If it's worth saying, but not worth its own post (even in Discussion), then it goes here.

Notes for future OT posters:

1. Please add the 'open_thread' tag.

2. Check if there is an active Open Thread before posting a new one. (Immediately before; refresh the list-of-threads page before posting.)

3. Open Threads should be posted in Discussion, and not Main.

4. Open Threads should start on Monday, and end on Sunday.

New Comment
217 comments, sorted by Click to highlight new comments since:
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

Luke Muehlhauser, or lukeprog, leaves MIRI and takes on a research position at Givewell.

Nate Soares, or So8res, takes over as executive director of MIRI.

The first thing I thought of was the clear advantages to putting Nate in that leadershippy position at the organisation, since he's a college graduate who used to work for google, which is reasonably high status (in a non-controversial, obvious to outsiders way compared to say, Luke/Eliezer's backgrounds).

The second thing I did was remember how huge Luke seemed to be for MIRI - the organisation really got its shit together under his executive direction.

Interesting news, anyhow. I wish everyone luck.


Luke: the wandering get-your-shit-together-er. Making organizations better since 2011.


Luke isn't joining GiveWell as a get-your-shit-together-er, though.

I can dream. After all 5% of my stipend goes straight to givewell.
Maybe he'll be a very effective researcher.
"I'm Luke Muehlhauser. I get your shit together." (semi-quote)
If he could establish such a reputation as a consultant for commercial organizations, I imagine he could make quite a lot of money to do EA with.

There was a discussion a little while back (I think in another open thread) about the game of looking at the titles of articles linked from the "Recent on rationality blogs" sidebar and guessing who wrote them. Usually this is pretty easy.

Right now, though, the top link in the list is "The Future is Filters", which seems like an obvious Robin Hanson title. But no! It's Scott, not Robin, and it's about "filter bubbles" rather than "great filters".

I wonder whether Scott is aware of the game and deliberately trying to tease...

5Scott Alexander
I liked the sound of "The Future Is Pipes" and saved that sentence structure in case I needed it.
My thoughts exactly! I sometimes use that sidebar to find out about SSC updates (which I read somewhat more often than OB), and when I saw that title, I never even thought to scroll over it to see the link. Obvious Robin Hanson. Something something Great Filter. Well, imagine my surprise...
I'd be more surprised if he wasn't.

Here's a blog post about how everyone hates each other over politics more than before. Eliezer commented on it on Facebook, hypothesizing that it's a slow-growing effect of the Internet.

I cursed aloud when I read that comment, because I've had that exact idea and an accompanying sick feeling for a while now, and this is the first time I've seen it repeated.

(it's never a good sign when Eliezer Yudkowsky is the one to express your deepest fears about why everything's and everyone's brokenness is unstoppably accelerating)

I wish to read more about the "The Internet Is Why We Can't Have Even The Few Nice Things We Almost Kind Of Once Had" phenomenon — hopefully from someone who thinks there's a way easier than developing Friendly AI to put even one evil back in Pandora's Box, but that's probably wishful thinking, and I want to read about it in any case.

(Note: I'm aware that the entire LW-affiliated rationalist community writes about how things are broken, and desires to teach people to be less broken. But right now I'm looking specifically for things about how the Internet's massive boon to free speech is way more double-edged than was anticipated.)

Anyone have any good links?

Here's my diagnosis of the problem.

Before the internet: "Hey Bob, here's why you're wrong."

On the internet: "Hey everyone, here's my witty response to Bob, explaining why he's wrong and evil."

You can see how that kind of discussion would make people radicalized.

I've been thinking of an online discussion site based on exchanges of personal messages, which eventually get released to the public only if both participants agree. Maybe that would work. At least there would be no name-calling, because that's useless in a one-on-one setting.

It seems odd to me that you'd think that way. Surely, before the internet there was radio, television, newspapers, books, and numerous other ways to say "Hey everyone, here's why Bob is wrong and evil." I suppose the internet might have had a more democratizing effect where ordinary people can broadcast their opinions to the world. But I'd be curious to know how much that actually matters. It seems to me that it's still the case that if you're a 'regular nobody', nothing you post on your facebook is going to have an impact beyond your immediate circle of friends. At the end of the day there's only a finite amount of supply of attention. Maybe the internet has had a 'reallocation of attention' effect where people who used to recieve more attention previously (such as honest journalists) cannot reach as an wide audience as they could, and vice versa. But then, the question becomes: Who is getting more attention nowadays, and what effect are they having on people?
How about Twitter? That's where the problem is worst, and that's where people are constantly in "talking to the crowd" mode.
The difference is that now the Bobs can organize and start saying why clique of Alices in traditional media are the really evil ones. Of course, unlike the Bobs, the Alices isn't used to being called evil so they completely flip out and start going after everyone, even each other, who appears to show the slightest deviation from the perceived party line.

If you went to a party (meaning a social event) and started loudly proclaiming that anyone who does not vote for your favourite political party is a selfish git, people would tell you you were being rude, and you might be asked to leave (unless everyone there shares your views).

But on facebook, this sort of behaviour is perfectly acceptable. And once you get used to this online, it carries over into offline life. Faced with this onslaught, people with descenting views either shut up about it or change their views to match the majority.

I dunno if I use anecdotal evidence too much, but from my experience, five years ago it was possible for people to have different political views, to have a civilised conversation about policies, to agree to disagree. Now virtually everyone I know has the same political views and no-one discusses policies (you can't fit policies into a tweet, its too complex).

More generally I get the impression that even physical violence in the pursuit of political aims seems to be argued as justified more frequently, from rioting to throwing stones at politicians to angry jokes about arson against people who support the wrong party.

Tomorrow its the general election... (read more)

Consequences of being rude to people who disagree with you about something:

in real life -- lose friends

on internet -- gain pageviews

I've found that there's a risk of losing real life friends if you aren't careful about what you say online about politics.
Does the risk vary with position on the political spectrum?
Probably, but I've just got sketchy memories of what people say, which seems to mostly be people on the left getting sick of hearing right wing views. I know more people on the left, so there might be just as much of the converse. Personally, I've got a friend who doesn't want me to defend rightwingers to her, and the friendship is worth enough to me that I'm not going to nag her about rightwingers being human, too. She can figure it out on her own-- or not-- without my help.
I think it depends on where you are; for example, ISTM that supporting animal testing, GMOs, and the like used to be pretty much taboo on Facebook in my country until a few years ago, and I would have felt very uncomfortable expressing ideas that could be construed no matter how broadly as speciesist on my Facebook wall, whereas today there is a sizeable metacontrarian current which is mostly tolerated except by extreme environmentalist wingnuts.
Interesting - where do you live?

(it's never a good sign when Eliezer Yudkowsky is the one to express your deepest fears about why everything's and everyone's brokenness is unstoppably accelerating)

What's his track record with these so far?


Here's a blog post about how everyone hates each other over politics more than before. [And so on.]

So...I suspect my beliefs on this topic are out of step with the rest of LW, and even if I limit myself to the empirical aspects (i.e. set aside my normative differences) it's going to take a bit of effort to explain & justify my disagreements/doubts.

The first thing I notice is that the blog post talks about political polarization, as well as hatred/intolerance. Something I didn't realize until I glanced at the political science literature on polarization is that political polarization is multidimensional, so it's risky to talk about a change in political polarization in general.

