localdeity

Karma: 1,964

localdeity 5 May 2024 1:31 UTC
36 points
18
on: Some Experiments I’d Like Someone To Try With An Amnestic
I had heard, 15+ years ago (visiting neuroscience exhibits somewhere), about experiments involving people who, due to brain damage, can no longer form new memories. And Wiki agrees with what I remember hearing about some cases: that, although they couldn’t remember any new events, if you had them practice a skill, they would get good at it, and on future occasions would remain good at it (despite not remembering having learned it). I’d heard that an exception was that they couldn’t get good at Tetris.
Takeaway: “Memory” is not a uniform thing, and things that disrupt memory don’t necessarily disrupt all of it. So beware of that in any such testing. In fact, given some technique that purportedly blocks memory formation, “Exactly what memory does it block?” is a primary thing to investigate.

localdeity 5 May 2024 1:15 UTC
4 points
0
on: Some Experiments I’d Like Someone To Try With An Amnestic
One class of variance in cognitive test results is probably, effectively, pseudorandomness.
Suppose there’s a problem, and there are five plausible solutions you might try, two of which will work. Then your performance is effectively determined by the order in which you end up trying solutions. And if your skills and knowledge don’t give you a strong reason to prefer any of them, then it’ll presumably be determined in a pseudorandom way: whichever comes to mind first. Maybe being cold subconsciously reminds you of when you were thinking about stuff connected to Solution B, or discourages you from thinking about Solution C. Thus, you could get a reliably reproducible result that temperature affects your performance on a given test, even if it has no “real” effect on how well your mind works and wouldn’t generalize to other tests.
This should be addressable by simply taking more, different, cognitive tests to confirm any effect you think you’ve found.

localdeity 2 May 2024 0:43 UTC
6 points
4
on: localdeity’s Shortform
Pithy sayings are lossily compressed.

localdeity’s Shortform

localdeity2 May 2024 0:43 UTC

6 points

2 comments1 min readLW link

localdeity 30 Apr 2024 0:03 UTC
6 points
2
on: Big-endian is better than little-endian
One aspect neither of you have explicitly addressed is the speaking of numbers; speaking, after all, predates writing. We say “one billion, four hundred twenty-eight million, [...]”.
Given that that’s what we say, the first two pieces of information we need are “one” and “billion”. More generally, we need to get the first 1-3 digits (the leftmost comma-separated group), then we need the magnitude, then we can proceed reading off all remaining digits.
Given that the magnitude is not explicitly written down, we get it by counting the digits. If the digits are comma-separated into groups of 3 (and “right-justified”, so that if there are 3n+1 or 3n+2 digits, then the extra 1-2 are the leftmost group), then it’s generally possible to get the magnitude from your “peripheral vision” (as opposed to counting them one by one) for numbers less than, say, 1 billion, which are what you’d most often encounter; like, “52” vs “52,193″ vs “52,193,034”, you don’t need to count carefully to distinguish those. (It gets harder around 52,193,034,892 vs 52,193,034,892,110, but manually handling those numbers is rare.) So if getting the magnitude is a mostly free operation, then you might as well just present the digits left-to-right for people who read left-to-right.
Now, is it sensible that we speak “one billion, four hundred twenty-eight million, [...]”? Seems fine to me. It presents the magnitude and the most significant digits first (and essentially reminds you of the magnitude every 3 digits), and either the speaker or the listener can cut it off at any point and have an estimate accurate to as many digits as they care for. (That is essentially the use case of “partially running the algorithm” you describe.) I think I’d hate listening to “six hundred sixty three, six hundred twenty-seven thousand, four hundred twenty-eight million, and one billion”, or even suffixes of it like “four hundred twenty eight million and one billion”. Tell me the important part first!

