding their relationship with one another. For example, uncertainty is hugely important in IR and in economics, but it seems almost impossible to measure. Even more obvious things, like the number of troops in a certain country or the number of jobs in a specific sector, are often debated intensely by people within these fields.

Without the ability to measure inputs or outputs of policy processes or the ability to compare those processes to the hypothetical effectiveness that other policies might have had, these fields are crippled. If there is any way to get around these problems or to minimize them, we really need to figure it out. This will be really really hard, if not impossible, but it's probably the most effective nonscientific thing that we can be doing to minimize existential risk.

TL;DR: I want to be Harry Seldon except in real life.

So we can see your vision, please describe how this would work?

I'd say that legal language, at least in America, is absolutely well within the bounds of natural language, with all the ambiguity that implies. Certainly lawyers have their own jargon and "terms of art" that sound unfamiliar to the uninitiated, but so do airplane pilots and sailors and auto mechanics. It's still not mathematics.

There are a lot of legislators and judges, and they don't all use words in exactly the same ways. Over time, the processes of binding precedent and legal authority are supposed to resolve the inconsistencies within the law, but the change is slow. In the meantime, statutes keep on changing, and human beings keep on presenting courts with new and unexpected problems. And judges and legislatures are only people within a society and culture which itself changes. Our ideas about "moral turpitude" and "public policy" and what a "reasonable man" (or person) would do are subject to change over time. In this way, the language of the law is like a leaky boat that is being bailed out by the crew. It's not a closed system.

One bottleneck here would be that the programmer would also have to be able to understand legalese. To find someone with both specialties could be pretty hard.

I honestly don't know enough about law to provide the kind of detailed mistake you're looking for. My belief that it is a somewhat 'important' problem is circumstantial, but I think there's definitely gain to be had:

1) It is often said that bad law consistently applied is better than good law inconsistently applied, but all other things being equal, good law is better than bad law. It is generally accepted that it is possible to have 'good' law which is better than 'bad' law, and I take this as evidence that it's at least possible to have good law and bad law.

2) Law is currently pretty ambiguous, at least compared to software. These ambiguities are typically resolved at run time, by the court system. If we can resolve some of these ambiguities earlier with automated software, it may be possible to reduce the run time overhead of court cases.

3) Law is written in an internally inconsistent language. The words are natural language words, and do not have well understood, well defined meanings in all cases. A checker could plausibly identify and construct a dictionary of the most consistent words and definitions, and perhaps encourage new law makers to either use better words, define undefined words, or to clarify the meaning of questionable passages. By reducing even a subset of words to a well defined, consistent definition, the law may become easier to read, understand and apply.

4) An automated system could possibly reduce the body of law in general by eliminating redundancy, overlapping logic, and obsolete/unreferenced sections.

Currently, we do all of the above anyway, but we use humans and human brains to do it, and we allow for human error by having huge amounts of redundancy and failsafe. The idea that we could liberate even some of those minds to work on harder problems is appealing to me.

Sorry. I didn't realize you really want to do the impossible. Good luck!

I thought you want to do something "nearly impossible", what a big money generator is. If you don't agree, if you think that's easy, set the constrains high enough. Like 10^12 USD of profit in 5 years. For example.

I'm not sure this it would require a full understanding of natural language. There's got to be an 80/20 rule method by which this can be done. Really, there are only so many logical fallacies, and there might be some way to "hack" out fallacious statements by looking for certain patterns in the sentence structure, as opposed to actual sentence interpretation.

For example: "All gerbils are purple." The computer only needs to understand: "All (gibberish) are (different gibberish)." Hasty generalization pattern recognized.

For another example: Gerbils are purple because purple is the color of gerbils. The computer understands: "(gibberish 1) are (gibberish 2) because (gibberish 2) is (blah blah blah) (gibberish 1)" Circular reasoning pattern recognized.

Yes, it would get more complicated than that, especially when people use complex or run-on sentences, or when the fallacy occurs after numerous sentences stack together. But, I still think it could do this with pattern recognition.

Hmmm... it would also have to detect statements where points are being made (looking for the words "is", "are" and "because" might help) and avoid sentences that are pure matters of opinion (I love ice cream because it's delicious! - this might look something like (blah blah blah) because it's (subjective term)).

I somehow doubt Google would appreciate the leeching of their resources - unless you mean they've made it open source or something. Making it dependent on them would be a liability - if they notice the leeching of their resources, they'd surely create a new limit that would probably break the program.

Sorry for sounding snarky, I mean this in a friendly way: which part of "Impossible Project Ideas" don't you understand? For that matter, contrast the last three words of my comment with the last word of yours ;-)

Also, if you want to get all technical, we do have technology necessary to revive cryogenized beings, just not for any kind of being. (Hint: for some kinds of cryogenized being the level of technology necessary for reviving is just "drop the ice cube in a puddle of rain water".)

Several problems:

Correlation is not causation. I'd have to see evidence that high testosterone was needed for trauma surgeons to be effective before I'd accept that it was a necessity.

What percentage of traumas are caused by violence? If excess testosterone were treated would the number of traumas decrease as well, making it unnecessary to have as many trauma surgeons?

As for whether I've considered what side effects would occur, no actually, I haven't. That was a good reminder. This isn't an idea I've thought about a lot yet so I haven't gotten very far. Up until this point, I'd been thinking about it like a disease - you don't justify failing to treat a disease by worrying about what society will be missing when those people are healthier.

Though, you could still wonder what might happen, sometimes consequences are unexpected. I don't know that much about testosterone. Do you have suggestions?