Posts
Comments
It's also that a lot of philosophy, including famous works, is just really bad.
Thanks for this. The work I really want to see from more forecasting projects is an analysis of how much things that typically impact people's lives can be predicted. Things like health, home-ownership, relationships, career, etc. Specifically, people's levels of cooperate/defect against their future self seems really inconsistent. i.e. people work really hard for their future selves along certain dimensions and then defect along lots of others. This is mostly just mimetic momentum, but still. Even rigorous research figuring out exactly what actuaries know that can be applied practically by people would be good. After all, actuaries have really good life outcomes along lots of dimensions, which means that most aren't taking advantage of the insights there.
My hope had been that 80k hours would have evolved to do more in this area but they've specialized narrower than that AFAICT.
Votes have to ultimately accrue to super cooperators (cooperators who take on enforcement costs for the payoff distribution) via some mechanism for stability.
This is neat. It makes me realize that thinking in terms of simplicity and complexity priors was serving somewhat as a semantic stop sign for me whereas speed prior vs slow prior doesn't.
Agree, seems like LW for normies circa ten plus years ago? Reaction for standard metacontrarian reasons, seeing past self in it.
(metameta note, I think going meta often comes off as snarky even though not intended, which might contribute to Why Our Kind Can't Get Along)
People's metabeliefs are downstream of which knowledge representation they are using and what that representation tells them about
- Which things are variant and invariant
- Of the variant things how sensitive they are (huh, actually I guess you can just say the invariants have zero sensitivity, I haven't had that thought before)
- What sorts of things count as evidence that a parameter or metadata about a parameter should change
- What sorts of representations are reasonable (where the base representation is hard to question) ie whether or not metaphorical reasoning is appropriate (hard to think about) and which metaphors capture causal structure better
- Normativity and confidence have their own heuristics that cause them to be sticky on parts of the representation and help direct attention while traversing it
Really looking forward to this sequence!
It's funny because I would argue the exact same thing from the opposite side of the aisle. If you take someone who identifies with memory, internal speech, and internal image and reports nothing else, they don't seem all that conscious to me. Consider the moment directly before and after going lucid in a dream.
Come to think of it, I wonder if 'lucid' dreaming would even be a sensible distinction for a q-zombie.
I think a similar attitude of pessimism persists today due to lack of legible paradigm shifts in several major fields:
physics: the standard model keeps fending off all challenges
medicine: biology is hard; we face big combinatorial explosions in many of the search spaces
materials science: although there is progress in exotic materials, very few make the transition to being scalably useful to build cool stuff we couldn't before. Instead we get mostly incremental improvements in specific dimensions. Also a big combinatorial explosion search space.
politics/coordination: Arrow's impossibility theorem and various game theoretic tools have shown our representational schemes for aggrageting messy preferences to have some unfortunate difficulties
philosophy of science: hasn't recovered from Kuhn's critique and the failure of the Vienna circle's approaches to find a new formalism that improves the actual practice of science.
math: math exists in dialectic with cutting edge applications (despite strenuously objecting to this) and the lack of progress in the others has likely contributed to the stagnant theorem proving regime we find ourselves in
AI: with the death* of GOFAI we are relegated to competing to see who can figure out how to copy existing innovations (biological brains) fastest. This isn't all that inspiring in the grander scheme of things. Is the best we can imagine an AI to do the thinking for us?
I don't think doom and gloom inclined metaphysics are all that surprising given this state of affairs. I think it's worth pushing back hard against this. What if huge innovations are still hanging out in plain sight? What if the penicillin of mental health exists? What if a heuristic as important as natural selection is waiting for the right person to come along to articulate it? What if coordination problems have enormous breakthrough potential such that once people have the right set of simple as-yet undiscovered ideas a level of coordination is possible that is scarcely imaginable at present, in the same way that the global coordination that modern finance make possible would have been unimaginable in a mercantilist world?
Obviously, given that I work with Qualia Research Institute, I think major breakthroughs in consciousness research are one such possibility in this space. I imagine a world in which huge insights into phenomenology (beyond recapitulating Buddhism or just understanding psychedelics) cause it to take its place next to things like the scientific method or rule of law or information theory as a set of heuristics that enable a civilization worth living in. But there is a wide open field for other approaches too.
"If a problem seems hard, the formulation is probably wrong." -David Chapman
I think big, new syntheses are coming. There is more than one Bitcoin (cryptography, finance, game theory) like thing on the horizon and I don't think they'll be exclusively AI related or enabled.
I invite you to imagine that you live in a civilization that has barely scratched the surface, yet promotes pessimism as a coping strategy for those who don't feel they can contribute to the Main Plotline.
whoops, I agree with the heuristic and didn't actually mean for it to go to AF instead of LW. Hadn't paid too much attention to how crossposting works until now.
