Posts
Comments
(Then perhaps build a second such solution that is orthogonal to the first. And so on, with a stack of redundant and highly orthogonal highly generic solutions, any one of which might be the only thing that works in any given disaster, and which does the job all by itself.)
This is excellent! Can this reasoning be improved by attempting to map the overlaps between x-risks more explicitly? The closest I can think of is some of turchin's work.
If it is trivial to do better with a few moments of reflection then make with the interesting comments. I see your near universal non-specific disdainful comments as a significant part of why LW is less pleasant to post to.
Strongly disagree. I would be more enthused about lesswrong if it had more attempts at futurism.
I recommend tabooing the word free in order to think more clearly.
detecting previously addressed ideas is a major impediment due to non-obvious terminology.
Increase the delay on reward loops with your phone by activating developer settings and setting colorspaces to black and white and setting animation speeds to 2x or 5x. I tried going back to 2x after months with 5x and it felt palpably neurosis inducing.
This is my favorite comment in a long while.
Doing yoga improved my rationality skills. If I were rewriting optimal exercise I'd add a section titled Retraining your Broken CNS.
Lossy compression isn't telos free though.
You can play with this right now and simultaneously dissolve some negative judgements. Think about the function of psychics/fortune tellers in poor communities. What do you think is going on there phenomenologically when you turn off your epistemic rigor goggles? Also try it with prayer. What might you conclude about prayer if you were a detached alien? Confession is a pretty interesting one too. What game theoretic purpose might it be serving in a community of 150 people? I've found these types of exercises pretty valuable. Especially the less condescending I manage to be.
and suffering-focused EAs do less stuff that tends to lead to the destruction of the world.
In support of this, my system 1 reports that if it sees more intelligent people taking S-risk seriously it is less likely to nuke the planet if it gets the chance. (I'm not sure I endorse nuking the planet, just reporting emotional reaction).
X-risk is still plausibly worse in that we need to survive to reach as much of the universe as possible and eliminate suffering in other places.
Edit: Brian talks about this here: https://foundational-research.org/risks-of-astronomical-future-suffering/#Spread_of_wild_animals-2
Related: perverse ontological lock-in. Building things on top of ontological categories tends to cement them since we think we need them to continue getting value from the thing. But if the folk ontology doesn't carve reality at the joints there will be friction present in all the stories/predictions/expectations built up out of those ontological pieces along with an unwillingness to drop the folk ontology on the belief that you will lose all the value of the things you've built on top. One model of the punctuated equilibrium model of psychological development is periodic rebasing operations.
Agree about creation:critique ratio. Generativity/creativity training is the rationalist communities' current bottleneck IMO.
Meta: if something has tons of evidence and you can't bring yourself to try it for a month ask yourself TDT-wise what your life looks like with and without skill of 'try seemingly good ideas for a month.'
Babbler reality has a strong pull because it doles out tasty treats.
It does have access to your nervous system since your nervous system can be rewired via backdriving inputs from your perceptions.
Olivia Cabane's books are where I'd start. Then Kegan's Immunity to Change.
Non-DSM: Opening the Heart of Compassion. People with psychotherapy chops explain the buddhist model of pathology in an entertaining way.
I think values are confusing because they aren't a natural kind. The first decomposition that made sense was 2 axes: stated/revealed and local/global
stated local values are optimized for positional goods, stated global values are optimized for alliance building, revealed local are optimized for basic needs/risk avoidance, revealed global barely exist and when they do are semi-random based on mimesis and other weak signals (humans are not automatically strategic etc.)
Trying to build a coherent picture out of various outputs of 4 semi independent processes doesn't quite work. Even stating it this way reifies values too much. I think there are just local pattern recognizers/optimizers doing different things that we have globally applied this label of 'values' to because of their overlapping connotations in affordance space and because switching between different levels of abstraction is highly useful for calling people out in sophisticated hard to counter ways in monkey politics.
Also useful to think of local/global as Dyson's birds and frogs, or surveying vs navigation.
