deepthoughtlife

Karma: 71

deepthoughtlife 14 Aug 2022 11:40 UTC
1 point
0 ∶ 0
in reply to: Morpheus’s comment on: Complexity No Bar to AI (Or, why Computational Complexity doesn’t matter for real life problems)
I feel I should mention an important implementation detail. If you do not know the range of possible values, it is often a good idea to check, as most generative processes only use a subset. This is also a linear time process, that only takes two values in memory, but does involve some branching. It is better to just know and put that in, but the ability to do this makes it a much more practical algorithm.

deepthoughtlife 12 Aug 2022 17:45 UTC
3 points
2 ∶ 4
on: Oversight Misses 100% of Thoughts The AI Does Not Think
This has far too many assumptions, and the final claims are too strongly stated. One of the important assumptions being that it can trivially destroy humanity without trying. Comparing humanity to a specific kind of flower is ridiculous. Even a strongly superhuman AI would be more like humans vs rats than humans vs. a flower. Could humanity eliminate all rats in the world if we wanted to? Maybe, but we wouldn’t accomplish much of anything else.
Another assumption is that whatever is overseeing the AI is vastly stupider than the AI, even with all the tools made for the purpose. If you can make a superhuman AI, you can make a superhuman AI and Human system too (which is an easier task).

deepthoughtlife 11 Aug 2022 1:53 UTC
2 points
1 ∶ 0
on: On Ego, Reincarnation, Consciousness and The Universe
Blank-slatism is so obviously wrong it is hard to see how you even got there. Even as babies people have completely different personalities and approaches even though all they are doing is lying there. When a baby is adopted by a completely different kind of family than the one they were born into, they bear remarkable similarities to the original family that they don’t to the adoptive in some areas. Older people always react differently than each other to identical approaches too (though you could argue path-dependence there.). Genetic factors are well established as making a difference to many lasting traits that clearly change how people interact with the world, including things such as height and other physical attributes, extraversion and other personality types, and things like IQ or other talents. There are reasons besides genetics that show blank-slatism is wrong, but those thoughts are less self-evident. I am not, and could not, be exactly the same as others (though I could have been, and could yet be, very different than I am now.)
Panpsychism is an obvious non-starter too. It is clear that consciousness is an attribute of things that can think, because consciousness is a certain kind of self-reflective thinking. Thus, it should not be attributed to anything unlikely to be thinking in a sophisticated manner. I doubt consciousness for animals in general, but at the very least, it is certain that only animals and up even have the chance. (Perhaps you might be able to make a conscious AI sometime in the future [I don’t have a strong position on it], but that wouldn’t make the computers it was running on conscious.)
Your claim that being used as raw materials later is somehow reincarnation falls apart the moment your panpsychism does.

deepthoughtlife 9 Aug 2022 14:58 UTC
1 point
0 ∶ 0
in reply to: Tom Shlomi’s comment on: Complexity No Bar to AI (Or, why Computational Complexity doesn’t matter for real life problems)
Morpheus is incorrect, I am actually referring to a generalization of ‘counting sort’ that I independently reinvented (where you are no longer limited to small positive numbers). I had never heard of sub n log n sorts at that time either. It’s actually quite simple. In fact, it’s how humans prefer to sort things. If you look at the object, and just know where it goes, put it there. The downside is just that there has to be a known place to put it. The O(n log n) lower bound is only for comparison sorts.
On a computer, we can just make a place to put it. If each value is unique, or we’re sorting numbers, or each value that is the same does not need further sorting, we can implement this in constant time per object being sorted. Also linear space, with space being O(m + n) where n is the number of objects being sorted and m being the range of values they can hold. This only works well if m is reasonable in size, not being too much larger than n. For instance, you might be sorting integers between 0 and 100, requiring only a hundred and one extra spots in memory, and be sorting 1,000,000,000 of them. (This is conceptually related to the bucket sort mentioned by Morpheus, but they are distinct, and this is truly linear).
Worst case is usually that you are sorting things whose values are an unknown 32bit integer, in which case you would need about 16GB of memory just for the possible values. This is normally unjustifiable, but if you were sorting a trillion numbers (and had a large amount or RAM), it’s nothing. (And the fact that it is linear would be quite a godsend.) Another positive is that this sort has virtually no branching, and is this much easier on a per instruction basis too. (Branching kills performance).
This can also be integrated as a sub-routine (that requires very little memory) in an overall sorting routine similar to (but better than) quicksort, and then it is more similar to bucket sort, but kind of reversed.