localdeity 29 Apr 2024 23:06 UTC
3 points
0
on: Big-endian is better than little-endian
The other big-endian algorithm is the one I observe myself as usually using. For “321”, it is:
- Count the digits (three), and convert that into an order of magnitude (one hundred). (Running total: ??? hundred ???.)
- Read the first digit, multiply it by its order of magnitude (one hundred), and add it to the total. (Running total: three hundred ???.)
- Read the second digit, multiply it by its order of magnitude (ten), and add it to the total. (Running total: three hundred and twenty ???.
- Read the third digit, multiply it by its order of magnitude (one), and add it to the total. (Arriving at three hundred and twenty one.)
I generally agree, except I find words like “multiply” and “add” a bit misleading to use in this context. If I read a number like 3,749,328, then it’s not like I take 3 million, and then take 7, multiply by 100,000, and get 700,000, and then perform a general-purpose addition operation and compute the subtotal of 3,700,000. First of all, “multiply by 100,000” is generally more like “Shift left by 5 (in our base-10 representation)”; but moreover, the whole operation is more like a “Set the nth digit of the number to be this”. If this were a computer working in base 2, “set nth digit” would be implemented as “mask out the nth bit of the current number [though in this case we know it’s already 0 and can skip this step], then take the input bit, shift left by n, and OR it with the current number”.
(In this context I find it a bit misleading to say that “One hundred plus twenty yields one hundred and twenty” is performing an addition operation, any more than “x plus y yields x+y” counts as performing addition. Because 100, by place-value notation, means 1 * 100, and 20 means 2 * 10, and 120 means 1 * 100 + 2 * 10, so you really are just restating the input.)
Also, I might switch the order of the first two steps in practice. “Three … [pauses to count digits] million, seven hundred forty-nine thousand, …”.

localdeity 29 Apr 2024 10:04 UTC
3 points
0
on: Constructability: Plainly-coded AGIs may be feasible in the near future
Take note of the Underhanded C Contest for inspiration on the problem of auditing code written by an intelligent, untrusted source. I think one takeaway is that, with some ingenuity, one can often put malicious behavior into ok-looking code that maybe contains some innocent mistakes. It seems, then, that fully guarding against malicious code implies fully guarding against all bugs.
Which might be achievable if your style guides put heavy limitations on what kind of code can be written. (The halting problem makes it impossible for a deterministic program to always detect all bugs.) Something perhaps like the JPL C standards:
- Predictable Execution
  Use verifiable loop bounds for all loops meant to be terminating.
  Do not use direct or indirect recursion.
  Do not use dynamic memory allocation after task initialization.
...
Rule 3 (loop bounds)
All loops shall have a statically determinable upper-bound on the maximum number of loop iterations. It shall be possible for a static compliance checking tool to affirm the existence of the bound. An exception is allowed for the use of a single non-terminating loop per task or thread where requests are received and processed. Such a server loop shall be annotated with the C comment: /* @non-terminating@ */

localdeity 15 Apr 2024 2:47 UTC
5 points
2
in reply to: gjm’s comment on: A High Decoupling Failure
Agreed on most of the above, but on this particular point:
This is pretty much parallel to a common argument for laws against euthanasia, assisted suicide, etc.: the easier it is for someone with terrible medical conditions to arrange to die, [...], or (this isn’t quite parallel, but it seems clearly related) to make it appear that they’ve done so when actually they were just murdered.
I would expect the opposite there. If assisted suicide and stuff is legalized, I expect that to come with high standards of “There should be a notarized signature, multiple witnesses, a video from the person in question stating their intentions, and they walk into a building where some official people first take the person into another room and say ‘Are these men coercing you? We can have our security staff subdue them and bring in the police’”, etc., designed specifically to make it hard to cover up a murder like that. And the existence of that option should push a chunk of regular suicides in that direction, making it less plausible that someone would commit suicide in the “traditional” way where they give no one any warning, may or may not leave a note, etc.