I'm unfamiliar with existing attempts at value decomposition if anyone knows of papers etc.
On predictions, humans treating themselves and others as agents seems to lead to a lot of problems. Could also deconstruct poor predictions based on which sub-system it runs into the limits of: availability, working memory, failure to propagate uncertainty, inconsistent time preferences...can we just invert the bullet points from superforecasting here?
Relevant term is judgmental bootstrapping in the forecasting literature if anyone wants to dive deeper. It is extremely practically relevant for many circumstances such as hiring, where adhoc linear models outperformed veteran hiring managers.
Time Braid and The Waves Arisen. Super fun reads, and also seem to put me in agenty mode better even than other rationalist fics. I haven't seen the naruto anime and I got on just fine with both.
As for why my model works this way: heavily influenced by the research on deliberate practice#Deliberate_practice). Essentially, it caused me to see expert performance as the combination of several core traits which are all predicated on perceptual skills. The first is generating the correct chunkings that mirror the causal structure in the domain in the first place, which are composed of distinctions that you must learn to make. If you've ever done something like music where you went from hearing complicated sounds to hearing specific 'phrases' this is what i'm pointing to with perception of chunks. In order to build these up one has to also isolate the feedback/reward loop that allows you to zero in on your performance of that chunk. Cleanly delineating the hits from the misses and having that information be on the smallest time delay possible. The other skill is navigating the chunked tree, which is predicated on perception of cues/proxies that indicate which decision paths to take in your knowledge tree. This structure then has the ability to get activated by experiences in the real world, where you notice something that looks like a chunk you've already seen. Normal self help techniques generally don't have these hooks that fire in specific times and places, meaning you likely just don't remember to use them.
Ontology lock in. If you have nice stuff built on top of something you'll demand proof commensurate with the value of those things when someone questions the base layer even if those things built on top could be supported by alternative base layers. S1 is cautious about this, which is reasonable. Our environment is much safer for experimentation than it used to be.
This is why I like Naruto as a rationalist fanfic substrate: perceptual skills are explicitly upstream of action skills in the naruto universe. I think this mirrors the real universe and explains much of the valley of bad self-help. Action skills are pointless if you don't have the cues on when where and why to deploy them.
Another frame on the same concept: don't keep teaching people spells when their mana pool size sucks.
This was what the research review on nootropics indicated is mostly the case. I've also encountered a similar conclusion in many other areas. Enough so that my prior in new domains is now that you can cut off the tail of bad outcomes but can't do much to the upside.
backslash escape special characters. Test Common knowledge)
done by adding the '\' in logic'\') without the quotes (otherwise it disappears)
Big five personality traits is likely the factor analysis most people have heard of. Worth reading the blurb here: https://en.wikipedia.org/wiki/Factor_analysis
Many many models can be thought of as folk factor analyses whereby people try to reduce a complex output variable to a human readable model of a few dominant input variables. Why care?
Additive linear models outperform or tie expert performance in the forecasting literature: http://repository.upenn.edu/cgi/viewcontent.cgi?article=1178&context=marketing_papers
Teaching factor analysis is basically an excuse to load some additional intuitions to make Fermi estimates(linear model generation for approximate answers) more flexible in representing a broader variety of problems. Good sources on fermi estimates (eg the first part of The Art of Insight in Science and Engineering) often explain some of the concepts used in factor analysis in layman terms. So for example instead of sensitivity analysis they'll just talk about how to be scope sensitive as you go so that you drop non dominant terms.
It's also handy for people to know that many 'tricky' problems are a bit more tractable if you think of them as having more degrees of freedom than the human brain is good at working with and that this indicates what sorts of tricks you might want to employ, eg finding the upstream constraint or some other method to reduce the search space first of which a good example is E-M theory of John Boyd Fame.
It also just generally helps in clarifying problems since it forces you to confront your choice of proxy measure for your output variable. Clarifying this generally raises awareness of possible failures (future goodheart's law problems, selection effects, etc.).
