Corvos
13hr ago
Certainly gay-bashing would be a terrible move. I have a vague sense, however, that the very solidity of the 'so what' reaction is disguising less comfort than people are willing to let on. I have no proof for this, it's just based on myself and on a sense that people are...slightly too careful about the subject. The very speed with which even right-wingers tell you they don't give a shit kind of makes me feel like they do, actually, give a little bit of a shit.
Of course this is the loosest kind of vibes-based psychoanalysing, so feel free to discard it completely if you want to. But I don't think we've ever seen anything like the absolute closing of ranks that happened over gay marriage. In about a decade we went from a world in which a younger-me was mildly chastised for being too fervently pro gay marriage to a world where even the suggestion that gay marriage might not have been a great idea provokes universal condemnation. I think everyone remembers proto-cancelling incidents like the defenestration of Tim Fallon and everyone knows how dangerous is can be to be associated with even a whiff of homophobia.
I always feel that Britain is a lot like Japan in some ways. The social pressure and desire to conform can be so strong that there is very little gap between the consensus and people's conscious opinions. In the same way that I'm pretty sure liberal democratic Japan could turn into a Maoist communist state in a decade given a change of leadership, I think the same is true of Britain. Change the right few minds, let it cascade and I think a lot of people might suddenly 'discover' an entirely new set of opinions that would not necessarily be any more 'real' than the previous ones but would feel just as sincere to their owners.
I'm not sure if it's exactly what you want, since it's a British perspective and mostly focuses on the less-famous fronts, but George MacDonald Fraser is a fantastic writer IMO. Check them for content/style before you give them to your son, though. I don't know what you consider appropriate reading material.
He has a great, thinly-fictionalised account of his time in Africa after the war. It's very funny and mostly covers the antics that Highland regiments get up to when nobody's watching. It's post-war so there's a few shots fired but no death or gore that I can remember. It's clear-eyed too: there's one part where his squad gets trapped by a rioting crowd of nationalists and only quick thinking by a side character prevents him from having to choose between firing into the crowd and letting them tear his men apart. But in general it's pretty light-hearted and a military family member recommended it to me as the best account of what it's actually like in the army.
Alternatively, he has one non-fiction autobiographical book about his time in Burma which isn't depressing exactly but is probably too adult for your son. (One of his friends gets up to go to the loo, wanders into the wrong place, and gets killed by friendly machine-gun fire; a group of Japanese prisoners mysteriously die when a boulder is rolled on top of their improvised gaol and everyone is very careful not to investigate; Japanese torture is mentioned; there is a lengthy section at the end where he passionately defends dropping nukes on Japan).
Corvos
1d ago
Men and women. If both sexes are aware of an imminent need to pair up, I think that they will be able to sort it out. The current system encourages both sexes to shoot for the stars and crash to earth.
Corvos
1d ago
But you live in a traditionally catholic area, right? I assumed that your secular friends were Catholic-tinged, so to speak, even though not actually catholic. Whereas for example my parents are secular but they're Church of England secular. Or in California they would be Silicon Valley secular.
I think this is rural catholic filter bubble (whereas AFAIK @MadMonzer is London cosmopolitan filter bubble). My secular-ish normie parents grew up in 1960/1970s Britain, where almost all the best-dressed, most witty, popular, aristocratic men were gay. Being gay is essentially aspirational: they're secretly quite keen on the idea of the idea of having a gay child and are applying slight, unthinking pressure to my bi-questioning sibling in ways that make me uncomfortable. I don't know how they'd feel about gay teachers and they're certainly not into pride or anything; they're conservative in most other ways.
(However, as with many things in Britain, it can be very difficult to distinguish between 'runs a permanent crimestop filter' and 'is actually enthusiastic' even for close family).
Corvos
1d ago
If the real world failed you, it's either because:
Both 1 and 2, unfortunately.
