Just this guy, you know?
Dagon
Karma: 12,940
Infinities are tricky. Once you get to “every possible local configuration of space-time will occur an infinite number of times”, you kind of lose the meaning behind the intuition that suffering is bad—everything just is, whether good or bad, it happens. Over and over, with no meaningful variation. What’s to judge?
If some configurations are more common than others (different degrees of infinite, I suppose), you could have beliefs about WHY they’re more common and preferences among them. But it still feels pretty abstract to me.
If there’s any significant chance that the universe is finite and some things don’t happen, then you’re back in normal intuition land, and you probably won’t be resurrected.
Do you make any distinction between “pain” and “suffering”? I tend to separate them, and I put a lot more weight on suffering, which I think of as an emotional/cognitive experience. Suffering quite plausibly is non-linear in brain-complexity, with more complex beings experiencing it trillions of times more intensely than simpler ones.
The confusion (in popular press, not so much among professionals or here) between censorship and alignment is a big problem. Censorship and hamfisted late-stage RL is counterproductive to alignment, both for the reason you give (increases demand for grey-market tools) and because it makes serious misalignment much less easy to notice.
Can you steelman the DR position a little? From this writeup (and from some lightweight previous explanation), I don’t understand why anyone sane and reasonably technical would think it possible or even sensible that “our senses give us direct experience of objects as they really are”. Our senses are very clearly intermediated by neural clusters that do a fair bit of interpretation/prediction, and then interpolated into mental models and beliefs.
I don’t see how anyone faced with any common optical illusion could think otherwise.
I think #4 is quite powerful. “identity” means many different things, and we haven’t had to distinguish them before, so many don’t even realize when they change topics.
Legal identity is likely quite distinct from any given continuity or branch/merge of memory. Memory identity and future-causality identity will eventually be distinct. Qualia would need to be measured before we could talk about experiential identity, but it won’t surprise me if we decide it’s different from either past continuity or future expected merges.
One nice side effect of these understandings (when we get to them) is it will answer age-old questions of harm under amnesiac drugs and a much better model of identity over long sequences of life/personality changes.
Ugh. This massively oversimplifies human desire or self-knowledge.
Procrastination is the broad fuzzy description of a phenomena where a person says he wants to do X, but does not do X.
Right. Or that parts of him want to do X, but other parts don’t want to do the elements that make X maximally likely. Or that he doesn’t want ONLY X. Or that X is the thing that is emotionally salient at the moment, but that won’t last. Or that X is what they want to be seen as wanting, but they don’t actually want it.
This feels a lot like the weight loss debate—Calories in < calories out is true, but useless. Defining “want” as “what you pursue wholeheartedly and consistently” is also fine, but useless.
procrastination is actually a broad label for seeming mismatch or contradiction of internal desires
Agreed, and that is real and difficult to deal with for most people. The solution (of integrating this self-knowledge and identifying/creating more consistent wants) is only ever achieved by a small subset of humans, and often takes decades of practice.
Interesting exploration, but I fear you’re selectively treating some preferences/beliefs as objective and not others.
Objectively, it doesn’t matter if her makeup washes off. There’s no one there that would judge her, and even if someone did judge her it’d just make them a jerk, which would be their problem. She doesn’t see it this way, obviously.
Incorrect. It is an objectively different world-state if her makeup is on or off. It’s less objective, but still true, that she will feel judged if she’s in public without makeup. It’s PROBABLY true that at least one person will treat her slightly differently, though is a question of weights and preferences whether that’s important.
It’s also quite possible that she doesn’t want to swim for other reasons, and is using her makeup as a means to avoid revealing those reasons.
The challenge is this: What do you say, and do, such that by the end of it she sees the truth here and is able to enjoy swimming with her friends?
Mu. There is no truth that I know better than her. There are framings and weights that I think may make her overall happier with the day (though they do have some risks that they won’t), and I’d likely try to help explain why I weight such considerations differently than she does.