localdeity 15 Apr 2024 0:25 UTC
2 points
0
in reply to: lukehmiles’s comment on: Alexander Gietelink Oldenziel’s Shortform
Yup. Many programmer applicants famously couldn’t solve FizzBuzz. Which is probably because:
[skipping several caveats and simplifying assumptions]
Now, when you get those 200 resumes, and hire the best person from the top 200, does that mean you’re hiring the top 0.5%?
“Maybe.”
No. You’re not. Think about what happens to the other 199 that you didn’t hire.
They go look for another job.
That means, in this horribly simplified universe, that the entire world could consist of 1,000,000 programmers, of whom the worst 199 keep applying for every job and never getting them, but the best 999,801 always get jobs as soon as they apply for one. So every time a job is listed the 199 losers apply, as usual, and one guy from the pool of 999,801 applies, and he gets the job, of course, because he’s the best, and now, in this contrived example, every employer thinks they’re getting the top 0.5% when they’re actually getting the top 99.9801%.

localdeity 11 Apr 2024 5:48 UTC
5 points
0
on: Is LLM Translation Without Rosetta Stone possible?
Without regard to anything specific to LLMs… Math works the same for all conceivable beings. Beings that live in our universe, of sufficient advancedness, will almost certainly know about hydrogen and other elements, and fundamental constants like Planck lengths. So there will exist commonalities. And then you can build everything else on top of those. If need be, you could describe the way things looked by giving 2D pixel-grid pictures, or describe an apple by starting with elements, molecules, DNA, and so on. (See Contact and That Alien Message for explorations of this type of problem.)
It’s unlikely that any LLM resembling those of today would translate the word for an alien fruit into a description of their own DNA-equivalent and their entire biosphere… But maybe a sufficiently good LLM would have that knowledge inside it, and repeatedly querying it could draw that out.

localdeity 10 Apr 2024 18:58 UTC
3 points
0
on: Sherlockian Abduction Master List
The Wikipedia link for claddagh rings goes to the Project Semicolon page.

localdeity 10 Apr 2024 18:54 UTC
7 points
3
in reply to: mako yass’s comment on: MakoYass’s Shortform
The true thing that Sagan’s line might be interpreted to mean is “A claim which is very unlikely on priors needs very strong evidence to end up with a posterior probability close to 1.” “Extraordinary evidence” would ideally have been stated as “extraordinarily strong evidence”, but that makes the line a bit clunkier. Unfortunately, there is often a tradeoff between accuracy and pithiness. Many pithy sayings require a bit of interpretation/reconstruction to get the correct underlying idea. I think anyone who invokes a catchphrase should be aware of this, though I don’t know how many people share this perspective.
Are there in fact a significant number of people who take it at the face value of “extraordinary evidence” and think it must mean it was obtained via super-advanced technology or something?

localdeity 10 Apr 2024 0:50 UTC
7 points
5
in reply to: romeostevensit’s comment on: romeostevensit’s Shortform
I think either I don’t know exactly what defines a “subculture”, or there needs to be a qualifier before “subculture”. Might “people who are enthusiastic about X sport / hobby / profession” be a subculture? Because I think lots of those can be highly successful while remaining what they are. (Perhaps you’d say that hobbies that succeed get eaten in the Geeks/MOPs/Sociopaths way, but that’s less so for professions.)
A “subculture of those dealing with X problem” sounds much more likely to fit what you describe, but that may not be your intent.