To pin things down, we can first ask whom we're talking about: citizens in general, or politicians specifically? Then we can ask, polarization of what: specific policy preferences, or party identification? Finally, we can ask, is the polarization we care about variance in its own right (i.e. has the policy/party preference distribution spread out?) or covariance (e.g. has geographic sorting strengthened, which would be a rise in spatial segregation by political belief?) or attitudes between partisan gro... (read more)

Online shopping or wikipedia isn't going to polarise people. I'm sure many people here were early adopters, and hung out on usenet or mailing lists, but this was not the norm. It was around 2005 when myspace turned online socialising into something mainstream, and 2008 when not being on Facebook was actively contrarian, and a few years later when even the contraians gave in. Furthermore, in the earlier days blogs could express complex opinions. It was only with facebook and twitter that opinions boiled down to one sentence.
Blogs can still express complex opinions. The political impact of a figure like Glenn Greenwald who writes long blog posts is much higher than it was 5 years ago.
True, its possible the political influence of blogs has increased, but as a percentage of online politics blogs have massivly decreased
That has some plausibility. Contrarian that I am, however, I have some more opposing evidence in my pocket. In one of life's little coincidences, a potentially germane Science paper appeared on the same day as your comment. I'm less interested here in the paper itself (though it has some relevance) than in a snippet from its supplementary materials: As with my earlier comment, the data represent the US alone, but imply that the vast majority (95%) of Facebook-using adults there don't care enough about politics on Facebook to put a recognizable political affiliation in their profile. This doesn't contradict the experiences of the posters here who regularly encounter fighty partisans on Facebook (and real life?), but I'm not sure they're typical. Perhaps a selection effect helps to explain LWers' encounters with Intense Facebook Politics: LWers may be unrepresentatively likely to run into (or even be) Facebook partisans. From the survey, we're young (median age 26 — hence more likely to spend lots of time on Facebook than middle-aged technophobes), 37% of us are students, and over a quarter of us have consistently strong left-wing views on stereotypical contentious political topics (with a further 6% who simply rate themselves 5/5 on their interest in politics). (The "over a quarter" comes from taking the 8 rate-this-from-1-to-5 items on "Abortion", "Immigration" and so on, and summing each person's ratings, reverse scoring the "Human Biodiversity" item because that's the only stereotypical right-wing item. 272 people left at least one item unrated, but the remaining 1157 people had total scores between 0 and 40, higher scores being more left-wing. 402 people scored 32 or higher, which means 28% of all respondents had an average rating of 4+. Not one respondent gave similarly right-wing responses; the lowest score was 9. This is perhaps unsurprising in light of the political self-identification data.)
Seems pretty radical to me (assuming it's real and not a measurement artefact, of course).
When I framed it to myself as "that's a doubling of the tails!" it did sound impressive, but I remembered how easy it is to make modest changes in a distribution's mean and/or variance sound extreme by focusing attention on the tails, where such changes have an outsized impact. My reaction was to roughly translate that tail doubling into the corresponding change in the whole distribution's standard deviation, and I got about 30%. Expressed like that, the change was clearly substantial, but it didn't strike me as radical. Opinions may differ! (For this reply, I thought I'd try estimating a standard deviation for each distribution in a more systematic way. From Pew's appendix I worked out the mid-interval value for each of the 5 ideological-consistency bins, then calculated the standard deviations using those mid-interval bin values. This is still inexact, but hopefully less so than my original back-of-the-envelope guesstimate. For 1994 I got 3.90; for 2004, 3.91; and for 2014, 4.80. That gives a 23% increase in standard deviation from 2004 to 2014.)
ETA: Just noticed skeptical_lurker's comment. Upvoted for question dissolution and thorough analysis; one thing: A binary Internet-use question seems like a bad metric; people were primarily consumers in the past, now everyone is producing content, or actively sharing the content that others have produced. Furthermore, a lot of people on this thread seem to be talking about Twitter and Facebook; even if everyone's using the word Internet, clearly they mean social media (blogs included), and that's what we should be talking about anyway. We wouldn't expect considerably more polarization from people switching their consumption habits from TV and newspaper to big Internet news sites; that's just a change of medium. (That is, unless a bunch of people who weren't getting news from anywhere started using the Internet during that time.) It's often through extended social interaction that we get various forms of polarization, and social media offers more opportunities for that. And indeed, Pew's social media use data lines up well with the hypothesis that growth of social media use positively correlates with polarization of policy preferences in the US. Seems like I've tripped this clause but I don't really get it. Did that count as the sort of argument you were looking for? Also, I don't have a strong position either way on this, just pointing out what I perceived as a weakness in what seems like an otherwise good argument.
My guess is that more and more of us are living in Ellen Ullman's"Museum of Me": Combine that concept with one from Alexis de Tocqueville, 167 year earlier: On the internet, he need not withdraw into silence. He needs merely to find where everyone else who stated his particular opinion has withdrawn to. Further and further towards the edges of Pew's graph, most likely.

(it's never a good sign when Eliezer Yudkowsky is the one to express your deepest fears about why everything's and everyone's brokenness is unstoppably accelerating)

You should be dubious about "unstoppably accelerating"-- prediction is difficult, especially about the future.

It's conceivable that people could get sick of the current level of nastiness.

Heh, I know. I chose that phrase to express despair more than to describe objective reality.

I think the article makes some sweeping and unjustified statements.

Just a few generations ago, everyone knew that there were two subjects not broached in polite company: religion & politics.

This is not true; in many parts of the USA for instance it was (and is) a quite common question to ask someone which church they go to. This might sometimes literally be the third question you get asked right after "what's your name?" and "where do you come from?"

And, of course, many other parts of the world openly debated politics and religion in public and with strangers, and in some places quite often. This is well-documented, and I have personal experience with this since I used to (in the 90's) live in such a country.

I think you (and the following chain you have gathered) are wrong on your prediction on this matter. I would not be unwilling to say that Eliezer's prediction (in your direction) is also an inaccurate prediction of the future. (before I go on; I should warn that this very topic has developed a gang opinion; with echo chambers where it is possible for many people to share the same view on the topic without being disputed) I believe that the world is still growing in its understanding of what is this internet thing. In doing so; I believe we are yet to get to the "less caring" point of time. if we look at politics ~40 years ago; there was a big politically active shift. Many joined in; many cared! Many protested. And then it all cooled off again. People cared less; people wanted to continue with their lives more and spend less time talking about politics. I predict many of the public are soon to hit the there is too much politic and not enough living in my life, at which point; people will care less of politics and share less politics. everyone will still have pet issues; and dagnamit! if you can't anger a dog-lover by telling them that dog-lovers are inferior; then it wouldn't be planet earth. Instead of; "this is all getting worse"; I predict its about to get better. if you disagree; feel free to discuss.
I like Venkatesh Rao's perspective on this, in his blog post about "escaped realities". He argues that our games and mental models of the world are becoming closer to reality over time, "less escaped", rather than more escapist. He points out that geography is a much bigger filter than online groups regarding the variety of ideas that a person is exposed to. If anything, the internet, overall, seems to be widening people's perspectives. (Although it also lets people get better organized with their tribalism.) I think the United States is undergoing a sort of cultural civil war right now, which I think makes the perspective of people living there somewhat different than people living in other parts of the world. I don't think that it would be as easy to make the case for "the internet encourages tribalism in politics" in other countries. I don't think that the internet is the reason for the social tensions in the United States right now, (and there are many.)
Here's my half-baked idea. Since the world is becoming safer, we have less real threats to prevent general ennui and so petty status games start to take on more importance.

Except that I would say this political hatred has taken off over the last five years, correlating far more with social media than with the world becoming safer. Has the world become safer over the last five years?