deepthoughtlife 8 Aug 2022 18:34 UTC
1 point
1 ∶ 1
in reply to: Tom Shlomi’s comment on: Complexity No Bar to AI (Or, why Computational Complexity doesn’t matter for real life problems)
Obviously, if computational complexity really forbid general intelligence...then we wouldn’t have it.
Technically, list sorting is O(n) if done a certain way...we just do it the O(n log n) way because it uses slightly less memory and is simpler for us (which is kind of related to Gwern’s rant, and he does technically mention a specialized one, though he doesn’t realize what it means, and puts it in a section about giving up generality for speed). I actually invented an example of that type of algorithm when I was teaching myself c++ using sorting algorithms and wanted to beat the built-in algorithms. In practice, sorting algorithms are often multiple different algorithms of different complexity classes mashed together...that then end up still being O(n log n) with better constant factors. We already optimize the hell out of constant factors for important problems.
I agree with you rather than Gwern. In my experience, complexity class really is a good indicator for how much compute something takes. It’s not a perfect proxy, and can definitely be goodharted, but it is still effective. Often, the constant factors are closely related to the complexity class too. Even describing an ‘NP’ problem is vastly more difficult than a ‘P’ problem, because there is just so much more to it.
I would claim that difficulty describing the process is actually a good proxy for how large the constant factors are likely to be, since they are what you have to keep in mind even for problem size 0, where we aren’t even trying to solve the issue, just keep it in mind.

deepthoughtlife 8 Aug 2022 17:56 UTC
3 points
0 ∶ 0
on: How Deadly Will Roughly-Human-Level AGI Be?
You aren’t really assuming human level AGI. You are assuming human level AGI that runs as much faster than a human as a computer is at doing arithmetic, which is a completely different sort of thing. It’s basically, ‘assuming we have an AI that is equivalent to all of humanity put together, but with perfect coordination’ which is literally a superintelligence. There is no reason to make that leap.
Computing technology could possibly top out at, human intelligence but fifty times slower, for a few decades after its invention. Clarity in knowing what your assumptions are is crucial when evaluating thought experiments. There is no reason to believe ‘ems’ would actually run fast (nor, necessarily, that they wouldn’t.)
For the record, once actual human level intelligence is believed close, I expect someone will spend a crazy amount of resources to get that human level AGI trained long before it could possibly run at decent speed. This may happen several times before real human level is reached.

deepthoughtlife 8 Aug 2022 17:44 UTC
1 point
0 ∶ 0
on: Steganography in Chain of Thought Reasoning
It seems easier to check for ‘steganography’ by simply comparing it doing chain of thought unimpeded versus having its steps rewritten to be superficially different by a person, and then going from there. For a math based system, you could do it with a ‘dumb’ program. Once you’ve established its existence, it seems like planning how to test it would be simpler at that point.

deepthoughtlife 8 Aug 2022 17:31 UTC
2 points
0 ∶ 0
in reply to: MSRayne’s comment on: The lessons of Xanadu
Oh, but that last one could be fun...
The obvious thing to do, of course, is to figure out why you are trying to design one thing that is all those things, and see if a simpler thing to do becomes obvious at that point. The reason could simply be that you like to plan complicated things, in which case settle on a problem that is genuinely complicated and doesn’t need any complications added, or it could be because your interests change rapidly, in which case you focus on a simple but expansive framework so that you can hit the ground running on whatever your new interest is, and achieve something (even just a beautiful thought) in that area before your interests change again. More interesting is if there is some commonality between all the things you switch between, and you can just focus on that part, because that is where your true interest lies.
Obvious, boring but surprisingly difficult to follow advice. Practice taking notes, organizing things, building working projects (which could be as simple as fixing a door that doesn’t close quite right, or a floorboard that squeaks, or as complicated as learning a new programming language to write a short program that does one thing.), and so on. Planning things successfully is a big deal for people that currently don’t. (I should take my own advice...)