localdeity 6 Apr 2024 18:36 UTC
5 points
1
on: Failures in Kindness
imagine a friend from a different country is visiting and will stay with you for a while. You’re exchanging some text messages beforehand in order to figure out how to spend your time together. You want to show your friend the city, and you want to be very accommodating and make sure all their preferences will be met. So you simply ask them: “What do you want to do”? And maybe you add “I’m completely fine with anything!” to ensure you’re really introducing no constraints whatsoever and you two can do exactly what your friend desires.
An additional angle on situations like this: Your friend may be hoping to choose something that’s positively enjoyable for you. Saying “I’m completely fine with anything” may not meet that bar, and doesn’t give any hints as to what would. To illustrate directly, compare “There are ten restaurants nearby and I’m fine with any of them” vs “There are ten restaurants nearby, I’ve been to them all and I love them all”. I think there are people who would respond to the second with “Great, I’ll look them up and pick my favorite” and would find the first frustrating (and may respond by probing, “Well, are there any that you particularly like?” [And if you really don’t care about food, then their hope to find a restaurant you enjoy is destined for frustration.]).
In this case there’s also the aspect that, since you live there (likely for some years) and they’re from another country, you likely know a lot more about the local offerings than they do (not guaranteed—perhaps you’re an ascetic who doesn’t explore such things and they’re a tourist who has researched your town—but likely), so in a division-of-labor sense it’s likely appropriate for you to volunteer info first.
That second aspect is indeed about the pure computational problem. The first aspect is a combination of the computation/search problem and an emotional negotiation element.

localdeity 4 Apr 2024 23:40 UTC
2 points
−1
on: On Leif Wenar’s Absurdly Unconvincing Critique Of Effective Altruism
A friend of mine mentioned the article, and here’s what I wrote.
Some of it seems pretty unfair. The early anecdote about charity work in Bali seems to be used to criticize the EA ethos, when “rich westerners flying to poor countries to do manual labor (and possibly post on social media about how virtuous they’re being)” is the classic case of something EAs consider to be an ineffective and wasteful charity. (Though perhaps EAs might not go so far as to expect it to be net harmful.) I don’t think most EAs would agree to the “bet the Earth on a 51% chance” scheme. As the author says, “SBF consistently made terrible choices” even according to SBF’s own goals, so I don’t think one can point to his bad outcome as evidence that his goals were bad, except perhaps via a psychological argument that having his grand goals and being a powerful player led him to think “good-for-me-now and good-for-everyone-always started to merge into one” (and subsequent self-serving bias), which would be a general argument against having grand goals and being a powerful player.
But the thing of “lots of aid money ends up in the hands of corrupt middlemen and oppressive rulers and might make the whole thing net negative” is a good point; the specific things that went wrong are good to know about; and the thing of GiveWell not taking seriously and honestly the harm caused by the aid (which, given how they operate, would indeed mean publicly writing up calculations) is a very good point. They should take “tracking the negative consequences” into their routine practice; e.g. there exists N such that >$N being given to a cause justifies having a person go and investigate what’s happening.
I hadn’t checked any of the specific claims about GiveWell’s charities going wrong, or about what they have or haven’t written about the downsides; I basically took the author’s word on that.

localdeity 28 Mar 2024 13:36 UTC
7 points
5
in reply to: Said Achmiz’s comment on: My Interview With Cade Metz on His Reporting About Slate Star Codex
Looking at Wiki’s Undercover Journalism article, one that comes to mind is Nellie Bly’s Ten Days in a Mad-House.
[Bly] took an undercover assignment for which she agreed to feign insanity to investigate reports of brutality and neglect at the Women’s Lunatic Asylum on Blackwell’s Island. [...]
Once admitted to the asylum, Bly abandoned any pretense at mental illness and began to behave as she would normally. The hospital staff seemed unaware that she was no longer “insane” and instead began to report her ordinary actions as symptoms of her illness. Even her pleas to be released were interpreted as further signs of mental illness. Speaking with her fellow patients, Bly was convinced that some were as “sane” as she was.
Bly experienced the deplorable conditions firsthand. The nurses behaved obnoxiously and abusively, telling the patients to shut up, and beating them if they did not. The food consisted of gruel broth, spoiled beef, bread that was little more than dried dough, and dirty undrinkable water. The dangerous patients were tied together with ropes. The patients were made to sit for much of each day on hard benches with scant protection from the cold. Waste was all around the eating places. Rats crawled all around the hospital. [...]
The bath water was rarely changed, with many patients bathing in the same filthy water. Even when the water was eventually changed, the staff did not scrub or clean out the bath, instead throwing the next patient into a stained, dirty tub. The patients also shared bath towels, with healthy patients forced to dry themselves with a towel previously used by patients with skin inflammations, boils, or open sores.
Interestingly...
While physicians and staff worked to explain how she had deceived so many professionals, Bly’s report prompted a grand jury to launch its own investigation[9] with Bly assisting. The jury’s report resulted in an $850,000 increase in the budget of the Department of Public Charities and Corrections. The grand jury also ensured that future examinations were more thorough such that only the seriously ill were committed to the asylum.
I can’t say I’m happy with failure being rewarded with a higher budget. Still, it may have been true that their budget was insufficient to provide sanitary and humane conditions. Anyway, the report itself seems to have been important and worthwhile.