I suspect there’s a LOT of handwaving in trying to apply theory to actual decisions or humans. These graphs omit time, so don’t include “how long spent in worse states, and how long the preferred state lasts” in their overall preference ordering. In known agents (humans), actual decisions seem to always include timeframes and path-dependency.
I think more technicallly-oriented history would benefit a lot of students. And more stories generally—not just facts, but world models of how individuals who were gifted in some ways and damaged in others navigated their society in order to produce lasting value.
But really, one size doesn’t fit all, and never has. Some students will bounce off some or all aspects of adulthood and competence, no matter how it’s presented. Some will be great, happy people regardless of classes. The really good teachers/professors and a well-fitting curriculum for a lot in the middle will be pivotal for them.
I think with modern interactive navigation support, the static implication of “map” is probably going away anyway. But to be clear, the POINT of the map-territory metaphor is that the map CANNOT be perfect, both because it’s less detailed, but also because it can’t change at the same rate as the world.
The real world is quite dynamic. Our mental models and descriptions of the world are less dynamic than the world itself.
I don’t mean to require perfect coherence—humans don’t have it, and if that’s required, NOTHING is conscious (note: this is a defensible position, but not particularly interesting to me). There’s enough difference between humans and the other examples that I’m not convinced by the analogies I’ve seen, and I believe this is the one important dimension of difference, but since this is all abstraction and intuition anyway, others are free to disagree.
In humans, there’s a lack of viability of independent subsets. It’s almost certainly the case that a partial brain still has some consciousness, and likely some differences from the whole being, but it’s not very divisible into truly independent segments. This is a kind of coherence that I don’t see in the other examples. Organs are not a good analogy for constituents or sub-organizations of a country, as organs DON’T have volition and world-models.If we take that critique seriously, we have to stop saying that corporations launch products, or that teams win matches.
No, that’s an isolated demand for rigor. We can definitely make different entity-analogies for different questions. When it matters, as it sometimes does, we can break the simplification and prosecute the officers or employees who are ACUTALLY responsible for a corporate action.
Under what conditions does it become useful or predictive to model a system as being conscious?
This is a GREAT framing for the question. Let’s not talk about “consciousness” as if it were a useful label that we agree on. Taboo the word, and the holistic concept, and let’s ask “when is it more useful to model a country as an entity that thinks and plans, as opposed to modeling it as a collection of groups of humans, who individually influence each other in thinking and planning”?
Both of these are quite believable. It took me a LONG time to get to a maturity/confidence level where I can say (still not comfortably, but I recognize that it must happen for good outcomes) “I’m just getting to this now, and I’m sorry I didn’t notice earlier that I don’t fully understand X and Y. Can you help clarify what’s the purpose/expectation/methodology for this?”
Good exploration, but I’m not agreed on some of the conclusions. I love Hofstadter, but remember his anthill is fiction, and one shouldn’t use it as evidence for anything. I don’t think anthill or country fit into my model of “conscious entity”. Though I suspect my main sticking point is “entity” rather than “conscious”. There’s a missing element of coherence that I think matters quite a bit. LLMs are missing coherence-over-time and coherence-across-executions. Countries are missing coherence-between-subsets.
When you say “countries do X”, it’s always the case that actually, some numbers of individual humans do it, and other numbers either don’t participate or don’t stop it. Countries do NOT state their right to exist. Humans state their right to be collectively recognized as a country. The preferences of the individual speakers may differ, but the actions don’t.
This is quite a different thing than noting “a human doesn’t say something, their chest, throat, and mouth muscles say the thing”. There are almost no muscle groups that act coherently without a brain to help coordinate.
Then I’m even more confused about the lack of cooperative-problem-solving between managers and employees. In fact, with fewer than 20 employees, why even HAVE a formal manager? You need some leaders to help prioritize and set direction, but no line-management or task breakdowns.