deepthoughtlife 8 Aug 2022 17:15 UTC
2 points
0 ∶ 0
on: The lessons of Xanadu
I can’t agree with your proposals for what to learn from this failure, which honestly read like cached thoughts to me. The actual issues outlined are hubris, which is perhaps the oldest known human bias, and an immense focus on centralized coordination with all the immense failures that has caused throughout human history not being solvable by fiat.
The lessons aren’t ‘evolve’ and ‘ship’. They are ‘pick an achievable goal’ and ‘let the people coordinate amongst themselves’. The former are simply methods aimed at achieving ‘pick an achievable goal’. Sometimes they are the right methods, and other times, they are not. The achievable goal of the web: ‘follow links’. The web is also obviously not centralized (no matter what the people complaining about social media tell you.). Local centers of coordination will pop up organically, without having to solve the hard problem of planning them in totality.
We tend to ‘learn’ what we want to learn, and that mix of hubris and lack of new thought makes it hard to learn when we fail due to our hubris.

deepthoughtlife 8 Aug 2022 16:36 UTC
1 point
1 ∶ 0
in reply to: Noosphere89’s comment on: chinchilla’s wild implications
You kind of can be informal though?
Suppose, 5x − 2 = 3b +9, thus
x = (3b + 11)/5 or b = (5x −11)/3
If b = 2, then
x = ¹⁷⁄₅
If x = 2, then
b = −1/3
This is obviously correct math, but formally you would do each step separately. The steps don’t necessarily need to be checked either, because it is an easy enough one that you can just check the result.
Math is a language, just a rigorous one, where it is simple to be right or wrong. It is a simple way to abstract away things that don’t matter, and talk about the underlying relations. Math is a subset of language with easier relations. For something with a pure general intelligence, math is probably much easier than a normal language.
I hold that we are story telling intelligences [and consciousness is us telling ourselves our own story as we compose it] that have been generalized through a deep understanding of the patterns in stories, which is why normal languages are easier for us -they were made to tell stories. (I also hold that you story of math is technically incorrect.)

deepthoughtlife 8 Aug 2022 16:19 UTC
2 points
1 ∶ 0
in reply to: Houshalter’s comment on: chinchilla’s wild implications
I literally noted that GPT-J, which uses said 7GB of math (assuming that number is right), usually fails at ‘2 + 2 =’. People can do several digit addition without pencil and paper. ’763 + 119 =’ probably doesn’t require pencil and paper to get ’882′. We do require it for many step algorithms, but this is not that. ‘Dumb’ computers do 64-bit addition trivially (along with algebra, calculus, etc.). I haven’t seen specialized math models, but I’m dumbfounded that general models don’t do math way better.
I haven’t tried coding using ‘AI’ tools, so have no real opinion on how well it compares to basic autocomplete.

deepthoughtlife 7 Aug 2022 13:07 UTC
1 point
0 ∶ 0
in reply to: Houshalter’s comment on: chinchilla’s wild implications
I am unsurprised it includes them, since it is an obvious thing. 7GB sounds like a crazy amount of math problems...but is only a tiny amount compared to what could be generated. Chinchilla was all about how they need more data, and it would be an easy way to increase that (correctly).
That don’t understand math on 7GB amount of examples is obviously related to the current extremely primitive state of logic in all such models. The big question, would it still not understand math and logic at 100x the amount of it. If it could learn basic abstract reasoning, that would massively improve its performance at all tasks. Since math and logic are literally just languages that express an understanding of the easiest (context-independent) of relations between things, that would prove modern techniques wholly unsuited to real AI. I suspect if it was 700GB of math, it wouldn’t fail so hard at math, but who knows?
(GPT-J even fails at things like ‘2 + 2 =’ on half the prompts I try, often giving strange results like ‘0’ or ‘5’ even with a temperature of 0, though often that is because it doesn’t even realize it is math, assuming that ‘2 + 2 =’ is somehow a programming thing even though the similarity is entirely superficial. Even when it knows it is doing math, it will often get the answer right at first, and then switch to ‘2 + 2 = 0’.).