Alternative hypothesis: it's about young adults from rich families, seeking status in the muggle world. In real life, if you are a spoiled rich kid, the best way to enjoy your wealth is to do things that most people don't do. Buy a private helicopter or a yacht, organize a huge party in your mansion, etc. You are invisible to the muggles, and the muggles are invisible to you; that contributes to social peace. However, this does not work on internet, because on internet the fun is where people are. Imagine that someone would create an alternative "Reddit Platinum" website where you would have to pay $1,000,000 every day to have an account. If you have more money than you can count, you could isolate yourself here from the muggles. The website would be wonderfully designed, and all bugs would be fixed immediately. The only problem is that there would be almost no interesting debates, because there would be not enough various people to bring new ideas. So even for a spoiled 1% kid, the muggle Reddit would be more fun then the "Reddit Platinum". On the other hand, the muggle Reddit would be frustrating for them, because they wouldn't receive the respect they get in the real life where everyone tries to kiss their assess. Instead, if they tried to pull the status card, people would make fun of them. The natural response would be to use their real-world resources to buy minions, and somehow use those minions as a support in online battles. And this is a problem that cannot be solved, because the parts of the online world where the rich kids can feel really good become boring as hell for everyone else, so most people move away, and the fun moves away, and the rich kids will try to conquer yet another part of the online world. It is a love-hate relationship towards the arrogant muggles: can't withstand their non-humble behavior, yet can't live online without them. EDIT: A weak-evidence datapoint: In Slovakia there was a website "booom.sk", something like Facebook for r
I don't think that playing status games online is reserved t people with parents rich enough to buy a helicopter or a yacht. I don't think that a community leads to be big to have interesting debates. LW isn't big.
Penny Arcade has just written about this as well. Comic strip, column.
I think programmers helping each other on Stackoverflow is an example of people being nice by helping each other. It's possible to create any kind of social norm in an online community. Often the problem is that there's often no moderation that enforces any community norms. Most newspapers don't invest the necessary resources that would be required to get decently moderated comment threads.
You're referring to the problem with people being mean to each other within a given online community. I'm thinking more of people hating each other more in real life because the Internet lets them seek out unfiltered outrage from people with similar beliefs, with nothing tempered by gatekeepers as in the days before the Internet and the rise of cable news.
I don't have the feeling that it's hard to speak about politics in the social circles in which I move in real life. But then I don't live in the US but in Germany.
More likely that means your opinions align with your social circle. As an example, do you think someone who supports the PEGIDA protests would feel comfortable expressing that support in your social circle?
I did read pro-PEDIGA posts on facebook from people where I wouldn't have expected it.
When was the first time you got online? For me I think, 1994-1995-ish. And it was a surreal place largely because the Internet was and still is a lawless anarchistic Wild West. Nazi types set up their Geocities websites because hate speech laws were not enforced. Restrictions on pornography in certain countries were overriden by looking at it from abroad. Age verification, a fairly important rule, was overridden so much that lately many porn sites don't even care anymore and now minors can access it. Hoax websites scammed people out of money. Socially inept nerds became millionaires. Filesharing and piracy. Uncontrolled free speech and under pseudonyms, so you are not only allowed to wear a smear campaign against a political big dog but even they don't know who you are. It was the Wild West where Anything Goes. I and many "old-timers" got attached this anarchy online because freedom. It is at some level romantic in the Samizdata / Doctorow sense and of course in many things even useful. But look, can we realistically happen it will always stay an uncontrolled Wild West or sooner or later the Real Powers in the Real World will bend the Internet to their world? I think they will. We will probably lose this marvelous, crazy, anarchistic freedom. But we will probably also overcome these problems as well, as in the longer run it will be corporate controlled, almost like a somewhat more interactive TV. For a lot of people, the Internet basically became Facebook. Much of what they are interested in is controlled by one corporation. And Facebook AFAIK is trying to become more media like.
I don't think we will lose the anarchistic freedom; it's too powerful. The internet has toppled enemy regimes; to shut down the anarchy would be to lose an incredibly potent weapon. Given that it's a weapon that can be used without significant diplomatic issue, I'd guess the internet is a more powerful weapon than nuclear bombs at this point, and anybody who actually takes significant steps to shutting down that anarchy is at least to some extent disarming themselves. A foreign state can't do much of anything when your citizens are fomenting revolt among their citizens. With anarchy, a state actor is indistinguishable from a citizen.
Looking at places like China, it doesn't seem like they are disarming themselves.
They didn't actually shut the anarchy down, they just made it mildly inconvenient. But a hacking attempt comes from China. Does anybody -not- think it was the Chinese government?
Me. China is big. In particular it has lots of people. Lots and lots and lots. And then some. A bunch of them are hackers for the same reason there are many hackers in Russia and Eastern Europe, for example. Some work for the government. Not all do.
In EE it tends to be connected with a certain... not openly rebellious but still anti-authoritarian attitude set, of gaming the system, getting away with bending or breaking the rules and going in the window when the door is closed. This is very similar to the Latin American https://en.wikipedia.org/wiki/Jeitinho and https://en.wikipedia.org/wiki/Malandragem although I don't know of a specific name. I don't know if there is anything similar in China. They look like a very disciplined type of culture... Also this thing should deserve an English name now. Perhaps System D. https://en.wikipedia.org/wiki/System_D
"by hook or crook"? "whatever it takes"?
True. But if they truly shut the anarchy down - prevented VPN and proxy connections to external anarchy - would that still hold?
No, not really, because you can't "truly shut the anarchy down". Consider, say, the night streets of New York City. Is there anarchy there? Certainly not. And yet there is some level of crime, you can buy drugs, you can get mugged, etc. If you get shot in the Bronx, it doesn't mean that a government agent shot you (though sometimes that is so). "Shutting down the anarchy" means getting it down to meatspace level.
Meatspace analogies don't apply to bitspace. Autonomous government agents are expensive in meatspace, but have only a marginal cost in bitspace if you prohibit secure-against-government-search communications, which is easy to do once you've committed to creating said autonomous agents to identify anybody attempting to engage in them.
That's a very big IF and I'm fairly optimistic that the cryptography genie will be hard to stuff back into his bottle. Even if you try to enforce plaintext-only communications (which by itself leads to a host of issues), I can stuff a wide communication channel into the lower bits of cat videos (and such) after which we are off to the arms races and your "marginal cost of governement agents" becomes not so marginal after all. In any case, if we get to THAT totalitarian society, we'll have bigger things to worry about than the freedoms of the internet.
You don't need to catch every secure communication. Even a 1% identification rate is enough; less if you're willing to toss in some traffic analysis into the mix. Your goal, after all, isn't to prevent secure communication, or even to identify what's inside it, it's to identify the people doing it, because the secure communication itself, rather than the contents of the communication, are what you've banned. But I don't think anybody -wants- this level of control. The Internet is too powerful a weapon.
That's why people are busy constructing dark nets.
Personal opinion: less has changed than people think. The walls of Pompei have preserved graffiti on them that amounts to 4chan trolling. Political parties once published their own newspapers and newsmagazines that pushed an explicitly partisan viewpoint and served as people's only access to information about world affairs. Most people are fine, provided that you get them in the right circumstances and mindset. What's amazing about the world's problems sometimes is how you occasionally find out How Problem X Happened, and while it wasn't a conspiracy of card-carrying villains in a smoke-filled room, it often was one person in an office who was feeling particularly malevolent or uncaring, and was able to make up rationalizations they could spout in public without actually being fired from their job. Like racism. Maybe there's an evil genie out of its bottle that causes epidemics of crime and violence and the police are stuck in a bad incentive gradient. Or maybe someone in power is, deep down, just as much a hateful jerk as your average YouTube commenter, and the ever-so-tragic Unavoidable Genie Problems are actually just that one hateful jerk forcing his hateful-jerk policies on whole groups of people.
Examples? The war on drugs seems to have been the result of a small number of people. So far as I know, the trail of Tears was something Jackson pushed through that the rest of the country didn't especially want.
Or maybe populations with low IQ's are simply more prone to violence. Especially, when there's a reluctance to punish them the same way as high IQ people for fear of being "racist".
And then the American Irish replaced their hardware from low-IQ to high-IQ in a course of 150 years. Wait, no, looks more like a software change: http://www.city-journal.org/html/7_2_a2.html
Your link says nothing about IQ. There are many ways to stop being poor other than raising your IQ.
See the recent discussion of echo chambers in here.

Is LessWrong interested in Bayesian machine learning introductory articles?

I would be interested in that, don't know about anyone else.
In a related matter, if I want to study the information theory of hierarchical Bayesian models, what definition of information should I be reading up on? Shannon entropy seems to only apply to discrete distributions; mutual differential entropy seems like the thing I want, but has "weird" properties like occasionally being negative; K-L divergence seems more like a tool for hypothesizing things about out-of-model error than for examining links between random variables. What I really want to do is: set up a statistical model, and ask a well-defined, well-informed theoretical question about "How much information do I get about random variable Y if I measure random variable X, where the relationships between their distributions are specified by this model I've got here?" Having asked the theoretical question, I want to maybe derive some maths (an analytical expression/equation for the question I asked would be nice), and then run an empirical experiment where I construct such a model and show how the equation holds numerically. The ultimate goal is to find out precisely how the "Blessing of Abstraction" mentioned in Probabilistic Models of Cognition is happening, and characterize it as a general feature of hierarchical modelling.
Short answer: use differential entropy and differential mutual information. Differential entropy and Shannon entropy are both instances of a more general concept; Shannon entropy for discrete distributions and differential entropy for absolutely continuous ones. KL-divergence is actually more about approximations, than about variables dependence; it would be strange to use KL-divergence for your purposes, since it is non-symmetric. Anyway, KL-divergence is tightly connected to mutual information: %20=%20KL(p(x,y)%7C%7Cp(x)p(y))) Differential mutual information is a measure of variables' dependence; differential entropy is a measure of... what? In the discrete case we have the encoding interpretation, but it breaks in the continuous case. The fact that it can be negative shouldn't bother you, because its interpretation is unclear anyway. As for (differential) mutual information, it can't be negative, as you can see from the formula above (KL-divergence is non-negative). Nothing weird occurs here.
Brilliant! Thank you so much!
Seems right in the core of things LW is likely to be interested in, to me.
I for certain would be interested.
Yes, at best with a how to of applying the concept in python or R.
The phrase, "DO IT!", spoken in an excited tone, comes to mind.

