Edward Pascal

Karma: 117

Edward Pascal 21 Mar 2023 1:19 UTC
1 point
0
in reply to: DPiepgrass’s comment on: Let’s make the truth easier to find
Okay, I think I understand what you mean that, since it’s impossible to fully comprehend climate change from first principles, it ends up being a political and social discussion (and anyway, that’s empirically the case). Nonetheless, I think there’s something categorically in the physical sciences than the the more social facts.

I think perfect knowledge of climate science would tend towards convergence, whereas at least some Social Issues (Ukraine being a possible example) just don’t work that way. The Chomsky example is Germane: prior to 92, his work on politics was all heavily cited and based on primary sources, and pretty much as solid academically as you could ask for (See for example “The Chomsky Reader”) and we already disagree on this.

With regards Ukraine, I think intelligent people with lots of information might end up diverging even more as to their opinions on how much violence each side should be willing to threaten, use, and display in an argument about squiggly lines on map blobs, given more information. Henry Kissinger ended up not even agreeing with himself from week to week, and he’s probably as qualified an expert on this matter as any of us. I think it’s fair to suggest that no number of facts regarding Ukraine are going to bring the kind of convergence you would see if we could upload the sum of climate science into each of our human minds.

Even if I am wrong in the Ukraine case, do you think there are at least some social realities that if you magically downloaded the full spectrum of factual information into everyone’s mind, people’s opinions might still diverge? Doesn’t that differ from a hard science where they would tend to converge if you understood all the facts? Doesn’t this indicate a major difference of categories?

Another way of looking at it: Social realities are not nearly as deterministic on factual truth as accurate conclusions in the hard sciences are. They are always vastly more stochastic. Even looking at the fields, the correlation coefficients and R2 for whole models in Sociology, at it’s absolute best, are nothing at all compared to the determinism you can get in Physics and Chemistry.

Edward Pascal 20 Mar 2023 18:57 UTC
2 points
0
on: Let’s make the truth easier to find
I think another issue that would arise is that if you get “into the weeds,” some topics are a lot more straightforward than others (probably delineated by being rooted in mostly social facts or mostly natural science facts, which all behave completely differently).

The Ukraine issue is a pretty bad one, given the history of the region, the Maidan protests, US history of proxy wars, and, and, and. It seems to me far from clear what the simple facts are (other than you have two factions of superpowers, fighting for different things). I have an opinion as to what would be best, and what would be best for people of Ukraine, and what I think sections of Ukraine undisturbed by US and Russian meddling for the past 30 years might vote in referenda. And at least one of those thoughts disagrees with the others. Add to this the last 70 years of US interventions (see Chomsky for pretty good, uncontroversial fact-based arguments that it has all been pretty evil, and by the standards of the Nuremberg Trials one might execute every president since Kennedy).

On the other hand, Global Warming is pretty straightforward (even allowing for seeming complications like Mars temperature rise, or other objections). We can handle the objections in measurable terms of physical reality for a home-run clear answer.

One of OP’s examples is an entirely social reality and the other is a matter of physics. Let’s face it, in some sense this war is about where we draw squiggly lines and different colored blobs on a map. It’s levels removed from something where we can do measurable tests. If you really made all the truth easy to find, bringing someone straight into the weeds of a social problem like a US/NATO intervention, in many cases the answer will not come out clear, no matter how good your tool is. In fact, a reasonable person after reading enough of the Truth might walk away fully disillusioned about all actors involved and ready to join some kind of anarchist movement. Better in some cases to gloss over social realities in broad strokes, burying as much detail as possible, especially if you think the war (whichever one it is!) is just/unjust/worth the money/not worth the money, etc.

Edward Pascal 20 Mar 2023 18:40 UTC
1 point
0
in reply to: Roman Leventov’s comment on: Will people be motivated to learn difficult disciplines and skills without economic incentive?
“When the economic factor will go away, I suspect that even more people will go into fitness, body-building, surfing, chess, poker, and eSports, because these activities are often joyful in themselves and have lower entry barriers than serious science learning.”