I think the “noticing” part can vary a LOT based on the implied reason for the manager’s request, and the cost/reward function of how close to “correct” the predictions are. There’s a whole lot of tasks in most corporate environments that really make no difference, and just having AN answer is good enough. An interested, conscientious employee would be sure this was the case before continuing, though.
The real puzzle is what is the blocker for just asking the manager for details (or the reason for lack of details). I didn’t work in big, formal, organizations until I was pretty senior, so I’ve always seen managers as a peer and partner in delivering value, not as a director of my work or bottleneck for my understanding. This has served me well, and I’m often surprised that much less than half of my current coworkers operate this way. “I need a bit more information to do a good job on this task” is about the bare minimum I’d expect someone to say in such a situation, and I’d usually say “do we have more functional requirements or background information on this? I can make something up, but I’d really like to understand how my answer will be used”.
Especially for the estimating parameters for a model question, I don’t understand why one wouldn’t ask for more information about the task and semantics of the parameters. If it were a coworker of mine, I’d mention it in a 1:1 that they need to take more ownership and ask questions when they don’t understand.
I think you need to define “effective” a bit more formally to answer this, and to state WHICH tests you’re talking about. I don’t know if the two dimensions you mention are the only (or even primary) purposes of most wide-scale standard testing.
They do seem to be effective at making funding a little tiny bit more transparent, and at showing parents which school districts they should be looking for when they’re choosing a city to live in. National standardized test are somewhat effective at showing relative academic/IQ strength, to a somewhat coarse degree (very diagnostic in the middle of the range, less so in the tails).
They probably do improve education in some cases, where the teachers/curriculum would otherwise wander into unimportant topics and ineffective methods. They almost certainly don’t improve it beyond a pretty middle-of-the road level. I think they probably help mid- and upper-level students in lower-level districts more than they help lower level students anywhere.
For most readers of LW, if you’re scoring near the top 10% of almost any standardized tests, they should mostly be ignored—the differnece betwen 90th percentile and 99th is mostly luck and approval-seeking-drive (note: this is overstated—it’s also conscientiousness, IQ, and social/innate support for wanting to learn that specific material). If you’re learning out of curiosity and self-driven reasons, you’re well outside the range where standardized tests can show you anything. If you’re scoring below that, then consider whether you should shift your study to include more common academic topics, or whether there are different metrics you’d prefer to measure yourself against.
I expect that there’s no simple relationship between these factors and success. Both are required, and it’s idiosyncratic which one is most lacking in any given margin between not-success and success.
Assuming all cars are traveling at a speed that gives 3 seconds of time between cars, any change to speed limit cannot affect the traveler throughput, and each car added lowers the speed of all other cars, including those at the front.
I don’t think this assumption holds. I don’t know what shape the actual speed-distance relationship is, but it’s not a straight line at a given number of seconds.
I also think the throughput measure (cars entering/exiting per hour) is rarely the most important thing for drivers or even planners. Average trip time outweighs it heavily.
European Roulette has a house edge of 2.7%. I think in the UK, gambling winnings don’t get taxed. I think in the US, you wouldn’t tax each win, but just your total winnings.
If you go the casino route, craps is slightly better. Don’t pass is 1.40% house edge, and they let you take (or lay, for don’t pass) “free odds” on a point, which is 0% (pays true odds of winning), getting it down below a percent. Taxes may not matter if you can deduct the full charitable contribution.
Note that if you had a perfect 50% even-money bet, you’d have to win 10 times to turn your $1000 into $1.024M. 0.5 ^ 10 = 0.000977, so you’ve got almost a tenth of a percent of winning.
Infinite time to run does NOT imply that every possible local configuration will obtain. If it’s cyclical, it could repeat the same few trillion years, which is a TINY subset of possible-experience-entity-space.
”Most things happen at the wrong time, and many things don’t happen at all. The conscientious historian will correct these defects.”—Heroditus