deepthoughtlife 5 Aug 2022 15:39 UTC
2 points
0 ∶ 0
on: Would “Manhattan Project” style be beneficial or deleterious for AI Alignment?
I’m not convinced that the government wouldn’t be the riskier approach. Their incentives often don’t really lie with the stated missions of the agencies in charge. The official purposes of government are more aligned with us, but the actual purposes are not aligned in many cases. For profit corporations, on the other hand, are forced by the profit motive to have substantially more inherent alignment (though it can still be pretty low.). Governments are likely to mandate solutions that won’t or can’t work.
That said, the corporations are only aligned at all with humanity by virtue of being full of humans, and relying on human customers.
In general, when it isn’t clear what government should do, it is usually better that it doesn’t act at all (though not in every case.).
That said, the actual Manhattan project was almost completely separate from outsiders, and the government could very well try out a whole bunch of alignment strategies in general, while releasing those that are likely to work for further scrutiny at the later stages. This would be especially useful for slow takeoff scenarios.
Many would argue that an AI arms race is already developing between US companies and China.
(Note: I am quite skeptical of near future AI danger. I clearly believe in slow or no takeoff.)

deepthoughtlife 5 Aug 2022 15:26 UTC
2 points
1 ∶ 0
on: Deontology and Tool AI
Obviously, this approach would rule out a large number of use cases for AI wherein it may need to make things much better, but it risks hurting someone else. It is thus not a fully general solution, but it could be useful anyway for many other uses.
Another, more pressing problem is that a tool AI is likely to not be sophisticated enough to realize what the actual likelihoods its actions have to harm humans are. For instance, a language model can’t conceive of the harm its words could cause (assuming they actually advance enough as to pose actual danger.).
This seems more like something to make an AGI that acts like a tool AI. It could be interesting to attempt that if the time comes, but it seems to have that little disconnect. Giving extra capabilities to an AI in order to reduce the risk it does something unexpected seems like a plan with a reasonably high chance of failure.
For use on actual tool AI, it seems we would be stuck with whatever narrow rules humans can come up with for it, but ‘harm’ is very difficult to define in rules. (A utilitarian making the rules might very well argue that it would check to see if it caused negative utility to any humans, and that would be the definition of ‘harm’, but it is not widely agreed to, especially deontologically, and it still has the issue of not knowing the values of the people it is affecting.)
I do agree with the sentiment that it is better to have humanity itself be the risk case, rather than an alien intelligence we don’t know how to deal with. (I am personally very skeptical that there is any real danger of super-powerful AI any time soon though.)

deepthoughtlife 5 Aug 2022 15:00 UTC
4 points
1 ∶ 0
in reply to: KatjaGrace’s comment on: What do ML researchers think about AI in 2022?
That doesn’t mean it wasn’t rounding though. People ‘helpfully’ round their own answers all the time before giving them. ‘0’ as a probability simply means that it isn’t going to happen, not necessarily that it couldn’t, while ’100′ then means something that will definitely happen (though it may not be a logical necessity).
In some cases ‘1%’ could actually be lower than ‘0%’ since they are doing different things (and 1% is highly suspect as a round number for ‘extremely unlikely’ too.) Ditto for ’99%′ being higher than ‘100%’ sometimes (while 99% is also a suspicious round number too for, ‘I would be very surprised if it didn’t happen’.)
I don’t think that it would necessarily be telling, but it might be interesting to look at it without numbers from ’99%-100%′ and ‘0%-1%’, and then directly compare them to the results with those numbers included..