Basically I think it is a fairly powerful unifying model for a lot of stuff. It seems like it might be closer to the metal so to speak in that it is something a bayesian net can implement.
Credit to Jonah Sinick for pointing out that learning this and a few other high level statistics concepts would cause a bunch of other models to simplify greatly.
Have you considered trying to teach factor analysis as a fuzzy model (very useful when used loosely, not just rigorously)? It seems strongly related to this and imports some nice additional connotations about hypothesis search, which I think is a common blind spot.
model uncertainty based discounting also.
Strongly related to Sarah Constantin's research review on nootropics: https://srconstantin.wordpress.com/2016/06/06/nootropics/
If drugs work, please investigate the underlying integrity of your life/self care.
Ie: coffee is less like throwing water on the out of control fire that is your sleep schedule and more like putting your fingers in your ears and and facing the other way.
The popularity of stimulants NOT being a flag is bothersome. If coffee and modafinil and methylphenidate seem like the greatest thing ever, investigate. There might be basement flooding, or load bearing column rot or other invisible hull integrity damage creeping in.
Having spent years thinking about this and having the opportunity to talk with open minded, intelligent, successful people in social groups, extended family etc. I concluded that most explicit discussion of the value of inquiring into values and methods (scope sensitivity and epistemological rigor being two of the major threads of what applied rationality looks like) just works incredibly rarely, and only then if there is strong existing interest.
Taking ideas seriously and trusting your own reasoning methods as a filter is a dangerous, high variance move that most people are correct to shy away from. My impression of the appeal of LW retrospectively is that it (on average) attracted people who were or are under performing relative to g (this applies to myself). When you are losing you increase variance. When you are winning you decrease it.
I eventually realized that what I was really communicating to people's system 1 was something like "Hey, you know those methods of judgment like proxy measures of legitimacy and mimesis that have granted you a life you like and that you want to remain stable? Those are bullshit, throw them away and start using these new methods of judgment advocated by a bunch of people who aren't leading lives resembling the one you are optimizing for."
This has not resulted in many sales. It is unrealistic to expect to convert a significant fraction of the tribe to shamanism.
It goes up at least one important meta level: fraction of the community willing to take on the (potentially high in ambiguous cases) cost of punishing free riders has threshold effects IIRC that determine which attractor you sort in to. Part of my S1 sense that EA will not be able to accomplish much good on an absolute scale (even if much good is done at the margin) is that it does not cross this threshold.
I'm not sure things have actually changed, I think some people's perception is getting sharper. But on the flip side, there is the notion of not regarding an entire person as bad or well intentioned. People are loose coalitions.
I agree with overcoming the curiosity stopper very strongly. I think the generalized form of this is an extremely important primitive action type to have. The decomposition hammer is like brainstorming, it seems to get stronger the more you habitually use it.
I disagree with the 'problem solving frame' which I think is the key sticking point. Lots of branches of psychotherapy and schools of buddhism seem to have independently reinvented the idea that you are not going to get anywhere on your problem until you acknowledge the ways it is helping you get your needs met.
'try being naively good first' has been a surprisingly useful thought.
If you really want people to engage with it you might want to do the hard work for them and strip out the preface and bold type chapter summaries from his book and upload and link them. Otherwise 99% of people aren't going to engage.
I'd recommend the second Harris podcast instead. They got bogged down on a side point in the first one as they mention. https://www.samharris.org/podcast/item/meaning-and-chaos
This wouldn't be a problem except for the fact that you have tacit beliefs installed by the path dependent process that is hugely religion influenced. It's useful to know about this. E.g. Nietzsche has useful insights for people who consider themselves non-christian but are still running the same OS. Etc.
Consider the difference between the frame of expected value/probability theory and the frame of bounded optimality/error minimization. Under the second frame the question becomes "how can I manipulate my environment such that I wind up in close proximity to the errors that I have a comparative advantage in spotting?"