When studying history I sometimes find the hardest thing for me is wrapping my brain around how people actually thought back then. I'm so ingrained with modern Western science-based thinking that it's really hard for me to envision how people outside that box actually think. Can anyone suggest some books or articles that explain the differences in modes of thought between us modern educated Westerners and other cultures / time periods?

Edit 1: I think what I'm looking for is something like the following book, just on current vs. past cultures and/or culture... (read more)

I have an anecdote related to the understanding of historical mindsets.

Firstly, I have spent the majority my evenings the last ten years either inside buildings or along well lit streets in cities. I.e. my description of the night sky would basically go: "it's mostly black, sometimes cloudy". Whenever I have read about celestial navigation, I've thought: "That's clever, but how did they figure out they could do that?"

Come last winter, I took part in a cabin trip. The air was very dry, and the sky was cloudless. When we arrived in the evening, more than an hour's drive from the city, it was pitch dark (you couldn't see your feet). What struck me -- the way a brick strikes one's face -- when carrying stuff from the car to the cabin (walking back and forth, turning around, etc.) was this: "Of course humans have looked at the stars since forever. The stars (and moon and planets) are the only things anyone can look at at night. My eyes are drawn to them whether I want to or not."

And: "When I turn around, the stars stay the same. Of course people could navigate by looking at them --- they should navigate by looking at them!"

And: "Of course t... (read more)

It's often important to appreciate that the ancients may not have been actively stupider than us, but in fact just had much less information and much less computing power to work things out with. Huge portions of human history, including much the present day, have to get filed under, "Well, they were doing their best!"
On the other hand we know a lot about mineral and vitamin deficiencies and their effects on IQ nowadays so in many cases whole cultures were on average actively stupider than us. Perhaps an explanation of the Great Man theory of history is that for a long time history was dominated by the few who happened to be lucky enough to avoid any severe nutritional deficits.

Eric Havelock's Preface to Plato is basically about the differences between pre-literate cultures and literate cultures. It is also a very engaging read. Ditto for Walter Ong's Orality and Literacy. And a very engaging novel about a contemporary pre-literate culture is Mario Vargas Llosa's The Storyteller.

O. Neugebauer's The Exact Sciences in Antiquity is a great overview of mathematical and astronomical thinking in ancient times, particularly in Babylon.

For an exceptionally engaging account of a melding of scientific discovery and religion in ancient times, there is David Ulansey's The Origins of the Mithraic Mysteries.

And Karen Armstrong is an engaging writer on religious thought in times past; especially The Great Transformation.

I'm not sure whether books and articles are the best. Talking to people is often more efficient when it comes to understanding people much different from you.

Visiting the Amish would provide you such an opportunity but you might even find people in your own city with radically different modes of thought.

If you go to some New Age event you find a lot of people who don't follow science-based thinking.

You seem to be interested in history of mentalities. Try something that is classified under this label. I can recommend "The Great Cat Massacre" by Robert Darnton [1], which is a short and interesting book about 18th century France that is quite easy to read since it is based on a course the author taught at Princeton University. A few years ago I have read "A history of European mentalities" by Peter Dinzelbacher, which covers a lot of topics and seems to be very close to what you want, but I am not sure if it is available in English. [1] Curiously, there is another book by the same name which is about a different cat massacre.
Look for original sources-- that way, you'll be getting samples of how people used to think rather than interpretations from more modern people. You might want to try Ex Urbe-- Ada Palmer is good about how much the way people think has changed.
I've actually done this on several occasions. But I mostly just get the feeling that "something is wrong here" and I can't quite put together how anybody could actually think what the source is saying makes sense.
Forget history, I have a lot of trouble understanding how a large part of the human population currently thinks.
Their subconscious thinking is mostly pretty normal and similar to yours, but they've been given very, very different maps of the world that make their verbal self-expression come out much, much differently from yours.
See the chapter on Synge in Declan Kiberd's Inventing Ireland, the references to his stay on the Aran Islands: '...Synge records all this with a terrified and terrifying accuracy, because he knows that, however spare and beautiful such a culture may seem to the outsider, its costs in human terms are just too high.'

Is nominalism pretty much the same thing as the map-terrain divide? Can someone try to steelman conservative anti-nominalism such as Weaver's Ideas Have Consequences or this ?

I mean, non-nominalism i.e. Platonic idealism is just trying to reify parts of the map, trying to project human categories into reality, and this is fairly obvious, isn't it? Or I am missing something? Currently nominalism - if I understand it correctly: the idea that categories are man-made - is a fairly obviously true idea, considered true by anyone who ever heard about a map-terr... (read more)

On Skeptics stackexchange there's an ad election. User can vote which ads are supposed to be shown.

At the moment top scoring entries are about http://stopavn.com and the wayback machine.

There's an entry for LW which is currently at three votes. If you have a Skeptics account voting is just a page away.

What happened to the rationality diaries? It's been a month since the last one officially ended.

No one's in charge of posting them, so it's probably just a case of "no one thought to post one". There are multiple possible reasons why that happens to rationality diaries and not to the open thread, but I don't think most of them imply that the rationality diaries are pointless or anything.
The latest post in that old one was on the 20th, so people are getting enough use out of them that continuing to have them posted regularly is worthwhile - I thought about it, but wondered about whether there were any rules about it, started thinking about posting in the open thread asking whether there was anything I'd have to do first, and then didn't bother to go to that effort.
Feel free to create one. You may do so even for your own diary entry.

What I did at work today: analyzed HPMOR and wrote a blog post, "Harry Potter and the Methods of Latent Dirichlet Allocation".

Starting a story-like object on ems

I'm experimenting with a text in the form of a FAQ that might be usable as a "story" in its own right, based on the idea of an em open-sourcing himself. I would welcome any and all feedback you might care to offer, either at https://docs.google.com/document/d/1nRSRWbAqtC48rPv5NG6kzggL3HXSJ1O93jFn3fgu0Rs/edit or here.

Thanks for bringing it to my attention! Having an interest in visual novels, interactive fiction and generally all forms of experimenting with good ol' prose that just might end up advancing 'state of the art' of fiction (I suppose I should call that ambition 'upgrading the prose' - have you seen this or that?) I was immediately attracted by your term 'a story-like object': I am interested in exactly the kind of stuff that might be hard to put a label on. I decided to take a peek, and read through it all. Didn't expect to enjoy it like I did. You know, something like this might lie at the beginning of a novel or game, but us common readers do not get to read it. You know, there's a peculiar game studio from London called Failbetter Games, makers of Escapist's 2009 "Browser Game of the Year" called Fallen London. That's a kind of game studio that, when deciding to make an expansion to their game, starts with hiring writers; Fallen London is over 1.5 million words. When asked how they keep track of their established canon and avoid overwriting other plots, they say: I presume you are not sure what it is you're writing? If you were to ask me, this story-like object of yours is not a story. Rather it has the trappings of just such a guideline for writing a novel or a series, maybe even written collaboratively by several authors. It has (i) and maybe (ii), too: these individual FAQ paragraphs could as well be made those spreadsheet cells, ranked from 'public' to 'unrevealable secret' and from that you would start picking and choosing details and making that into a story, game or whatever you call it; short story or novella won't even have place for all of the detail; whatever gets shown would be decided as much by the laws of narrative as by the inner logic you present here.
Thanks for bringing it to my attention! Having an interest in visual novels, interactive fiction and generally all forms of experimenting with good ol' prose that just might end up advancing 'state of the art' of fiction (I suppose I should call that ambition 'upgrading the prose' - have you seen this or that?) I decided to take a peek, and read through it all. Didn't expect to enjoy it like I did.

I noticed a thing that I do. When I rush; I have a tendency to do clumsier versions of actions I know really well. I have now trained myself to notice moments of rush, and purposefully "slow down to normal speed" on tasks to allow them to happen in the efficient most possible time.

Simple example, searching for a key in a bundle. Where rushing causes fumbling which takes longer, slowing down to "normal speed" makes the finding the right key happen sooner.