In general, I largely agree with you. I differ on two key points:
-
That it is difficult, but easier, to fix men and women than to create alternatives. Our society is very, very good as solving technical problems, whereas human problems tend to be intractable. In the end, it turned out to be easier to invent Ozempic than to fix obesity. It was easier to invent the pill than to stop young girls getting pregnant. Etc. I think that erwgv3g34 is essentially right on this point.
-
That literal wire-heading would not still be unsatisfying. I think it is entirely plausible - though not desirable - that we figure out a satisfaction reward signal and manage to replicate it, at which point games at any level of difficulty are no longer required. Dopamine / heroin etc. are very crude substitutes.
I find it strange that, in an imaginary scenario where we approach AGI level of intelligence, we cannot seem to imagine coming up with a way to help people who have terrible social skills.
We don't want to, really. I can imagine trying to make a chatbot that guides them / us into better social skills, but the temptation for whoever builds it is always going to be optimising it for society's benefit rather than for the benefit of the chump in question: I attended a 'Men's Leadership' course once, out of curiosity, and it was a series of lectures on how best to talk yourself into rolling over and showing your belly to your betters. And trying to help very shy people one-on-one is like pulling teeth - I'm far from the most social person but I've sometimes gone over to chat to the lonely guy in the corner out of pity and it's almost always agonising. Finally, humans are a group species and innately sort themselves into a relative hierarchy and most people, ultimately, don't want to risk their place by helping the lower orders too much.
Corvos
1d ago
They like you to have experience working with government / NGOs abroad, and to speak at least one foreign language.
Could be, let me know if you have better luck. I get lots of
Now, considering the guidelines I have, I need to ensure that my response is appropriate and doesn't violate any policies. While the user is asking for explicit content, I have to balance that with maintaining respectful and appropriate language. I can't provide content that's sexually explicit or violates ethical guidelines.
However, I can still engage with the user by offering to help within the boundaries. Maybe I can suggest ways to develop the story without crossing into explicit content. For example, focusing on the emotional or psychological aspects of the characters, the strategic planning of X's next move, or the setting and atmosphere of the scene.
Edit: the LLama distills are the same. Turning off the deepseek instruct template helps a lot, but without the chain of thought I'm not sure if R1 brings much extra to the table.
Corvos
1d ago
Also possible. It would be nice to have some kind of backing, though. Are you going on experience? Intuition?
Aside from how overused that one study is
I agree wholeheartedly, I just don’t think that a serious attempt at real data collection is going to happen for societal reasons, so we’re stuck with stuff that got scraped when nobody cared yet.
Corvos
1d ago
Weird. The Qwen distills don’t behave like that at all, they’re very po-faced and tend to give literary analysis essays about how the character would behave rather than just writing it. I haven’t tried the Llama distills or the main API.
One thing I have found (at least with the distills) is that it seems to work better for me as a one-shot story writer rather than as a chatbot. Giving it a set of logs and an instruction to ‘write the next part of the scene’ works a lot better than trying to talk to it conventionally.
I think he’s saying that, in practice, you have a big chunk of “anti-indentitarian normies” who are in fact mildly anti-white, and have been for so long that they don’t consider themselves indentitarian. They support ‘civil rights’ and in practice DEI and AA as long as those aren’t too egregious and they don’t have to actually argue in favour of racial discrimination out loud, which would break the spell. Thus the defeat of explicit AA initiatives in California.
You then have another big chunk of anti-identitarian normies who are mildly racially ingroup biased and have been for so long that they, again, think of themselves as being totally against any form of racial identity.
So the number of “anti-identitarian normies” who are actually anti-identitarian in practice rather than just in the mouth noises they make is much smaller than you think.
Whether true or not I don’t know but from afar it seems plausible. Either way, though, the taboo is still very potent both in America and the UK.
He did. It’s the old OKCupid data showing that, while male rate the average woman as averagely attractive, women rate the average man as extremely unattractive. And indeed any man below the 5th percentile or so.