localdeity 28 Mar 2024 3:56 UTC
6 points
2
in reply to: Said Achmiz’s comment on: My Interview With Cade Metz on His Reporting About Slate Star Codex
The ones that come to my mind are “Person or Organization X is doing illegal, unethical, or otherwise unpopular practices which they’d rather conceal from the public.” Lie that you’re ideologically aligned or that you’ll keep things confidential, use that to gain access. Then perhaps lie to blackmail them to give up a little more information before finally publishing it all. There might be an ethical line drawn somewhere, but if it’s not at “any lying” then I don’t know where it is.

localdeity 27 Mar 2024 21:37 UTC
4 points
0
in reply to: Richard_Ngo’s comment on: My Interview With Cade Metz on His Reporting About Slate Star Codex
Ok, let’s consider the case of shadowy influencers like that. It would be nice to know who such people were, sure. If they were up to nefarious things, or openly subscribed to ideologies that justify awful actions, then I’d like to know that. If there was an article that accurately laid out the nefarious things, that would be nice. If the article cherry-picked, presented facts misleadingly, made scurrilous insinuations every few paragraphs (without technically saying anything provably false)—that would be bad, possibly quite bad, but in some sense it’s par for the course for a certain tier of political writing.
When I see the combination of making scurrilous insinuations every few paragraphs and doxxing the alleged villain, I think that’s where I have to treat it as a deliberate cancellation attempt on the person. If it wasn’t actually deliberate, then it was at least “reckless disregard” or something, and I think it should be categorized the same way. If you’re going to dox someone, I figure you accept an increased responsibility to be careful about what you say about them. (Presumably for similar reasons, libel laws are stricter about non-public figures. No, I’m not saying it’s libel when the statements are “not technically lying”; but it’s bad behavior and should be known as such.)
As for the “they’re probably robust” aspect… As mentioned in my other comment, even if they predictably “do well” afterwards, that doesn’t mean they haven’t been significantly harmed. If their influence is a following of 10M people, and the cancellation attempt reduces their influence by 40%, then it is simultaneously true that (a) “They have an audience of 6M people, they’re doing fine”, and (b) “They’ve been significantly harmed, and many people in such a position who anticipated this outcome would have a significant incentive to self-censor”. It remains a bad thing. It’s less bad than doing it to random civilians, sure, but it remains bad.

localdeity 27 Mar 2024 15:42 UTC
2 points
0
on: The other side of the tidal wave
I don’t trust that the replacements will be actually good
If ∑(people who agree with you on this * how much they agree with you) is high enough, that constitutes market demand for making an actually good replacement, or remaking the original. Then the cost-benefit calculation determines whether the market will incentivize someone (or an AI) to solve the problem.
If you, personally, have preferences that are truly unusual, then that could be a problem. Though I do expect the technology to make individually customized solutions cheaper and more feasible (in the vein of 3D printing). Perhaps, among all the truly unusual preferences you have, the net move will be towards more satisfaction of them.