deepthoughtlife 5 Aug 2022 14:41 UTC
1 point
0 ∶ 0
in reply to: AllAmericanBreakfast’s comment on: For Better Commenting, Stop Out Loud
I reject your claim that what I said about needing to neutrally use good evidence as an editor was a dig, since it was simply an explanation saying what the duties of an editor are (and I very clearly stated that I believed you were likely to do it well, unless you’re somehow objecting to me calling it frustrating, which is just obvious?), because it was necessary when talking about what the policy should be. This failure of communication is exactly what I was talking about with people being unable to tell good vs. bad engagement on a topic. You assumed I was making a dig when I was doing nothing of the sort. (It is, of course, possible someone else would say exactly that as a dig.) My arguments should be evaluated on their own merits, not on whether or not someone guesses they are made for good or bad reasons.
Your duties as an editor do, in fact, determine what you should be doing, for the most obvious of reasons. The neutral use of good evidence is vital to the official reasons for Wikipedia’s very existence. (It is very explicit that there should be neutrality, though that is often not true in practice since everyone involved is human.)
I’ll expand the sentence you said you wanted rephrased into an equivalent paragraph: I am advocating a difficult policy on how to deal with changing Wikipedia pages explicitly based on levels of evidence on controversial parts. It would be much easier to use the SOL approach than to do this. The levels of evidence policy, however is not significantly different than what I believe should be the policy on every page, not just on controversial ones. It does not simply seem like a burden (though it clearly is one), but a way to improve Wikipedia. (It is also very much in keeping with the intended spirit of Wikipedia).
I do agree that telling off a person acting badly in the sci-fi example can be a good thing, (and would note that the satisfaction it engenders is a positive under a utilitarianism), and I’m perfectly willing to accept that you want it to be that way as a reason for supporting it -we’ll simply disagree that it somehow makes up for all of the obvious negative effects. You don’t seem to have engaged at all with the idea that intentionally making it get to that point was a clearly bad thing under most moral systems (which was the main point I made). Your assumption that things have to go to the very end is what seems like a ‘dodge’ to me.
To make something clear, how we get to a specific situation matters. When one side has intentionally provoked a conflict, they don’t get to act the same way they could reasonably act when they didn’t; this is true even if the side that engaged in it is morally correct to do so. It matters that the student chose this conflict. The teacher acts very badly, but is merely continuing things. When you choose conflict rather than cooperation in a shared endeavor, you have a stronger moral duty to use means that are clearly good to the extent possible. The student had better means available.
I made multiple points on alternatives to bare stopping which I endorse, all of which are clearly related to the further purposes you claim SOL has. In the sci-fi example, I advocated making an actual defense of sci-fi earlier in the string of events, preferably a calm and well-reasoned one, rather than intentionally violating expectations on an assignment to pick a fight. Then, when they’ve said their piece, they can stop. The ball is then in the teacher’s court, and they can choose how they will react. The student can then stop if there is no point in dialogue with the teacher, without having lost, nor having implemented a bad policy, or they can continue if it is worthwhile to do so.
In the Wikipedia case, making a real policy on how, when, and why engagement will happen is the correct approach. This is much more in keeping with the ideals of a community made encyclopedia than simply ignoring people whose viewpoints pattern-match ones that have been annoying before, and provides a good reason for why failure to respond is not an good proxy for losing. (Care must be taken so that it is not a despotic policy.)
Online harassment is a guaranteed thing. SOL gives more attention to the simple troll than it deserves as compared to simply ignoring it. More dedicated trolls can simply stalk you across the site, and SOL would encourage them, so unless you enjoy being trolled, it doesn’t seem worth using. You can continue to talk to other participants in the discussion while ignoring them, if you are still interested in the rest of the discussion.
The only place explicitly stopping that way has a use case is when you have a duty to be involved with all genuine participants in the conversation (this duty can be self-imposed), and need to explain why you think someone isn’t, but that is definitely a niche case.
Alternately, you can simply restrict the size and emotional tenor of your responses, and let the conversation die naturally in many cases. This doesn’t work if others are feeding the troll, but most trolls rely on the reactions of the people they are trolling directly.
If there is a whole pack of trolls, simply avoiding them seems clearly best (unless it rises to the level where you can and should remove them.).