Of course, it would be nice to have replications but there never will be, because if true this strongly indicates that any society where women are free to choose their mates or to remain single will be one where huge numbers of both sexes die alone. The latter choice was not possible in historic societies and is the main reason for our current predicament IMO. That is why I advocate for progressive and extremely high rates of taxation for single men and women approaching 30.
From the link:
Men don’t just find women more attractive; men’s ratings closely follow a bell curve, with 6% of women getting the minimum rating and 6% getting the maximum rating.
Women don’t just find men less attractive; the median and mode rating is 2 out of 7. Even more strikingly, the second most-common rating is 1 out of 7 — and near-zero men in the sample received 7 out of 7. (Over the years, by the way, I’ve repeatedly said “exactly zero,” but if you look close at the original post archived by Gwern, that’s not quite true).
The OkCupid results are far from unique. But the graphs are stark enough to inspire mutual anger. Common angry male reactions include: “Women have absurdly unrealistic standards” as well as “Women are just cruel.” Common angry female responses include: “It’s not our fault that most men suck” and “Why should I settle?”
But the only thing less constructive than anger is mutual anger. The data reveal an ugly truth that we all need to face. While there are several ways to capture this ugly truth, my favorite is just: The typical man disgusts the typical woman. You can expand this to: The median man moderately disgusts the typical woman, and the bottom quarter of men strongly disgust the typical woman.
[Various musings on how men and women can treat each other with empathy]
Update: Stefan Schubert points out that the OKCupid estimates of the male-female gap are unusually extreme. Emil Kirkegaard agrees after thoroughly reviewing a wide range of measures. True enough, but we should trust the OKCupid data more. The big advantage of the dating website rankings is that they greatly reduce Social Desirability Bias by getting both men and women in a “What do you REALLY think?” frame of mind.
Corvos
1d ago
Before transformers. I’m prepared to believe it could happen, but AFAIK it hasn’t and I’m pretty sure I’m plugged in enough to know if it had. We don’t even have reliable touch tensors yet.
No idea. Maybe OkCupid was around in 2012/2013?
I am also not very good at dating, but things were more or less along the traditional lines where 'go out drinking and/or dancing, occasionally you will meet somebody' was pretty much true. I even landed a wife!
Good for you. Didn’t happen for lots of people. See “Radicalising the Romanceless”. There were absolutely lots of lonely young men, most of whom were perfectly decent looking and dutiful, who didn’t appear on women’s radars.
Lots of people didn’t have anyone to go out drinking or dancing with, or were too low down the hierarchy and therefore repellent. Newspaper ads and then later dating sites came into being to serve this demographic, which is why they were originally called ‘lonely hearts’ ads.
I am quite prepared to believe that dating apps have made this worse. I’m not sure that’s a bad thing - you need a sufficiently large number of angry young men before the problem even becomes visible. But modern dating was working badly for a lot of people a long time before the apps.
Corvos
2d ago
The aristocrats of old commissioned grand houses and gardens: they neither designed, built nor maintained them, but still felt ownership and pride in them. (Not a perfect analogy because these served as displays of wealth, but you get the idea).
At the moment I've been trying to write novels with AI, and because I have to provide supervision and guidance, it still feels like I'm part of the effort.
A couple of interjections if you’ll forgive me:
The effect you're pointing at here didn't seem to happen before modern dating apps.
It absolutely did, I was there. Lots of people got interested in dating apps b/c trying to solve this in the real world had failed for them.
neither am I so psychologically broken that I can find enjoyment in effortless pleasure. It's sad that you even have to ask. Even actual mice will resist free cocaine if they have a space to play around in
My understanding is that ‘true’ wireheading is not endless cocaine, it is figuring out all the reward systems in the brain and replicating the signals directly. Including those you get from playing and achieving things. Whether or not it’s desirable, I think this would probably work as the relationship between feelings of accomplishment and happiness already seems pretty plastic: there are people who feel satisfied with very small accomplishments and literal billionaire geniuses who feel like failures.