localdeity 27 Mar 2024 15:10 UTC
14 points
9
in reply to: Richard_Ngo’s comment on: My Interview With Cade Metz on His Reporting About Slate Star Codex
But one difference between doxxing normal people versus doxxing “influential people” is that influential people typically have enough power to land on their feet when e.g. they lose a job.
It may decrease their influence substantially, though. I’ll quote at length from here. It’s not about doxxing per se, but it’s about cancellation attempts (which doxxing a heretic enables), and about arguments similar to the above:
If you’re a writer, artist or academic who has strayed beyond the narrow bounds of approved discourse, two consequences will be intimately familiar. The first is that it becomes harder to get a hearing about anything. The second is that if you do manage to say anything publicly — especially if you talk about the silencing — it will be taken as proof that you have not been silenced.
This is the logic of witch-ducking. If a woman drowns, she isn’t a witch; if she floats, she is, and must be dispatched some other way. Either way, she ends up dead.
The only counter to this is specific examples. But censorship is usually covert: when you’re passed over to speak at a conference, exhibit in a gallery or apply for a visiting fellowship, you rarely find out. Every now and then, however, the censors tip their hands.
And so, for everyone who says I can’t have been cancelled because they can still hear me, here’s the evidence.
The first time I know I was censored was even before my book criticising trans ideology came out in mid-2021. I had been asked to talk about it on the podcast of Intelligence Squared, a media company that, according to its website, aims to “promote a global conversation”. We had booked a date and time.
But as the date approached I discovered I had been dropped. When I asked why, the response was surprisingly frank: fear of a social-media pile-on, sponsors getting cold feet and younger staff causing grief. The CEO of Intelligence Squared is a former war correspondent who has written a book about his experiences in Kosovo. But at the prospect of platforming a woman whose main message is that humans come in two sexes, his courage apparently ran out.
Next came the Irish Times, my home country’s paper of record. Soon after my book came out a well-known correspondent rang me, said he had stayed up all night to finish it and wanted to write about it. He interviewed me, filed the piece, checked the quotes — and then silence. When I nudged by email, he said the piece had been spiked as it was going to press.
Sometime around then it was the BBC’s turn. I don’t know the exact date because I only found out months later, when I met a presenter from a flagship news programme. Such a shame you couldn’t come on the show, he said, to which I replied I had never been asked. It turned out that he had told a researcher to invite me on, but the researcher hadn’t, instead simply lying that I wasn’t available. I’ve still never been on the BBC to discuss trans issues.
Next came ABC, the Australian state broadcaster, which interviewed me for a radio show about religion and ethics. This time, when I nudged, I was told there had been “technical glitches” with the recording, but they would “love to revisit this one in the future”. They’ve never been back in touch. [… several more examples …]
Now, the author has a bestselling book, has been on dozens of podcasts, and now works for an advocacy organization that’s 100% behind her message (she’s not technically listed as a cofounder, but might as well be). She has certainly landed on her feet and has a decent level of reach; yet, clearly, if not for a bunch of incidents like the above—and, as she says, probably a lot more incidents for which she doesn’t have specific evidence—then she would have had much greater reach.
In Scott’s case… if we consider the counterfactual where there wasn’t a NYT article drawing such smears against Scott, then, who knows, maybe today some major news organizations (perhaps the NYT itself!) would have approached him for permission to republish some Slate Star Codex articles on their websites, perhaps specifically some of those on AI during the last ~year when AI became big news. Or offered to interview him for a huge audience on important topics, or something.
So be careful not to underestimate the extent of unseen censorship and cancellation, and therefore the damage done by “naming and shaming” tactics.
What links here?
- localdeity's comment on My Interview With Cade Metz on His Reporting About Slate Star Codex by Zack_M_Davis (27 Mar 2024 21:37 UTC; 4 points)

localdeity

lo­calde­ity’s Shortform

localdeity’s Shortform