After digesting for a few days my intuitive response is to add the handle 'virtue fatigue' to this concept cluster. Virtues are a means by which the commons are policed. When you have runaway virtue signaling this is essentially defecting against the commons. You get what you want from scrupulous people who take public virtues seriously in the short term, but create virtue fatigue in the long run as more and more gets piled on to this working behavioral modification channel. Eventually the channel fails. This might turn ugly.
play feels nice doesn't seem separate from play is probably useful.
Camaraderie is death spiral among anxious social group because a lot of people loudly object to such things. OMG you did not get their consent becomes an issue among people who are bad at exuding and reading non verbal signals about how welcome such things are.
Instead of submission and dominance I think an entirely different frame is helpful. Play fighting helps animals build the procedural knowledge for real fighting. In the same way that you would run or go to the gym with friends you might engage in playful status jostling with your friends so that when someone tries to actually knock your status down a peg it isn't a totally new reference class of experience and you just fire right back like you might with friends.
3.1.4 seems totally ungrounded from analysis and fairly random in its speculation.
If durability is hard for a consumer to evaluate, then the manufacturer will push costs there in order to devote more optimization power to the dimensions that consumers actually evaluate. Manufacturers that don't do this will be left behind for two reasons. 1. They will appear worse in terms of the features that are visible and 2. Their competitors will have larger advertising budgets.
The general principle is similar to the idea of an Aether variable. Costs get pushed into harder to evaluate dimensions. The products available in the market are only 'designed' by humans in a loose sense. After a market has existed for awhile it is more accurate to say the product is the output of a selection process. This is part of the reason new areas appear to make ultra fast progress initially, proxy measures haven't diverged from utility yet keeping the feedback loops tight.
Additionally: if only one member seems enthusiastic about thinking/planning/enforcing this kind of stuff that is a very bad sign. In such a situation when that person burns out the community slowly dies.
Common limiting beliefs can be seen as particularly strong attractors along certain dimensions of mindspace that coping mechanisms use to winnow the affordance space down to something manageable without too much cognitive overhead. Regularities in behaviors that don't serve a direct purpose could also be seen as spandrels from our training data clustering things that don't necessarily map directly to causality. Ie you can get animals to do all sorts of wacky things with clicker training which then persist even if you start only rewarding a subset of the actions if the animal has no obvious way of unbundling the actions.
Thanks a lot for sharing. I think there's something valuable to be learned from how you've managed to maintain option value.
I've looked into housing prices for multi family complexes and they scale sublinearly with number of bedrooms. The biggest obstacle is that people aren't really willing to invest significant fractions of their income in them currently (because you don't want to have to gather 8 investors for an 8 unit, chaos/life happens). Ideally something like 3 people/couples who think they are relatively stable would take on responsibility for an 8 unit with a significant fraction of their income. This is a risk, but one of the top regrets of old people is becoming socially isolated. I think investing a significant fraction of ones income in what will eventually turn partially into semi-passive income (once the mortgage is paid) and partially into their community it is okay to invest a larger than usual fraction of income in. This will still likely take an individual slightly more wealthy than your average techie to eat a larger chunk of the down payment than others and thus own more of the equity in the income stream.
I suspect this is fairly impossible in the bay area which has the lowest conscientiousness people in the US AFAIK.
Edit what I mean by pointing out low conscientiousness is that many people are incredibly short sighted and will defect when short term opportunities look better ie they will not tough out a few years of sub-optimal financial arrangement ie people don't actually grasp the concept of investing in a community. Related to why our kind can't cooperate.
Not all noble excuses are bad.
I think excuses in general are bad. They reinforce the narrative fallacy. All excuses collapse to I didn't care enough about the thing to take into account the factors affecting it. Including variance in factors you couldn't control. People don't like this frame because it makes them a lot more responsible for what happens to them. Note this should be deployed exclusively internally and not as an excuse to have less empathy for others. Everyone is doing the best they can. Note also that there is a distinction between responsibility and fault. It isn't your fault that random misfortunes befell you, but dealing with them is your responsibility whether you would like to think about it or not.