[internal screaming intensifies]
Can we somehow make Metaphors We Live By mandatory reading for these people? Reference class tennis plus analogical reasoning is only comforting in the sense that maybe someone stupid enough to be arguing that way isn't smart enough to build anything dangerous.
I agree but the steel man (not sure actually intended) is a mean variance issue and whether you're introducing a more sensitive parameter. i.e. you get the mean you want using the new control variable but variance is now higher and you don't actually understand the new parameter space this puts you in.
Consider various utilitarian fixes to classic objections to it: https://www4.uwsp.edu/philosophy/dwarren/IntroBook/ValueTheory/Consequentialism/ActVsRule/FiveObjections.htm
In each case, the utilitarian wants to fix the issue by redrawing buckets around what counts as utility, what counts as actions, what counts as consequences, and the time binding/window on each of them. But these sort of ontological sidesteps prove too much. If taken as a general approach, rather than just as an ad hoc approach to solve any individual conundrum, it becomes obvious that it doesn't specify anything about agents' actions at all as discussed in Hoagy's post: https://www.alignmentforum.org/posts/yGuo5R9fgrrFLYWuv/when-do-utility-functions-constrain-1
Another way to see it is as a kind of motte and bailey issue with domain/goal specific utility as the motte and god's eye view as the bailey.
Through this lens it becomes obvious that a lot of population ethics problems, for instance, are just restatements of the sorites paradox or other such problems with continuums. You can also run this the other way and use 'utility' to turn any conflict in mathematical intuitions into a moral puzzle.
Utility arguments often include type errors via referring to contextual utility in one part of the argument and some sort of god's eye contextless utility in other parts. Sometimes the 'gotcha' of the problem hinges on this.
I always have a window open to workflowy. It got me to consistently take notes in a way no other system did because of the low overhead. Downsides for some is that tabbed lists might not capture everything they want to do. Since access time trumps everything else when it comes to getting your brain to treat it like an extension, I find the lack of features to be a feature. It's a non starter for pictures.
Four Causes of Aristotle, Levels of Abstraction of Korzybski, and Levels of Analysis by Marr (elaborated later by Floridi), and Hayakawa's Ladder of Abstraction are all relevant. It's also been in use in the philosophy of medicine and philosophy (and methodology) of anthropology for a long time but I don't know who popularized it in those fields or recall what names it tends to go by.
Relevant: https://en.wikipedia.org/wiki/Global_workspace_theory
In particular, I think under this formulation knowledge and onotology largely refer to the same thing. Which is part of the reason I think this formulation is mistaken. Separately, I think 'reality' has too many moving parts to be useful for the role it's being used for here.
Yeah, that's part of it, plus having some model uncertainty about thinking of major parts of your system as a reinforcement learner.
> Seeking pleasure as a goal is goodharting.
I think this frame is subtly off. Something more like: using what feels good as evidence about what sort of thing you are.
Or, dis-identifying with a part of yourself often hides a tacit stronger identification with a different part of yourself, and is thus part of your ongoing war against yourself. This process is unavoidable, most of us have to pass through building a spiritual identity on the way towards a more widespread dis-identification.
Agree that this seems like a significant part of it.
Frozen peas and carrots add significantly to this and I do this a lot.
Meta note: I think it's a pretty big problem that even with reports by many high performing people that finding a particular creativity technique that resonated with them after investing some effort in trying that boosted their output by a multiple, people mostly don't seem to be able to take such claims seriously enough to invest the effort of trying. Secondly, such techniques usually give a boost for some time before dropping back towards baseline as you mine out the novel connection types that that technique causes. This also points towards having the meta-heuristic in place to regularly invest in trying new ones important.
A service where a teenager reads something you wrote slowly and sarcastically. The points at which you feel defensive are worthy of further investigation.
This is a good start: https://lifehacker.com/five-steps-that-make-it-easier-to-skim-through-non-fict-1758111229
Skimming a much larger volume of material with fewer commitments to a 'full stack read' (reading with note taking and later iterated summarization) allows one to more easily identify top quality information.
"ChuckMcM 3 days ago [-]
I am always amazed when people make comments like this: "The results of the University of Texas at Austin’s first full-semester foray into massive open online courses, or MOOCs, are in."
"Professor Michael Webber’s “Energy 101,” which had an enrollment that peaked at around 44,000 students, had 5,000 receive a certificate of completion — about 13 percent of the roughly 38,000 students who ultimately participated."
So let's unpack this a bit. Professor Webber created a class called "Energy 101" and processed 5,000 students through it to completion. Your typical 100 level undergraduate class might have anywhere from 50 to 200 students in it.
UT Austin this year had 8690 freshman total.
So assuming the largest possible class of 200, this professor in one semester taught the equivalent of 25 semesters of 200 student classes, in one semester.