This strikes me similar to the death of the darkroom. Yeah, computers do it better, cheaper, etc. However, almost no one who has ever worked in a darkroom seriously producing photography is happy that this basically doesn’t exist at all anymore. The experience itself teaches a lot of skills in a very kinaesthetic and intuitive way (with saturation curves that are pretty forgiving, to boot).

But more than this, the simple pleasure of math, computer programming, and engineering skills are very worthwhile in themselves. However, in John S Mills style utilitarianism, you have to do a lot of work to get to enjoy those pleasures. Will the tingle of the lightbulb coming on when learning PDEs just die out in the next 20 years like the darkroom has in the past 20 years? Meanwhile maybe darkrooms will make a big comeback?

I guess people will always want to experience pleasures. Isn’t learning complex topics a uniquely human pleasure?

Edward Pascal 20 Mar 2023 14:19 UTC
1 point
0
in reply to: Qumeric’s comment on: Wonder about the hard parts of the alignment problem
Thanks for that. In my own exploration, I was able to hit a point where ChatGPT refused a request, but would gladly help me build LLaMA/Alpaca onto a Kubernetes cluster in the next request, even referencing my stated aim later:

“Note that fine-tuning a language model for specific tasks such as [redacted] would require a large and diverse dataset, as well as a significant amount of computing resources. Additionally, it is important to consider the ethical implications of creating such a model, as it could potentially be used to create harmful content.”

FWIW, I got down into nitty gritty of doing it, debugging the install, etc. I didn’t run it, but it would definitely help me bootstrap actual execution. As a side note, my primary use case has been helping me building my own task-specific Lisp and Forth libraries, and my experience tells me GPT-4 is “pretty good” at most coding problems, and if it screws up, it can usually help work through the debug process. So, first blush, there’s at least one universal jailbreak—GPT-4 walking you through building your own model. Given GPT-4′s long text buffers and such, I might even be able to feed it a paper to reference a specific method of fine-tuning or creating an effective model.

Edward Pascal 18 Mar 2023 18:38 UTC
5 points
2
in reply to: baturinsky’s comment on: Wonder about the hard parts of the alignment problem
Has anyone worked out timeline predictions for Non-US/Non-Western Actors and tracked their accuracy?

For example, is China at “GPT-3.5” level yet and 6 months away from GPT-4 or is China a year from GPT-3.0? How about the people contributing to OpenSource AI? Last I checked that field looked “generally speaking” kind of at GPT-2.5 level (and even better for deepfaking porn), but I didn’t look close enough to be confident of my assessment.

Anyway, I’d like something more than off-the-cuff thoughts, but rather a good paper and some predictions on Non-US/Non-Western AI timeframes. Because, if anything, even if you somehow avert market forces levering AI up faster and faster among the big 8 in QQQ, those other actors are still going to form a hard deadline on alignment.

Edward Pascal 17 Mar 2023 1:32 UTC
1 point
0
on: The Waluigi Effect (mega-post)
Am I oversimplifying to think of this article as a (very lovely and logical) discussion of the following principle?:

In order to understand what is not to be done, and definitely avoid doing it, the proscribed things all have to be very vivid in the mind of the not-doer. Where there is ambiguity, the proscribed action might accidentally happen or a bad actor could trick someone into doing it easily. However, by creating deep awareness of the boundaries, then even if you behave well, you have a constant background thought of precisely what it would mean to cross them at any and every second.

I taught kids for almost 11 years so I can grok this point. It also echos the Dao: “Where rules and laws are many, robbers and criminals abound.”

Edward Pascal 24 Feb 2023 16:20 UTC
1 point
0
in reply to: Razied’s comment on: Building and Entertaining Couples
I think after all you will end up spending so much time together, there has to be something that overcomes the general human crazies that will pop up in that large amount of time. I remember a quote from one guy who went on an expedition across the arctic with a team: “After two months in close quarters, how do you tell a man you want to murder him for the way he holds his spoon?”