Is there a name for this process? Has anyone recorded it before? Is this a suggestio... (read more)

When I'm in a rush and I'm about to do something that requires carefulness, attention, and "normal speed" as you call it, I tell myself that, if my purpose is to get done with it as quickly as possible, I should deliberately slow down, because more often than not, botching up something and then fixing it takes significantly longer than doing it right at a normal speed.
I suppose it depends on the model of risk associated: where the whole task might be a minute; taking another minute to fix things up is a 100% time increase. Where a task that might not be rushed; something that takes an hour; an extra minute won't be as big a change. so not as bad a change? (but this is something of a different effect)
I think this is a commonly known thing, and the common advice is to calm down and breathe deeply,
I am not talking about "big rush" situations; but rather "small rush" situations. not sure where to find more about it. Surely a rational brain would have concluded the inefficiency and stopped doing "small rushing" (not that we are rational or anything - but maybe someone documented it before me)
Do you have a heuristic for differentiating big rushes from small rushes? I think any time you are trying to perform a task, and some epsilon greater than zero of your conscious capacity is focusing on the ticking clock, then that represents a deficit from maximal focus. I think the deep breathing advice is good for any rush. https://intelligence.org/files/CognitiveBiases.pdf in part 8 has a note on time pressure increasing the effect of the affect heuristic, but it doesn't quite fit with what you are talking about (fumbling for keys).
I don't think speeding up processes is always bad. It's just not always an appropriate emotional response to the problem one faces. Speeding up the typing of password makes sense when you know them well enough to make no errors. It's often problematic when an uncomfortable emotion causes one to rush and one tries to rush to escape the uncomfortable emotion. When it comes to task such as putting on clothing it's also useful to do them a few times in slow motion to allow the body do find better ways of doing them.
(I get my password wrong when I try to type it too fast, conversely I have made one spelling error when writing this sentence, I have probably typed this particular password upwards of 1000 times)
"Slow is smooth, smooth is fast."
The following proverb comes to mind:
Another proverb: "More haste, less speed"

A claim that local violence (individual crime and enslavement) is now a primary cause of extreme poverty, and it's crucial and possible to improve police and government enough to make a big difference.

Details needed. Social interventions often fail, so it is important to explore what exactly was done right in this specific project. EDIT: Seems like the strategy is telling the target country "okay, we will show you how to do it our way, plus we will provide resources and oversight". And if the country agrees, their policemen will receive training and resources (e.g. cars), the government will receive expertise on how to discover corrupt policeman, and the organization will apply some pressure to make sure some of the criminals are punished.
Echoing "details needed" on the method of measuring "commercial sexual violence against poor kids." Some agencies have reduced reported crimes by reducing reports rather than crimes or otherwise changing the map rather than the territory. Assuming that question away, related measurement question: do efforts that reduce commercial sexual violence against poor kids generalize well to reducing other crimes, and would we expect these results to replicate and generalize in other communities? Pilot programs (that we hear about; publication bias) are often the best case scenario for program effectiveness, rather than results we can expect elsewhere. ... which is a lot of words to ask how serious we are about the "and possible" in Nancy's one-sentence summary.
This seems to be about third world poverty. They speaker seems to be from the charity International Justice Mission. It's on GiveWell's list as "Considered not contacted"

Repost due to lack of reply: I’m a fourth year PhD student in the life sciences, and I need mentorship, preferably from a Slytherin, or at least someone with a Slytherin hat. My advisor doesn’t want me doing “mercenary collaborations”, or quick experiments with researchers outside my field in exchange for secondary authorships. He says I need to focus on my thesis research in the next year so as to publish and graduate. Are there any academics in the LW readership who have the insight to tell me whether this is good advice or whether he just wants me pumping out papers with his name on them so he can get tenure?

Given the lack of reply the last time how about posting the question to http://academia.stackexchange.com/?
Thanks, doing so.
No one else is probably going to tell you this, but go ahead and do those collaborations. At the end of the day, your thesis is only going to be read by 3 people, and a doctorate degree is just a degree and isn't important in and of itself. The important things are the work and publications you get done while doing your degree. So unless you doubt your ability to get your thesis done on time, you should be looking on doing all the collaborations you can get your hands on. You are the best judge of your own capabilities here.
I am a slytherin academic supervising a couple of PhD students now and I endorse the above message. If you finish your PhD, collaboration with others is key to future career development. If you don't finish, though, it's another story --- so make sure you get through. You can always forget to tell your supervisor about some of your collaborations. If the "mercenary" papers are unrelated to your field, it would also be advantageous to start thinking of ways to portray them as related - this will be helpful in presenting your research as a coherent whole, rather than giving the impression you are just helping on others papers for the sake of your CV. Or, as identifying a specific area of expertise that has cross-disciplinary applications. Finally if you are not able to portray your mercenary papers as related.... maybe your supervisor has a point and you should work on developing collaborations that offer more than just a one-off publication. I have a couple of "mercenary" publications from my PhD period (no further collaborations beyond one paper) and they stood out like sore thumbs in my CV when I applied for postdocs.
I think the advice "do whatever you have to do to publish and graduate" is good advice. Whether this is better achievable through secondary authorships (which also are very valuable for networking -- you'll need a job once you get your Ph.D.) or through focusing on your thesis, I can't say. It depends on your field and the specifics of your situation. But you need to be focused on graduating -- "almost there but not quite" for many years is a very common failure mode for grad students.
Are you sure you have the right advisor? If you are so concerned that his interests may be that divergent from your own, it might be better to switch advisors than to continue with this one. Ideally, you want an advisor who already has tenure and looks at you like a son or daughter, and will advocate for you as your career progresses. If that's not the vibe you are getting from your advisor, then look around and see if there's someone whose personality and style of work better suits you. There will be costs to switching advisors, but they may not be as big as the costs of continuing with a mismatch between advisor and student for more than a year.
I'm in a PhD program in the life sciences and although I haven't graduated myself (still in my first year), based on almost all of the advice I've read, this is good advice. The way to prove to your advisor that you deserve to be able to do quick secondary experiments with people outside of your field is to submit a paper of your own, and then you can do those experiments in the interim while you're waiting for the reviews. That said, I don't think I'm all that Slytherin (although I admit that this is what a Slytherin would say).
I already am published. When we got the paper accepted, instead of congratulating me, he said "Now you need to get another one before you graduate."
Ugh, that's awful, sorry. It seems like you have a pretty complicated and frustrating situation -- feel free to PM me or email me (gmail: amckenz) if you want to talk more.

Kurz Gesagt explain the Fermi Paradox in a nutshell.

I think this wouldn't be a bad introduction to the concept for people new to it. It seems to be based on Tim Urban's Wait But Why post - maybe they'll do superintelligence next.

Can anyone recommend good sources on the social dynamics of witch-hunts?

Not necessarily about witches, of course. I'm interested in the hand of Moloch in these situations: social incentives to go along, status rewards for being more morally outraged than your fellow citizen, self-protection by avoiding looking insufficiently outraged, the not necessarily intended but still unescapable prosecutorial traps, the social impossibility of denying the actual existence of the outrageous facts...

In sociology literature this will show up as 'moral panic'.

I'd like to see clearer in the issue of narcissism in the broader, not strictly in the clinical definition sense. It often argued that it is a typical problem in the current age. Lot of young people believe their parents are. But outside the typical stereotypes of narcissism, such as having flashy looks, in the broader sense, even something like being shy can be interpreted as a form of narcissism, as extreme self-consciousness, extreme self-awareness, thinking everybody is looking at you, in a disapproving way.

Can anyone recommend an article or ten to so... (read more)