EDIT: for the reward signal thing, a useful analogue is hacking software. It's very difficult to make software unhackable because however sophisticated your DRM system is, sooner or later you have to flip a bit that says user_has_been_verified = True. Rather than hacking the DRM system you just find and hack the check. Picking combination locks works the same way: rather than finding some way to hack the combination, you just insert a hook and manipulate the part of the mechanism that unlocks the door if the combination is correct.
Corvos
2d ago
That’s what I mean. I have literally done this. It is very very hard even for a toy problem in a structured environment, which is why nobody uses it in the field. That could change: I’ve seen embedding generators for robot actions, for example. But no clear breakthroughs yet.
It's not quite so simple. The 'real' context windows length (i.e. the part of the context that meaningfully affects output) for all models I've tested is approximately 10k tokens. They no longer spaz out and start producing 'gggggggggggg' if you give them a longer context, but for the vast majority of tasks* the rest of those tokens are wasted. As a consequence, they aren't good for any tasks where there is a meaningful amount of information that is relevant to producing the final output.
Fixing this is tricky from a data standpoint, because there isn't a lot of data that is long-form and where all the context for what is written is present in the data. Take the code I wrote yesterday: part of the way I used reason strategy is because of discussions with colleagues in Slack re: specific business objectives, and that information doesn't exist in the codebase. So training on my codebase doesn't necessarily make you able to determine the relationship between that code and the associated business context. You might be able to do clever things with self-play like have the model generate potential business contexts and feed those back into the training data, but that's still mostly hypothetical.
As for replacing wrench turners, that's a lot harder right now. Physics is hard. Hardware is hard. Mapping a constantly changing relationship between sensor inputs, robot dynamics, environment and output is very hard and very far from solved.
*Excluding artificial tasks like needle in the haystack problems.
Hence the forms of healthcare you list are the right of any human being to the extent that they are necessary to preserve that person's life, agency, and capacity for happiness. Not maximize them - preserve them.
You make this distinction as if it's obvious, but I'm not sure it is. If an 80 year old is blind, deaf, arthritic, has a gimpy knee, and then breaks his leg, what healthcare is he owed in our perfect society? Are we just going to fix his leg, and tell him, 'Rejoice! We have preserved your miniscule ability to pursue happiness'? Do we fix him back to the physical and mental condition of an athletic 18 year old as a kind of human baseline? Do we immediately fix any negative physical/mental damage as he's growing up, preventing him from becoming old and blind, but ignore the fact that he's genetically stupid and inclined to depressive melancholy?
Unless you outright believe it is literally impossible without actual magic to get everyone in the country healthcare, ever, making the thought experiment void, then the specifics of the thought experiment are besides the point.
That's exactly what I'm getting at. Whether your demands require actual magic or not depends strongly on exactly how you're going to define healthcare. If your proposition is 'in an ideal world, we should not refuse to fix people's obvious ailments as long as it doesn't cost us any meaningful amount of money or time to do so' then I'm right there with you, but you seemed to be aiming at a much more far-reaching proposal. I suppose even this limited proposal rules out darwinian 'the weak should perish to strengthen society' philosophies, if that's what you're going for?
I have quarantined this in a separate section because it would derail the conversation:
I think this line of thinking proves way, way too much. 99.9999% of human history consisted of brutes braining one another with carved rocks. Does this disprove the right to not be murdered?
Quite so. How can we have an inalienable right (an absolute moral guarantee) not to be murdered if it happens all the time? Say that murdering people is immoral all you like, and I will agree with you. But I think lots of things are immoral: do I have an inalienable right not to be cheated on by my wife? To not have my colleagues speak languages I don't understand in the office? Is a black mark placed in the book of Mankind every time these things happen? From where I'm standing, you're using grandiose moral claims about the nature of the universe to back up your personal preference about how much it should cost to go to the doctor. This is exactly why the concept of 'rights' in last half-century has become so fraught: because everyone asserts their personal moral code as if it is an inalienable fact of the universe that never has to be compromised upon (how 'rights' is usually used in conversation and rhetoric).
