I think I’m prone to falling into demon threads. Some of the attributes feel very familiar (such as consumed energy, ever growing rudeness, terrible miscommunication, eventual lack of progress) but some feel quite foreign (such as everything in your text involving the substring “soc” or more than two people). Apparently we’ve all been in such threads, but I wonder if we’re all talking about the same things. I wonder if there is meaningful classification of different kinds of demon threads. I wish people actually shared the daemon threads they’ve ran into so we could talk about them, what was done right and what was done wrong. I’m not going to start though.
In the past I just assumed that demon threads are terrible, because the kind of people who participate in demon threads are terrible (note the self deprecation). Recently, however, I had a different idea. When you see people terribly miscommunicating in a demon thread, it might at first seem that they’re morons, but it’s possible that what you see is the true limit of the human ability to understand each other. In this view the reason most other threads aren’t demon threads is because they are too short for the participants to discover how poorly they are really understood, the participants are too dumb or too charitable to realize they’ve been misunderstood, or they care too little to correct the misunderstanding. In this view, a demon thread is merely the most likely outcome for any discussion where all parties are willing to put in some effort (at least the emotional kind) and not give up, and where the disagreement is not completely trivial.
There is a solution to this problem I can see in my dreams—a perverse dedication to formal reasoning. There can be no disagreements and wild claims when the arguments can be formally verified. There can be no dick measuring if ideas are verified one at a time, and admitted into the Long List of True Statements, instead of fighting with other ideas one-on-one. And if the idea can’t be formalized, perhaps it’s not worth thinking about. Surely I’m not the only one?
3 seems like it would sneak unjustified premises into everything, and make it prohibitively expensive to challenge them. Philosophers already tried this, and all we got was analytic philosophy, which is not very interesting, and still doesn’t do anything to solve problems of emphasis, which are another way in which words can be wrong (your schema might treat rare events as central cases and common ones as exceptions, for instance).
Obviously, I wouldn’t know how to do formal reasoning correctly, even if I seriously tried to do it. I’m sure there are many problems with the idea that don’t have known solutions. I believe that complete and correct formal reasoning is easier than full AI, but not by much. Having that in mind, it’s hard to make claims about what this reasoning would look like.
I’m not sure what you mean by unjustified premises and problems of emphasis, so I’ll make a guess. You might worry that some people would dedicate a lot of time and effort into constructing increasingly convoluted proofs showing how, e.g. flat earth, is consistent with various observations and experiments. Such proofs might be admitted into the Long List of True Statements. However, as long as these proofs lead to no implications about what NASA should be working on, they are not a problem. Another possibility is that the proofs are of the form “if lizard people run NASA, then it’s most likely that the earth is flat”. Again, if you don’t share the assumptions, there is no harm from such proofs, they might even be beneficial in some ways (e.g. displaying our sensitivity to bad priors). In this framework, building perverse proofs for “the outgroup is stupid” might actually be a productive activity.
Well, that’s a long list, but I don’t see why formal logic would make any of those problems worse, and it seems many could be solved. Do you have some specific worries?
Social-related-stuff isn’t the only cause of demon threads, it just is a more-reliable-way-than-average to cause them. (I think it’s possible that what I’d call malignant demon threads require some element of social stakes, even if it’s just two people devolving into personal attacks)
I have vague plans (I’m not sure if this will turn out to be a terrible idea), about attempting to have a followup post that walks through specific examples while treading very carefully to avoid re-opening the original threads in question.
while treading very carefully to avoid re-opening the original threads in question
Is that even worth worrying about? One one hand this seems to assume very little faith in LW users, and on another hand, if it did reignite conflict, would that really be bad?
I think I’m prone to falling into demon threads. Some of the attributes feel very familiar (such as consumed energy, ever growing rudeness, terrible miscommunication, eventual lack of progress) but some feel quite foreign (such as everything in your text involving the substring “soc” or more than two people). Apparently we’ve all been in such threads, but I wonder if we’re all talking about the same things. I wonder if there is meaningful classification of different kinds of demon threads. I wish people actually shared the daemon threads they’ve ran into so we could talk about them, what was done right and what was done wrong. I’m not going to start though.
In the past I just assumed that demon threads are terrible, because the kind of people who participate in demon threads are terrible (note the self deprecation). Recently, however, I had a different idea. When you see people terribly miscommunicating in a demon thread, it might at first seem that they’re morons, but it’s possible that what you see is the true limit of the human ability to understand each other. In this view the reason most other threads aren’t demon threads is because they are too short for the participants to discover how poorly they are really understood, the participants are too dumb or too charitable to realize they’ve been misunderstood, or they care too little to correct the misunderstanding. In this view, a demon thread is merely the most likely outcome for any discussion where all parties are willing to put in some effort (at least the emotional kind) and not give up, and where the disagreement is not completely trivial.
There is a solution to this problem I can see in my dreams—a perverse dedication to formal reasoning. There can be no disagreements and wild claims when the arguments can be formally verified. There can be no dick measuring if ideas are verified one at a time, and admitted into the Long List of True Statements, instead of fighting with other ideas one-on-one. And if the idea can’t be formalized, perhaps it’s not worth thinking about. Surely I’m not the only one?
3 seems like it would sneak unjustified premises into everything, and make it prohibitively expensive to challenge them. Philosophers already tried this, and all we got was analytic philosophy, which is not very interesting, and still doesn’t do anything to solve problems of emphasis, which are another way in which words can be wrong (your schema might treat rare events as central cases and common ones as exceptions, for instance).
Obviously, I wouldn’t know how to do formal reasoning correctly, even if I seriously tried to do it. I’m sure there are many problems with the idea that don’t have known solutions. I believe that complete and correct formal reasoning is easier than full AI, but not by much. Having that in mind, it’s hard to make claims about what this reasoning would look like.
I’m not sure what you mean by unjustified premises and problems of emphasis, so I’ll make a guess. You might worry that some people would dedicate a lot of time and effort into constructing increasingly convoluted proofs showing how, e.g. flat earth, is consistent with various observations and experiments. Such proofs might be admitted into the Long List of True Statements. However, as long as these proofs lead to no implications about what NASA should be working on, they are not a problem. Another possibility is that the proofs are of the form “if lizard people run NASA, then it’s most likely that the earth is flat”. Again, if you don’t share the assumptions, there is no harm from such proofs, they might even be beneficial in some ways (e.g. displaying our sensitivity to bad priors). In this framework, building perverse proofs for “the outgroup is stupid” might actually be a productive activity.
I’m worried about what happens before people start putting time and effort into proofs.
Related: 37 Ways That Words Can Be Wrong
Well, that’s a long list, but I don’t see why formal logic would make any of those problems worse, and it seems many could be solved. Do you have some specific worries?
Social-related-stuff isn’t the only cause of demon threads, it just is a more-reliable-way-than-average to cause them. (I think it’s possible that what I’d call malignant demon threads require some element of social stakes, even if it’s just two people devolving into personal attacks)
I have vague plans (I’m not sure if this will turn out to be a terrible idea), about attempting to have a followup post that walks through specific examples while treading very carefully to avoid re-opening the original threads in question.
Is that even worth worrying about? One one hand this seems to assume very little faith in LW users, and on another hand, if it did reignite conflict, would that really be bad?