I've seen that picture, but I didn't take it as bragging. IMO the poster just thought it was funny how the dog looked, and he just mentioned the cleaning because he thought if he didn't, he'd get hundreds of replies telling him he should've helped the dog instead of laughing at it.
Self consciousness =/= narcissism IMO. A person might be self conscious for many reasons. I don't think you can really get at the underlying reality of what you're trying to explore if you use words like "narcissism". I'd take a step back and look at Cluster B personality disorders as a whole to get a better idea of the clinical dimension of "narcissism" in a broader context, and to dissociate the popular meaning from the clinical one. There's general deficits in emotional processing, the ability to take the perspective of others, and various vmPFCish thingies of that nature, and sometimes those present in a certain stereotyped fashion labeled "narcissistic". I'd characterize it as more a deficit in self awareness and other awareness, with the rationalizations filling in the blanks left by that deficit sometimes but not always creating an inflated self perception. Rejection sensitivity, abnormally harsh self evaluations and feelings of inferiority can also fall into Cluster C, especially "avoidant", and depressive states. I think the difference is that with Cluster B, there is at least some level at which a person is trying to defend a positive image of themselves even if they have deep-seated insecurities. There might be periods triggered by negative external events causing them to express some extremely negative feelings about themselves (which they later recover from and can be talked out of) but for the most part they like themselves. In contrast, an anxious, avoidant or depressed person just consistently insists that the low self evaluation is correct and it's much harder to talk them out of it. I think the briefest way to say it is that Cluster b is associated with basically normal emotional states, but those emotional states do not react normally to environmental cues, and the specific abnormality is what further characterizes them. In the stereotypical and stigmatized case of narcissism, a person fails to accurately self monitor their own behavior and mo
I don't know if it is the official definition, but after reading the links I guess it is about thinking "me, me, me" all the time. And the trick is that it does not have to be "look how great I am" -- because some people believe that it is okay to spend 24 hours a day thinking and talking about themselves (and how other things are related to them) as long as they avoid talking about their own greatness. The authors of the linked texts believe that "greatness" is a red herring. For example, the narcissist can also talk a lot about how they feel bad -- but when a narcissist does it, the focus is always on their emotion of feeling bad, not on what makes them feel bad. It is difficult to define exactly. For example, two people can say "I feel so sad about children in Africa starving!", while one of them means "please look at those children, and perhaps try to find out something that would reduce their suffering" the other means "look at me and admire me for how altrustic and noble I am". If you are not careful and don't know where to look, you might miss the difference, because technically both of them are talking about starving children in Africa. -- Actually, I suspect that people who didn't have previous bad experience will usually miss the difference, because they will automatically assume that the only reason why anyone would mention starving children is because they care about the starving children. While the narcissist is only thinking about how expressing care about the starving children could make the narcissist more awesome. Depends on context. If someone did it for the first time, they could be happy for having the experience, and want to share the happiness. -- I am proud of myself for being able to change my baby's diapers quickly. Before I had the opportunity to practice, I didn't know whether it will be easy or difficult, so it was a pleasant surprise to find out that it is trivial. That's good for me and good for the baby. And I am not writing this to
Of course if you look at a term in a broader sense you can look broadly enough that everything qualifies. What purpose do you want to achieve by having a broad notion of narcissism?
Mainly whether it is more widespread now, a generational illness or not, does social media seem to increase or enable it, and do I or my loved ones "suffer" from it and to what extent. Maybe focus on the social media aspect, perhaps it is the best approachable. Suppose social media is drifting towards validating posts like "When I was done laughing and taking pictures, I helped him clean his beard." What class of behaviors are this an instance of? Of course an LW favorite would be "status seeking" but I think is, at best the other kind of status because nobody gets actually useful, usable social status through this. It just feels that way - and is this felt-status as opposed to real status something that maps to narcissistic supply?
It's a fair starting point to assume people are always wrong when they talk about generational pathology. It's the sort of thing that's possible, insofar as the social environment matters, but nostalgia goggles and technical changes totally dwarf it in terms of plausibly explaining any particular generation gap.
If you wan to know whether narcissism is now more widespread, why not use clinical definitions of narcissism. Do people on average score higher or lower on the Narcissistic Personality Inventory than they did in the past? That's a specific measurable question. There no point in having a broader notion. Playing reference class tennis is not useful as an end in itself. If your goal is to judge people as being bad because they are narcissistic but wouldn't be labeled that way by well researched academic scales by more clear about your goal.
I think that the point is just to be able to talk about the thing that people mean when they talk about somebody being narcissistic, which generally is not the clinical diagnosis.
It usually means that the person doesn't like the person they call narcissistic. If Alice calls Bob narcissistic that might tell you more about Alice then it tell you about Bob.
On the off chance that you haven't heard about it, I would recommend Christopher Lasch's The Culture of Narcisicism as background reading if you enjoy TLP. Lasch treats narcissism as a more general cultural phenomena instead of a strict clinical diagnosis. It is over 30 years old now, it probably doesn't directly apply to modern topics like social media.

Has anyone here used fancyhands.com or a similar personal-assistant service? If so, what was your experience like?

(context: I have anxiety issues making phone calls to strangers and certain other ugh fields, and am thinking I may be better off paying someone else to take care of such things rather than trying to bull through the ugh fields.)

Since I am interested in forecasting, anyone has links to good analyses as to why UK pollsters fell straight face into the mud this time?

Link one, mostly mea culpas, a bit of analysis.


Anyone esp. @CellBioGuy having anything interesting to add to new method to predict cancer? http://www.telegraph.co.uk/news/11574893/New-test-can-predict-cancer-up-to-13-years-before-disease-develops.html

Strangely, shorter telomeres is not actually a new idea here AFAIK. I would also be interested in a very rough guess of whether this may be an expensive or cheap test.

Check the following figures: http://ars.els-cdn.com/content/image/1-s2.0-S2352396415001024-gr1.jpg Biiiiig cloud of points showing a statistically-distinguishable trend by which those in the cancer group had telomere length attrition with age slightly higher than those that did not get cancer. But the variation from patient to patient is much larger than the average variation with age. http://ars.els-cdn.com/content/image/1-s2.0-S2352396415001024-gr2.jpg In contrast to the previous trend with age, average blood telomere length in patients who later went on to get cancer was statistically distinguishably longer for ~3 years before cancer diagnosis. Again, population averages. The distributions are not shown but I would not be surprised if variations are large enough that for most values a diagnostic test wouldn't be terribly conclusive. Blood is a funny tissue, its stem cells maintain telomerase activity and the balance of telomere extension and wear-down can be shifted one way or another by conditions. This could be a readout of other deeper things happening elsewhere in the body at a systemic level or might reflect a generalized telomere extension effect in the early stages of cancer as a contributing cause or an effect.
What's the contrast? Isn't the situation exactly the same, statistically distinguishable, but practically meaningless?
The contrast is the type of trend. In samples from people who got cancer age had a greater correlation with short telomeres. However, in samples from patients in the three or four years immediately before cancer diagnosis, there was a significantly longer telomere length. People who will get cancer's telomeres shorten more rapidly with age, but are longer right before they get cancer. It's a dynamic process and it's apparently got some weird complications to it. Mostly interesting to me because it suggests body-wide mechanisms of telomere regulation that have something to do with cancer genesis.
Oh, whether short vs long is associated with cancer. I guess that should have been clear from De Vliegende Hollander's query. On a different note, that second graph looks like nonsense to me. The cancer group holds steady, while the control group bobs up and down. The effect is in the control group, not in the cancer group. This graph tells nothing about cancer. 95% it's pure nonsense, but maybe it tells something about the control group - it's predicting their censoring by death. Which is why it's crazy to use time to censoring as the metric for the control group.
Just because a mainstream news article says that something is a new method doesn't mean it is. If you use the word "strangely" when reading a mainstream media article about science it's a good heuristic to instead go to the actual study. In this case http://www.sciencedirect.com/science/article/pii/S2352396415001024 (open-access) Rather cheap. They describe their method: Taking blood samples and running PCR are both not complicated things. A bit googling around finds 400$ as a retail price (http://www.repeatdx.com/products/pricing-and-cpt-codes/) for telomere testing. There seems to be a failed Indigogo campaign that used 99$ as the default price but was target at salvia instead of blood.

One month ago, I started being treated with Humira for my Crohn's disease, which put and end to a roughly three-month period of waking up several times every night to go to the loo, and since then I seem to have needed fewer hours of sleep per night than ever before. In the past it's always seemed like the natural amount of sleep I'd have if it wasn't cut short was about 9 hours, and I've recently seemed to need so much less that on two occasions in the last month I've been almost frightened by how early I've naturally woken up - "I thought I was slee... (read more)