deepthoughtlife 5 Aug 2022 13:28 UTC
1 point
0 ∶ 0
in reply to: Steven Byrnes’s comment on: Moral strategies at different capability levels
I never talked about respect in my points for a reason. This isn’t about respect. It’s about how it is not an agent if it doesn’t do anything in an attempt to make the world more to its liking. If it does nothing, or does things randomly (without regard to making things better), that is hardly agentic. If I don’t care at all about colors, then picking between a red shirt, and an otherwise identical blue shirt is not an agentic choice, merely a necessary choice (and I will not likely have even bothered thinking about it as a choice involving color.). Identically, if I just have to always wear a specific color, I am not being an agent by wearing that color. There are obviously degrees of agency, too, but the article is genuinely assuming that beings that do basically nothing are still agents.

deepthoughtlife 5 Aug 2022 13:11 UTC
2 points
0 ∶ 0
in reply to: TAG’s comment on: Conservatism is a rational response to epistemic uncertainty
That’s the thing. Conservatism? Inherently moderate. It is a specific kind of moderation. It’s an impulse to do things carefully, knowing the danger of doing things any other way. It’s moderate in terms of goals, and paths to get there. Epistemically? Very humble. We haven’t just figured everything out suddenly in this moment. The past was not that dissimilar to today, they’ve had our problems before, and there are groups that did the correct things and succeeded, while most failed. Learn from that. Having seen history, they know that there are ways things could be better too, and want to make those changes, but also that they could easily make things worse.
The difference with centrism is that centrism doesn’t really have a position on why we should be moderate, or how, just that something near the center between the current poles should work. (Individual centrists will likely have their reasons, of course, they just aren’t conservatism necessarily.) Many centrists are conservative, but many aren’t. Conservatives very often prefer and vote for centrists.
Staunch Leftists obsessed with progress can actually be quite conservative too, it just doesn’t come up in our modern political environment (even though they exist, they won’t get much air time, and I’d guess they aren’t a significant political force or someone would market to them.). (Conservatives amongst the democrats would likely talk about electability rather than their actual concerns, but that is just a guess.)
Rightists pay lip-service to conservatism, while leftists generally refuse, so rightists get most conservative votes, but conservatism doesn’t have an inherent place on the left-right scale (neither near a pole, nor at the center). Conservatism is an approach, not a set of policies.
Note: I’m more conservative-adjacent than actual conservative. I would say that I am more of a centrist than a conservative, though I think actual conservatives have a big point these days given the turmoil terrible non-conservative governance has wreaked upon our world. I am not temperamentally conservative; I am fairly obsessed with how things should be better, with less emphasis on caution, but we still need to be aware of all the conservative ideas on how to do that. The people arguing for progress should have been easily able to persuade me, but often fail to take even basic precautions.

deepthoughtlife 3 Aug 2022 17:03 UTC
1 point
0 ∶ 0
on: Conservatism is a rational response to epistemic uncertainty
I think you are missing a key insight into conservatism. Conservatism is about using things that are known to work, in the ways they are known to work (virtues). Conservatism does not blindly keep what it has now (because it could very well be or become malign), but rather looks and sees how it works repurposing or replacing things in a careful manner. The means by which the changes are made undergo the same scrutiny to avoid going down bad paths even when the result looks appealing. Conservatism has a virtue ethics approach to problems both moral and practical, because it knows we don’t just get what we’re trying for.
In other words, conservatism is just being rational in a world where things are likely to go wrong, but we have the experience culturally and societally to handle that if we’re careful. We have a lot of evidence in favor of certain ways of being, acting, and believing working well, and the newly proposed ways are very lacking in evidence that they are good, even if they sometimes have visceral appeal. A conservative is usually still against old things that are shown to be negative (by their value systems.). They can obviously still be wrong on any point, of course.
The wisdom of the ages is hardly faultless, but it’s a very good place to start (and also helps you understand what your policies really mean, since they are virtually never truly new.)