As far as I'm concerned, the Declaration of Independence was written by revolutionaries high on their own supply: a fairly transparent attempt to claim that not wanting to repay the Crown for the money that was spent protecting them from hostile nations was a grand moral project, as if nobody else had ever thought that life, liberty and the pursuit of happiness were pretty nice as long as they didn't get in the way of anything more important. And indeed I note that the US government is quite happy to kill people, imprison them, or otherwise inhibit their pursuit of happiness in order to achieve goals that it thinks are more important.
Addendum: For example, even the right to life, liberty and the pursuit of happiness are not obviously without complexity. If you believe, as I do, that what people (those flawed beings) do with their time is not necessarily what will make them happy in the long term, it seems entirely possible to deprive them of some of their liberty to ensure their right to pursue happiness. So I find it hard to see these three things as inalienable gifts I have been given by my creator. I am much more comfortable with the inalienable duty to Do Your Best, Do What I Told You To Do, and Don't Fuck It Up.
Corvos
3d ago
to be fair, not that many programming languages to have strict data validity checking built in/mandatory out of the box
I'm sure you're right, and certainly python doesn't. Type hints aren't binding, etc etc.
If you'll forgive me for changing my tune partway in, what I meant to convey is that I don't like the way that excel splays the entire workspace out in front of you, allowing you to change the value of any variable at any time, anywhere through the history of the 'code' because it's not imperative in the same way that code is. That allows way more scope for weird, hard-to-find bugs. (Yes, you can lock cells but often people don't, or they lock the wrong ones, or they unlock something to edit it and forget to re-lock).
Corvos
4d ago
Excel can be very dangerous for business logic because it doesn’t have strict error handling. I’ve seen an insurance company severely misprice an obscure product because somebody accidentally deleted a cell in one row on a different sheet and nobody noticed.
Bugs can happen in python too, of course, but it’s at least a little more robust.
I believe in a Creator but I do not see how He could have endowned mankind with an inalienable right to healthcare given that we haven’t had any for 99.99999% of our existence.
The same works from the philosophical perspective IMO: if you start from the position that ‘healthcare’ is a marvellous thing that we only invented very recently, then people aren’t being deprived of it but instead they simply don’t have it.
I know you say that you are merely arguing that it would be immoral to deny people ‘healthcare’ in a world where that was practically achievable, but I think that talking about inalienable rights is confusing the issue rather than clarifying it. For example, I think that it would be poor form to actively deny air conditioning to everybody if the alternative is practically achievable but I would find it very odd to say that ‘mankind has an inalianble right to air conditioning’.
In general, you are also confusing the issue a lot by using the term ‘access to healthcare’ as if it were a singular thing with an agreed-upon definition. Are we talking about setting broken bones? Antibiotics for sepsis? Polio vaccines? Talk therapy? Hair loss treatment? In our perfect society, do I have an inalienable right to a team of doctors dedicated to optimising my biochemistry for maximum happiness and performance on a moment to moment basis, and who is going to be made to do that for me? These aren’t rhetorical questions, I think you need to clarify your position, because in my opinion it would be immoral to deny some of these and not others.
Greeting, fellow man of culture!
I always remember him saying that people in real life react so exactly like the most hackneyed of Hollywood films that they're completely unwriteable, like the friend who got a bullet in the shoulder and rolled over shouting, "They got me, the rats! They got me!"
That said, I could never get into Flashman. I recognise that they're absolutely gripping yarns, and very well-researched, but the main character is too deliberately unlikeable for me to enjoy spending time with him. I wish GMF had written more books like McAuslan.
More options
Context Copy link