In general unhealthy people need more sleep than healthy people. Removing a factor of stress for your body can reduce sleep needs.
I suppose I should have been more explicit than just to use the wording "in the past it's always seemed"; I'm not comparing how much I sleep now to how much I slept while I was clearly less healthy, that wouldn't be a weird thing I'd need to ask about. I'm comparing to about a year ago, when my condition was about the same as it is now. The effect seems to be [slightly unhealthy, sleeping 8-9 hours a night] -> [very unhealthy, sleeping 8-9 hours a night with frequent interruptions] -> [slightly unhealthy, sleeping 7-8 hours a night]. And I'm not sure what my priors should be for whether that's actually what's happened.
How old are you? If you are 18-25 years old, you are still physiologically completing the transition from adolescence to adulthood, which means at some point your circadian rhythm will "shorten" and you'll start tending towards early to bed / early to rise, with both a shorter natural sleep phase and a shorter natural awake phase. I noticed it happening to me pretty suddenly - my sleep patterns noticeably changed sometime during college despite no major lifestyle change. I wouldn't be surprised if dramatic changes can happen within a year. As far as I know it's permanent! The nasty downside is that, whereas the adolescent body would refuse to sleep on time but once asleep continued sleeping even until noon and I always woke up refreshed, the adult body has no problem sleeping early but will wake up at a set certain time and refuse to go back to sleep, regardless of whether I'm well rested or how late I went to sleep. So on adult-mode I actually have to go to sleep on time if I want a productive day tomorrow, which was never an issue for me in adolescent-mode. "Going to bed too late" becomes equivalent to "waking up too early". So, if you buy the circadian explanation, then don't take "waking up naturally" as a sure sign that you've slept enough!
I'm 21, so I was considering something like that as the most likely alternative hypothesis, and the extra details about waking up are also consistent with what I've noticed, so yeah it's probably that.
I might suggest that seasonal change in sleep-requirements is enough to add 2 hours; or take 2 hours away. Don't get too excited, and also - start sleep tracking; easiest way to find out if you really do need less sleep. I use a Basis watch to observe my sleep cycles and a fitbit to track total sleep (and the two of them in case of failure). There are apps that also track but I find they don't work with my pre-bed lifestyle (be awake in bed till I fall asleep, usually on my phone); and remembering to set the app was causing it to not work for me. (I also tried a zeo and I managed to take it off during the night) The 3rd thing I would say is - trust your body. try to optimise for a lifestyle that does not need alarms; and trust your body to wake up when you are rested. still sleepy? go back to sleep. I can send you a graph of my sleep/time if you are interested.

Interested in what you guys think about this. Jayman(hbd blogger) say's parenting has no effect on how children turn out. Seems empirically incorrect to me and it's just probably difficult to encapsulate the results/hard to see non-linearities to make it easy to reference.

He insists on twin-adoption studies contrary to my views.

Thoughts? This sort of seems like the two cultures divide we agree on. I might make a thread just for this.

Argument: Does parenting have any effect on child outcomes?

  • His view: Zero effect & Breast milk confers no advantage eit

... (read more)

Jayman is correct that adoption studies typically show negligible parental effects. But remember the studies can only talk about the environmental variation present in their data, and are generally done on normal, western, middle class cohorts. In studies where they include stronger environmental variation - e.g. Turkheimer et al 2003, you find that it matters.

So basically, the kind of parenting choices that people typically worry about are probably meaningless, but severe trauma, poverty, abuse etc. do matter. That being said, You can't just say "X is difficult to encapsulate" with studies. This is a fully general counter argument to any evidence you don't like.

That's not what I'm saying. I'm just saying that even if a small percentage of the population took that advice as definitive and executed over it, all latent small probabilities would be realized and the mistakes would be immediately manifested. Stuff like that isn't being difficult as any pharmaceutical company or any medical company has to know for sure that their studies/devices are correct for health effects. In fact in statistical decision theory texts pharmaceutical examples are standard. tl;dr this would be a big lawsuit of some sorts
I think you need to be more specific about your criticisms here... I don't see why the twin study design, and extensions thereof, don't allow such inferences. The cool part about them is that they exploit some very unusual properties to do what they do. (Unusual enough that when I went looking for uses of twin designs outside of behavioral genetics to get inspiration for a diet self-experiment, I couldn't find any.) If there is some variable effect, then there should still be some average effect. I don't think that's true. Could you explain how the twin design requires that, keeping in mind the fundamental logic of the design?
Thanks for the reply I'm saying in general with probabilistic inference not necessarily to twin study design. I think I worded it relatively poorly. I'm saying there is that is not necessarily encapsulated in the study because there would have to be some relatively exhaustive statistical machinery applied. When someone says something with such strength that there is zero effect(ask him on twitter) that means there is no uncertainty and we know all relevant variables and have assigned probabilities and utility values to them(everything is known). Which is false. We do not know if we have missed something related to for , because we have not done exhaustive statistical studies on it. we probably only have so much empirical knowledge related to it. His statement is even stronger that there is no uncertainty, he is saying there is no risk to executing over the advice that parenting has no effect and therefore we can just go do whatever we want with parenting. Which is equivalent to assigning the utilities to all equal zero or the integral to zero. He is saying there is neither uncertainty, nor risk. A bit long winded, I think I might ask Mrs.Mayo if I have it right. To say a statement with such strength of and that it is advice that could be executed over, say ten thousand people stop parenting their kids would mean that any negative effects that were not captured in the study would be realized no matter how small the probability, that we missed something not encapsulated in the study or something in general. It's just a problem with executing over empirical advice in general. It's hard for me to state declaratively. For example I think we could argue that the fact that these people exist to be studied means that there is a survivorship bias accordingly(for those that did not get studied are dead/did not show up) and that that weights the probabilities in the favor of the people who survived but if we evaluate the entire sample space(dead+alive) it would be in fav
Could you link to the twitter post where he describes his position? Without a clear reference to the claim under debate this exercise isn't useful.
He is just straight up naively saying parenting has no effect and the consequential statement that any one can act as if this is completely true. This is not an exercise, this is trying to convince an actual person why he is wrong. I'll ask him for a clarified picture right now. https://jaymans.wordpress.com/category/parenting-2/ https://jaymans.wordpress.com/2014/12/21/iq-and-birth-order-effects-real-no/
That's different then claiming that the effect is zero and that you can completely mess it up without bad effects.
The first comment is equally as strong. I'm having a running dialogue with him and he will come here and correct/clarify if necessary. Please focus on the topic at hand. For once can we actually have a discussion? edit: https://jaymans.wordpress.com/2013/09/25/no-you-dont-have-free-will-and-this-is-why/ My framing is coherent and correct editedit: he said read this post https://jaymans.wordpress.com/2014/03/31/the-son-becomes-the-father/
When the goal is to focus on the topic at hand it's vital to understand the position on the participants. Do you agree with the claim "Parenting is simply less important than most people think"?
Breast milk "confers no advantage" compared to what? (baby dying from starvation? water? cow milk? soylent? infant formula -- which one?) All that stuff that exists in breast milk -- such as immunoglobulin A, lysozyme, blood albumin, creatine -- it's just completely useless, I guess. I wonder why evolution even bothered to design such system, when it provides no advantage. (This is a sarcastic way to say "my priors for this hypothesis are very low".)
Wait did you mean for Jayman's hypothesis is low or breast milk?
(I'm not Viliam, but:) For Jayman's hypothesis, obviously. The argument is: Here's all this complex machinery put in place by evolution; it's terribly unlikely that what it does is actually useless. (That would argue that breast milk is a good food for babies, which I don't think anyone denies. It's only a strong argument against feeding them infant formula in so far as we have reason to think that formula doesn't have in it those things that evolution helpfully put into human breast milk.)
Yes, low priors for Jayman's hypothesis. Although it wasn't sufficiently specified what exactly his hypothesis was, so I could be arguing against a strawman. I am not an expert on infant formulas, but I think that if someone could factory-produce a drink that updates your immune system (as breast milk does), that would have huge implications in medicine. Essentially, we could replace vaccination by drinking soylent.
There have been many randomized controlled experiments of breast milk vs formula. Every single one of them has shown no effect.
I just did a little googling and that appears to be untrue. What's the source of your information? The first hits from a Google search for <> were: * Scholarly articles (Google pulls out a few of these at the top of its search results): * A study on promotion of breastfeeding in Belarus. Intervention appears to have increased breastfeeding and also substantially (and significantly) reduced the incidence of two of the three health problems they measured. * A study on how often breastfeeding passes on HIV from mother to child. (Conclusion: about 1/6 of the time; using formula is an effective way of reducing this. Interesting but not relevant here.) * A study on the effect of "peer support" on making mothers breast-feed for longer. Not relevant here. * Ordinary search results: * First five hits are for the same study, looking at very premature babies. Conclusion was that mother's milk appears better than both donor breast milk and formula, and the latter two aren't much different. * Meta-analysis of studies comparing donor breast milk with formula. Conclusion is that unfortified donor breast milk (i.e., no extra nutrients added) leads to slower growth than formula, but formula leads to more necrotizing enterocolitis. Apparently donor breast milk is generally fortified nowadays. * Different meta-analysis looking specifically at necrotizing enterocolitis. Found only four relevant trials, from 20 years before. They were too low-powered to find significant results, but aggregating them finds a substantial and significant reduction in NEC risk for donor breast milk versus formula. * CBS News report on a study that found that giving a little formula to babies who lose a lot of weight shortly after birth may result in longer breast-feeding. Interesting but not obviously relevant. And that's the end of the first page. So it looks to me as if * There haven't been that many randomized controlled experiments -- fully half the first page of Google resul
http://www.bayesianrisk.com/chapters.html http://www.bayesianrisk.com/sample_chapters/Chapter%201%20There%20is%20more%20to%20assessing%20risk%20than%20statistics.pdf I agree with Viliam & gym. This just points to the limits of statistical knowledge & that we need to supplement with other logical-experimental knowledge, such as arguments from evolution. Risk cannot only be based on statistical knowledge, as chapter one of bayesian risk argues.
Always a good questions to ask.
You might be interested in a fairly well-known book, The Nurture Assumption.
I will read it but I must insist I do not have any 'nurture' assumption.
"Nurture" here is a synonym for parenting (nature vs nurture as in genes vs upbringing). The nurture assumption is that parenting matters. You seem to believe that parenting matters.
I mean, obviously parenting matters, right? You can't just drop kids in the wilderness. What's actually being said here is that if we remove the extreme cases, the people who aren't really fulfilling the role of "parents" as they should, the normal diversity we see among various types of parenting isn't really a big deal as far as outcomes are concerned. (This is still a bold claim, because, as you said, we might just not be aware of the slice of diversity that matters.)
What do you mean by "extreme cases"? One culture's "extreme case" is anther's typical parent.
In this cases, I think "non-extreme cases" basically means "the child manages to survive until the age of psychological maturity with no debilitating injuries". The claim is that the main role of parents is just to get the kid to adulthood in one piece, and anything else is extra and might not seriously help or harm commonly measured outcomes. One might add the assumption that the parents at least didn't actively inhibit ordinary non-parental exposure, to rule out things such as never learning to talk or something. (But remember, I'm not claiming this at all, just stating what I think the real claim is. I don't even think the more conservative version is correct: even if you assume that childhood experiences with parents controlling-for-differing-non-parent-related experiences aren't that important, how could variations in ongoing parental support in adulthood possibly not make a difference?)
I think dogs give us the clearest insight on this problem. Dog breeds tend to be vastly different from each other despite being genetically close enough to interbreed. Breeds have certain tendencies common that breed in terms of behavior, but if you really want complex stuff to happen you HAVE to train them. And on the converse side, no matter how well you train a chihuahua it will never be a great sled dog. I think the same logic can apply to humans. Humans are capable of self-training (eg an ambitious child can go and learn physics on his own), but parents have a huge amount of control of the kinds of training kids get as well as the kinds of social contexts they live in (which Judith Rich Harris in No Two Alike argues is the main explanation for why children sometimes end up very different from each other even if they are both born to and raised by the same parents). Humans (like dogs) learn and adapt to the kind of environment they end up, with their natural genetic advantages and disadvantages playing a hugely important role in how well they do so. Jayman looks at the failure of all the studies looking for parental influence and finds little effect, but I think he goes too far in dismissing it because he (like most of the researchers in the field) ignore the importance of context to behavior. This is not to mention that the kind of contexts and environments parents are allowed to put their children in these days are extremely constrained. Measuring different parenting styles on how well children do on standardized tests after going to standardized Prussian schools is of course not going to going to find that big a difference. But imagine arguing that kids will end up the same whether their parents apprentice them to a Blacksmith, send them off to join the Army, or give them to the Priesthood.
The 'dog breed' analogy seems to a common pop-sci way of talking about differences between human beings but there's no evidence that it is a useful analogy at all. The average genetic variation between dog breeds is far greater than between humans. Even the most dissimilar humans are more genetically similar than the average pair of dogs. But really, it's more correct to say that the genetic variation of humans and dogs is not comparable at all. Human genomes vary in quite different ways than dog breeds. On the converse side, it does lend credence to the idea that nurture is important.
I agree with your sentiment, but presumably there's even greater variation among wolves than there is among dogs, yet most wolves are pretty similar. The stunning variation among dog breeds is due to variation in selection, not variation in genotype.
Perhaps you mean that despite large genetic variation in wolves, there appears to be small phenotypic variation in them. In terms of morphology and appearance this is true, but there's no reason, in my view, to think that it's true in terms of behavior and intelligence and so on.
Here is a government study showing a markedly higher risk for perpetrating violent crime among children who were abused or neglected by their parents. So that's one effect. Bad parenting leads to bad outcomes.