Desire and chemistry have a nice effect of countering at least some of that.

Edward Pascal 14 Feb 2023 14:20 UTC
1 point
0
on: SolidGoldMagikarp (plus, prompt generation)
Mathematically we have done what amounts to elaborate fudging and approximation to create an ultracomplex non-linear hyperdimensional surface. We cannot create something like this directly because we cannot do multiple multiple regressions on accurate models of complex systems with multiple feedback pathways and etc (ie, the real world). Maybe in another 40 years, the guys at Sante Fe institute will invent a mathematics so we can directly describe what’s going on in a neural network, but currently we cannot because it’s very hard to discuss it in specific cases with our mathematics. People looking to make aligned neural networks should perhaps invent a method for making them that doesn’t use fudging and approximation (“direct-drive” like a multiple regression, rather than “indirect drive” like backpropagation).

All this is known, right? So given GPT-3 has dozens of billions of x variables driving its hyperdimensional vector-space, I reckon we should expect this kind of thing to lurk in some little divot on some particular squiggly curve along vectors 346,781 and 1,209,276,886. I guess there should be vast numbers of such lurking divots and squiggles in the curves of any such system, probably that do way worse things than get the AI to say it likes Hitler and here’s how to make meth. Moreover, SolidGoldMagiKarp seems like a mundane example that was easily found out because it was human-readable and someone’s username.

Edward Pascal 1 Feb 2023 2:38 UTC
1 point
0
in reply to: jasoncrawford’s comment on: Why didn’t we get the four-hour workday?
The exceptions to what I said above, which are very bad are always the waiting. I hate it when I have 28 minutes of work to do, but it ain’t gonna happen until Joe gets that other thing on my desk. Then Supervisor Jake wants me to help him pick up a rental car. The inefficiencies in those two processes in the worst case, might eat a whole day and have me home late. This kind of stuff is demoralizing.

I think in the past, factory workers might savor that. It’s variety from the line, and it’s “easy.” For us management and information worker types, or at least me (lets say it’s just me) this makes me want to punch holes in drywall. Between people wanting to have meetings in rooms with chairs, and processes involving waiting, those office jobs can get very taxing. Working for myself I mostly avoid the meetings, but I still have those days of time-eating activities.

Perhaps a common culture here on lesswrong is jobs where “we’re gonna be here until everything is done” (including entrepreneurs and consultants) and so waiting is painful. Maybe for something like a government bureaucrat or a factory worker or similar, it would still be a boon.

Edward Pascal 7 Jan 2023 14:37 UTC
23 points
17
on: Why didn’t we get the four-hour workday?
I do not know how to explain this properly, but there is some amount of “non-work” work hours in every job I have done. If I were allowed to do everything that needed to get done and then go home at the end, no question, no raised eyebrows, etc, then most office jobs I have had would have been 2-4 hour work days.

Indeed, it’s hard to get more than four solid hours of cognitively intense work done on any given day anyway, and if I have done this, I consider it an especially productive day. I mostly work for myself now and typically do my 2-4 hours of intense cognitive “real production” work starting a half an hour after I wake up, with all the benefits of a night of sleep, good blood sugar, and a fresh pot of coffee. After lunch, I might work on a couple of hours of boring housekeeping type stuff, answering emails, ordering supplies, talking to providers, cash i/o, etc.

But those office jobs, really, most days were spent filling 8 hours and a half with what could have been a single uninterrupted 4 hour block. I think some of the filler is also “meetings.” I think a lot of people in broadly “middle level administrative” roles have probably experienced something similar.

Edward Pascal 23 Dec 2022 14:10 UTC
7 points
−2
on: Let’s think about slowing down AI
(1) The framing of all this as military technology (and the DoD is the single largest purchasing agent on earth) reminds me of nuclear power development. Molten Salt reactors and Pellet Bed reactors are both old tech which would have created the dream of safe, small-scale nuclear power. However, in addition to not melting down and working at relatively small scales, they also don’t make weapons-usable materials. Thus they were shunted in favor of the kinds of reactors we mostly have now. In an alternative past without the cold war driving us to make new and better weapons, we hit infinite free energy for everyone back in 1986 and in 2022 we finally got our Dyson Sphere running.