Why should we care that 32,000 people signed up and then said "Woah, really don't have the time to commit to this right now?"
I experimented with some of these techniques for a while but found much bigger gains from improving my information searching, filtering, and skimming (random and structured sampling rules of thumb) heuristics.
Upvoted for trying to communicate something hard to communicate. This can be really frustrating leading to people often not trying, so thanks for trying.
>"I know it when I see it." but we might not be able to describe what "it" is.
hard to generate easy to verify functions. Related: Gendlin's 'sharp' blank, or a blank that knows what it is looking for, eg tip of the tongue phenomena, or forgetting what you're looking for and then remembering when you see it.
Dumping all distractions to an empty sheet is often helpful.
http://www.overcomingbias.com/2007/05/policy_tugowar.html
Made me realize that the things that are optimized for attention aren't the things optimized for accomplishing anything useful. I've since pursued many contrarian rabbit holes, some of which were of enormous benefit.
https://www.lesswrong.com/posts/X7G3HfRuMzCSBzgbM/87-000-hours-or-thoughts-on-home-ownership
Also, betting on things continuing to get better contra all the doom and gloom that is mostly generated by class tensions between the American middle class and upper class (using moralizing about the lower class as a weapon/distraction) who are both absurdly well off. Expecting to retire abroad because retiring to the US is too much a random gamble given health care costs. Also things like not expecting children to cost much money because I don't think spending money on education accomplishes anything (outside the scope of financial literacy per se, but affects things a lot). That and some various FIRE tropes decreases magnitude and variance on expected money spent to achieve a given level of well being over a life time.
Don't have it on hand but noticeably condensed in Sur/Petition.
Financial literacy enormously reduced worries about things like retirement and buying a house which increased my risk tolerance which lead to me working on projects that I actually care about.
>Peter Thiel's smart and highly original book
A lot of it struck me as incredibly similar to Edward de Bono's material from the 70's and 80's albeit applied to startup memes, which is still valuable.
>unmediated-by-ontology knowledge of reality.
I think this is a confused concept, related to wrong-way-reduction.
A short heuristic for self inquiry:
- write down things you think are true about important areas of your life
- produce counter examples
- write down your defenses/refutations of those counter examples
- come back later when you are less defensive and review whether your defenses were reasonable
- if not, why not? whence the motivated reasoning? what is being protecting from harm?
When young you mostly play within others' reward structures. Many choose which structure to play in based on Max reward. This is probably a mistake. You want to optimize for opportunity to learn how to construct reward structures.
Deference networks seem underrated.
> "How much value has this organization created?"
can insights from prediction markets work for helping us select better proxies and decision criteria or do we expect people to be too poorly entangled with the truth of these matters for that to work? Do orgs always require someone who is managing the ontology and incentives to be super competent at that to do well? De facto improvements here are worth billions (project management tools, slack, email add ons for assisting managing etc.)
One thing I don't understand about cycles is that they seem fine as long as you have a generalized cycle detector and a single instance of a cycle getting generated is fine because the losses from one (or a few) rounds is small. I guess people think of utility functions as fixed normally, but this sort of rolls in fixed point/convergence intuitions into the problem formulation.
One frame is that utility functions as a formalism are just an extension of the great rationality debate.
Not an explicit exercise and likely mentioned or alluded to somewhere in the forecasting stuff, but remember that any time you are about to run any sort of outside view/reference class forecast you have the opportunity to get some calibration by first trying to answer the question yourself based on what you know, including your confidence bounds on your model. When you subsequently look up the data and get any surprises, you can ask yourself why you are surprised which helps you to figure out what generators the experts have that you don't.
Since you can do this all the time (how many google searches do you do a day?) it gives you a lot more data than one time exercises.
It could use a clever anchor phrase for memory purposes. Open to suggestions.
See the butterfly and the net for generativity.
"The moment we eliminate identification we become conscious of abstracting, and permanently and instinctively remember that the object is not the event, that the label is not the object and that a statement about a statement is not the first statement." -Alfred Korzybski
emphasis mine.
I really like combinatorial explosions as the reason for superstition/cargo culting as a concept handle. Thanks.
QRI wants to simultaneously develop the penicillin of mental health and solve AI alignment.
Disposable plates and flatware are cheap and save significant time.
Cambridge handbook on expertise probably has some useful info.
Bicycles didn't make sense until there were smooth roads. I often wonder what road like things we are missing because they are non obvious, but would enable obvious inventions once created. The world wide web was pretty similar (warning: very longform)
> or we need some kind of controller to make the system behave-as-if the causal arrows line up.
This seems like a toe-hold for thinking about counterfactuals. i.e. counterfactuals as recomputing over a causal graph with an arrow flipped or the coarse graining bucketed differently.