deepthoughtlife 3 Aug 2022 16:12 UTC
1 point
0 ∶ 0
in reply to: AllAmericanBreakfast’s comment on: For Better Commenting, Stop Out Loud
I don’t really think it will help, but even in your version of the student teacher interaction, it is still obviously negative.
1)Of minor but obvious importance, it is very confrontational and not beneficial. It in no way helps the student, it in no way helps the teacher, it doesn’t help any bystanders that hear about it. All it does is add a moment of satisfaction at telling off the teacher, and continues burning bridges (which the student obviously doesn’t care about, but isn’t good utilitarian thinking for a student, nor taking into account the feelings of the teacher for more universalist types.). (Telling off a person who is acting badly can be virtuous though.)
2)Crucially, the student knew exactly what they were doing by writing the science fiction for a class where the teacher was obviously telling them not to write science fiction. They picked this fight intentionally. (This is not necessarily a bad thing, but it needs to be justified too.)
The student could have also given a defense of sci-fi in class when the teacher kept saying negative things about it, that would have been more on point without being the one to actually escalate. (It would not necessarily have gone any better, of course.)
3)There is no internal difference for the student between the teacher mocking it for being science fiction, and making good points about typical failure modes of science fiction and then the student’s writing that said student didn’t want to hear, so there is no principled way to determine whether it really is this circumstance. The student can only see these things from the inside. It would be much more obvious if the student had previously asked the teacher their exact reasons for being so anti-sci-fi. (This is not an expected solution though since it runs counter to most people’s natures.)
Definitely not defending the teacher in this setup, of course. Taking the hypothetical as objective, it is an extremely unvirtuous way for a teacher to act. I don’t have a strong objection to telling off the teacher here, which this definitely is, but setting up things for it to go this way is clearly a bad thing, and the student did that at least semi-intentionally.
The defense of sci-fi being put this way makes it much less likely to be heard. Telling the other students why sci-fi is good has a much better chance of working. If everyone is persuaded about the incorrectness of the teacher’s bigotry against the genre, that pretty effectively neuters the effect of the teacher’s view without the negative side effects.
The Wikipedia example is harder to know how to interact with because I have little experience with being an editor of any kind, and only occasionally use Wikipedia more than for a moment, but it is well known that Wikipedia has some despotic editors too, that rule by fiat and keep pages wrong. It makes perfect sense for the people that when people want to fix something clearly wrong with a page, that they will be very frustrated when not allowed to do it. Many people will be wrong when thinking so, but so will many of the people assuming they don’t have a point just because it looks similar to a previous rejected point.
Better than SOL for Wikipedia would probably be to come up with a reasonable policy for the level of evidence required to change a particular controversial part of the page, along with a list of reasons why it is the way it is (this should also include self-reference on the level of evidence required to change the policy for when to change the policy on changing the part). This should be based on the current level of evidence that established it the way it is now (including both the evidence for and against). Also an explanation of what kinds of argument are given what kinds of weight, and an explanation of that (though this can be more general) would be good.
This is much more difficult in the short term than just stopping, but I don’t really see this as much different than what the policy should be even on parts where there is no disagreement on completely non-controversial pages. This should lead to higher quality pages. A reasonable policy on when things deserve a response would also be a good thing. (These all can obviously be abused too.)
Off-topic trolling can of course just be removed, but on topic trolling should just be treated by the strength of its (non)arguments. People are notoriously bad (because of normal human motivations) at telling the difference between good and bad versions of on topic engagement in places where there is intended to be mass use. Your job as an editor is to neutrally use good evidence, not just that which you agree with. (I expect based on our exchanges that you find this frustrating, but probably do a reasonable job.)
Does engaging in your exact examples really help bridge the gap between our positions or really increase comprehension? I think it is clear how things could be done better than SOL, by never letting it get to that point, by being thorough, and/or by simply side stepping it, and making the arguments to the other people.
(Side note: By this point in our many exchanges, I do think I’ve fairly effectively determined why I had an initial reaction that it was obviously wrong to endorse the technique, and I believe my reasons are clearly correct. There are simply better approaches, and even just being okay with bare stopping is still better in most circumstances.)