So yeah, the sad thing about it being an arms race with China and DARPA et al that is the AGIs will look a certain way, and that might become “”“what AGI is””” for several decades. And the safeguards and controls that get put around those military-grade AGIs will prevent some other kind, the equivalent of Molten Salt reactors, from getting developed and built.

But we have to accept the world of incentives we have, not the one we wish for.

(2) As a PD defector/exploiter without much in the way of morals or shame, what I like about all this slowing down is that it gives smaller players a little opportunity to catch up and play with the big guys. I suspect at least a few smaller players (along with the open source community) would make some hay while the sun is shining and everyone else is moving slow and ensuring alignment, which is democratizing and cool. I put this up here as a selling point for those people who crave the pedal to the metal on AI dev. The slowness the OP is talking about allows a different pedal to different metal if you are looking for it, perhaps with your foot on it.

Edward Pascal 21 Dec 2022 18:43 UTC
3 points
0
in reply to: nem’s comment on: How to Convince my Son that Drugs are Bad
“His knowledge of what was ‘safe’ and what wasn’t didn’t stop his drug usage from turning into a huge problem for him. I am certain that he was better off than someone thoughtlessly snorting coke, but he was also certainly worse off than he would have been had he never been near any sort of substance. If nothing else, it damaged some of his relationships, and removed support beams that he needed when other things inevitably went wrong. It turns out, damaging your reputation actually can be bad for you.”

I have a friend similar to your buddy here. He was vastly vastly experienced with drugs and “should have known better” but at age 40, with a great-paying programming career, he started taking meth occasionally. Stupidest thing I have ever heard of someone doing. The story ends with him trying to “buy” a 13 year old girl and showing up and the FBI vans were there for a sting op, and now he’s sitting in prison. Because meth can seriously skew your perspective on reality after a surprisingly short while.

The weirdest part to me is he would have been the first person to say meth is the worst drug and can skew your perspective into something beyond your worst nightmares. But it didn’t help him. Maybe his knowledge made him overconfident. Who knows? I cannot ask him until he’s out of the federal pen.

Edward Pascal 21 Dec 2022 18:23 UTC
6 points
2
on: Sazen
Much of what you have said here about capturing ideas is why I (and perhaps others) tend to prefer deep narrative as a means of conveying a lot. I mean, read Puzo’s original Godfather—more is in there than the movie. And ye gods, the movie has a lot in it. A summary of the movie is more or less meaningless to capture the multiple highly-coherent gestalts and meta-models available in the movies. I’m not even sure it points well at them later. I don’t know if you could even make something less than 1k-4k words that even points well at pieces* of the high-coherency embedded information in a powerful story (make it Dune, Breaking Bad, or Harry Potter as you like). At some point the best one might be able to do is write several paragraphs that point to pointers that point to the platonic ideal of an elegant pointer to whatever is conveyed. (And maybe I’m pointing well at something in this paragraph, that people who have been blessed by being consumed in a beautiful story world understand.)

So, with any real deep narrative, maybe it conveys something like living in the territory for awhile, or maybe it shortcuts your mind to something like a complex hyper-dimensional surface through back-propagation. Basically, you might be able parse what would or would not be consistent in the Harry Potter universe (depending on how well you read the books/watched the movies) pretty well, even in some really odd details, after your neural network was trained by the narrative. Fanfic is about someone building their own models of possibilities given the training data, and (not having read much) I assume part of the satisfaction in it is providing an extension to people that is consistent (“rings true”) within that world. People also often feel uneasy when a character or world becomes inconsistent (such as when the storyline of the TV show passes the original writer’s material and the committee just cannot nail it or has inconsistent intentions in their approach or agendas beyond total honesty of intention to the story and its characters). We humans are pretty good at noticing narrative consistencies, I think.

(I know I have gone a bit divergent from what you were saying, but you did push me to thinking about all this, so thank you)

EDIT: Are Koans attempts to do this? Maybe sometimes they come close to non-lossy compression for what amount to really vast hyper-dimensional surfaces in just a couple of sentences. To some extent they may still depend on context (though I can feel Hakuin hitting me with a stick when I suggest this).

Edward Pascal 20 Dec 2022 13:40 UTC
2 points
1
on: Notice when you stop reading right before you understand
“With school, often the way to get adequate grades with minimal time spent is to learn just enough so that you can do shallow keyword matching—full understanding is not needed.”

I once made $1000 from a desperate housemate during Undergrad, who had not written papers for his Epistemology course along with some other philosophy course. He met me at 10pm, the night before everything was due. Now, I did not understand Hume or half of what I wrote, but I literally bunched words according to meaning blocks and slapped them together with a consistent meaning ruleset with some reasonable logical jumps on that same ruleset. I was in the zone enough for him to get a low A and a high B.

Since then, I have been an educator myself and I try very hard to make that sort of keyword matching totally not possible as a final stopping place in my classes. However, it can be used along the way (see below).

And a note on the rest of your paper: Lately one thing I have been doing is “just finish the book.” Like, I was reading Quantum computing books and kept getting hung up on some points. So I went through A Student’s Guide to Vectors and Tensors to refresh my vector analysis. That was nice, however, it just didn’t get me all I needed to comprehend the Quantum stuff. What finally worked a lot better was just setting aside (in fact, shallow keyword matching) understanding of the points of quantum computing and just doing an entire book all the way to the last page. Now I am working another one (which is more programmer oriented) and kind of getting it. Again, I’m just chugging through the book, and mentally holding spaces for the things I still haven’t grokked.

Learning is weird. I kind of need to juggle ambiguity in a creative tension to construct the knowledge for myself Maybe this is a-la Vygotsky? I cannot say for sure because I didn’t finish the Vygotsky parts of the learning theory book. I got stuck re-reading and trying to really understand one or two concepts and never picked it up again after awhile. At the same time, I had enough to finish my paper and get an A. In fact, for my Instructional Design degree, I got nearly a 4.3 GPA in +/- grading. Yes, all of this really happened, including WRT Vygotsky, but as a result I don’t have enough knowledge base about Vygotsky’s theories to be certain one way or another if it’s ironic. I have a little suspicion it is based on my surface-level reading, and so I’m laughing nervously.

Anyway, everything you said was useful and cool. I am going to be thinking about this.

Edward Pascal 18 Dec 2022 21:31 UTC
1 point
0
in reply to: Dagon’s comment on: Edward Pascal’s Shortform
Do you believe in the existence of win-win? If so, why wouldn’t they tend to behave as I am suggesting? Also if you believe win-wins exist and think they do not behave this way, then how do you understand a win-win?

Edward Pascal 18 Dec 2022 17:21 UTC
6 points
3
in reply to: concerned_dad’s comment on: How to Convince my Son that Drugs are Bad
If, after all you discuss with him, he still seems hell-bent on it, would it be sensible that he do it legally and above the age of majority under the care of a professional? I would think that in many parts of the country, a psychiatrist can now assist with a psilocybin treatment. That is certainly in a different league from “I bought this shit in a bag from a guy named Lou and I’m going to put some Tool on and take it. Should be fine.”

The variables you’re controlling (set, setting, dose, PURITY OF PRODUCT, safe plan in case of abreaction, etc, etc) would seem all very important for taking mind-altering chemicals. The arguments that it’s probably safer than alcohol really do fail on some of those matters (particularly purity of product, in my opinion—if I buy Bacardi rum, I know it’s a controlled production. Lou’s LSD could have, nBOME, for example.)

Some of this you might bring up even if he’s not so hell bent. The studies done in medical settings on psychedelics have vastly more certainty of purity than the product you bought from some guy, no matter how you slice things. Meanwhile LSD synthesis is both illegal and highly non-trivial. I guess if he were to really synthesize pure LSD, you are talking about him doing it after he finishes his degree in Chem, so you have a few years to go. By then I would think he would have enough knowledge of the legal risks to steer well clear of that.

And I will say the next part as someone who likes occasional psychedelic use in safe settings (usually with an MD present and overseeing the use, though less formally than a clinical setting). I even use it sometimes to help with mild depression. I have not personally met anyone who takes psychedelics, or any drugs, a lot who isn’t really screwed up. I mean, I have taken them a grand total of seven times in my 38 years on Earth, and I think they have been net good.

However, to the last man, everyone I know who takes psychedelics often each year, or recreationally and frequently on weekends, has got problems. Just like everyone I know who smokes pot regularly has got problems. I’m sure there are regular illicit drug users out there who are fine, but I have never seen one in the wild, and I have known dozens of regular drug users. The ones I know who are functional also take it extremely rarely and with the same MD present as me, thus with a known controlled product and in a safe setting.

Meanwhile, the people I know who drink alcohol 50 times a year, or drink coffee 400 times a year, are generally all functional, and I know dozens of these people currently. All this is to say there may be something not well captured in, for example, the NIH chart of drug risks that puts LSD and Mushrooms way down at the bottom and alcohol up near the top. Here on Lesswrong, the contrarian view that the mainstream might be wrong is embraced, well I am suggesting that the mainstream NIH drug impact assessment might have issues as well.

Edward Pascal 18 Dec 2022 16:43 UTC
1 point
0
in reply to: Dagon’s comment on: Edward Pascal’s Shortform
I suppose this is technically true, but all concrete choices are not created equally.

Some policies tend towards win-win, for example “Let’s pave the cowpaths.” In that case, they are only going to bother someone with a systemic interest in the cowpaths not getting paved. Not to dismiss their interests entirely, like “they have some job that depends on routing people around the long way” or something, but this is going to, on balance, tend to be less people and less intense opposition (and more easily answered) than more zero-sum competitive approaches, for example.

I guess this is getting into a separate argument though: “Win-win thinking is fundamentally more Utilitarian than competitive zero-sum thinking.”

Edward Pascal 17 Dec 2022 18:23 UTC
3 points
0
in reply to: Dagon’s comment on: Edward Pascal’s Shortform
I did not say engineer something so that no one wants to destroy it. Just that if you have actually reached towards the greatest good for the greatest number, then the fewest should want to destroy it.

Or have I misunderstood you?

My argument is going something along the lines of the Tautological argument that ( I think) Mills (but maybe Bentham) made about Utilitarianism (paraphrasing much), “People who object to Utilitarianism that it will end up with some kind of calculated dystopia where we trade off a few people’s happiness for the many actually prove the principle of utilitarianism in their very objection to this. Such a system would be anti-utilitarian. No one likes that. Therefore it is not utilitarianism at all.”

Edward Pascal 16 Dec 2022 13:19 UTC
1 point
0
on: Edward Pascal’s Shortform
1. Any position that requires that group A not be equal before the law compared to group B, who get the full benefit of the law, means that group A probably has rational grounds to fight against the position. Thus that position has built into it a group that should oppose it, and if one applied the golden rule, if the second group B were in their shoes, they would also oppose it.
  Given how hard it is to make any very large and operationally functioning system, it is a lot to ask for it to also withstand the fact that for an entire group of people, it must be stopped. Thus with racism, sexism, nativism, etc, a lot of energy must be expended defending ideology and policy.
2. This is one major strength of Utilitarian ethics: The system you designed should have as little built-in rational opposition as possible. The converse of “The greatest good for the most possible” is “There will be the minimum possible number of people with an automatic morally-aligned drive to stop you.”

Edward Pascal’s Shortform

Edward Pascal16 Dec 2022 13:19 UTC

2 points

8 comments1 min readLW link

Edward Pascal

Ed­ward Pas­cal’s Shortform

Edward Pascal’s Shortform