That study is observational, not experimental. Maybe genes for disagreeableness make parents abuse their children, and they pass those genes on to their offspring. Probably both nature and nurture contribute.

Certainly. Correlation isn't causation. One hurdle is that any experimental study of this phenomenon would be highly unethical. But all is not lost. Single-parent households are also associated with higher risk of juvenile delinquency. I'll see if I can dig up a study of children abused by foster parents or step-parents.
Single-parent households are also associated with higher rates of parental divorce and teenage pregnancy. I think I remember reading what happens if you only look at single-parent households where the other parent got sick and died, but I don't remember the answer.
Not always. It might be possible that there are orphanage systems where children are randomly assigned to either orphanage A or orphanage B. If you go to Africa to set up such a system you can add special funding to one of the two orphanages to raise it's quality much higher than that of the average African orphanage. You can also split test different educational philosophies that way.
Here is a different study which says:
Indeed. One could also look at Sariaslan's population registry studies. Sadly, most of these sorts of studies can be written off with a single sentence: "includes no family design, therefore is useless and merely shows that heritable traits gonna inherit."
Certainly there are some things affected by parenting. A baby born in Japan to Japanese parents, but adopted in infancy by English speaking Americans will grow up speaking English, not Japanese. If her bio-parents were Buddhists and her adopted parents were Christian, there is a good chance she will be a Christian at adulthood. While I would say these parenting affects are substantial, I get the feeling this is not what you and Jayman are talking about. It would be helpful if you specified what precisely you disagree about.
The fact that he claims so authoritatively and certainly that there is zero effect which is hubris manifest. it is also incorrect for the arguments I have or at best rather incomplete.
The null hypothesis is always false, and effect sizes are never zero. When he says it's zero you should probably interpret zero as "too small to care about" or "much smaller than most people think". I'll bet the studies didn't say the effect was literally zero, they just said that the effect isn't statistically significant, which is really just saying the effect and the sample size were too small to pass their threshold. People say a lot of things that aren't literally true, because adding qualifiers everywhere gets annoying. Of course if he doesn't realize that there are implicit qualifiers, then he's mistaken.
Yeah but his abrasiveness of delivery is contrary to your goodwill. I'm being polite in his favor.
I think you're essentially right, but just out of curiosity, what are the 'twin-adoption studies' he refers to? As far as I know, there are really very few twins where one twin has grown up entirely separate from the other.
Yes, separated twins are rare. But twins and adoption by themselves are common and can be studied separately. Either one alone can separate nature from nurture. And they both show the same result: 50% genetics, 0% parenting.
In what way? As knb mentioned above, parenting has a direct effect on e.g. what language you learn. On the other hand, parenting has zero effect on, for example, the color of your eyes. So you have to first specify what variables you're looking at. I hope you realize that this is an extremely strong claim.
I'm not sure, I just took his word on it. I'll ask him in the morning and edit it in later
Hint: when the result of scientific studies is confusing, not conclusive, it is probably better to take a step back and try to see things from a common-sense angle, because it helps deciding what would we exactly want kinds of hypotheses we want scientists to test. So let's generate typical parenting moves that we may think could have an impact. Positive: * getting children hooked on reading (worked for me and I guess for 75% of LW) * getting children hooked on sports (discipline, mature thinking, a friend's 13 year old athlete daughter is literally the most adult thinking child I ever saw) * an athmosphere of ambition and confidence (don't think CEOs are a separate kind of people who reproduce amongst themselves, think like you can become one) Negative: * the usual kind of violent, abusive, drunken non-parenting, the chaotic environment of parents with problematic personality disorders * not insisting on things like homework, not caring about grades * anti-intellectual athmosphere at home, against studying, "why care about geography just be a miner like your dad" * unpredictable parenting The question is, are they testing these?
The case is made over twin studies. If you believe that parents equally try to get all their children hooked on reading, it's factored in. You assume that the people here wouldn't have getting hooked on reading if their parents didn't encourage reading.
It's too far against nurture. This is pushing against the limit for hard reductionism there are definitely non-genetic emergent effects while maintaining the absolute good taste of genetic arguments.

Can some heroic math-savvy LWer compile a poll about 'the most difficult kind of math I had to learn in high school' and 'in uni'? A separate issue would be 'probable reasons for failure', a multi-choice question (please include 'well-meaning family help'). And somewhere there'd have to be an option 'never experienced this KoM before'. This way we'll get some data about what we struggle(d) with. Also we can do it for 'the clicking' KoM. ... Actually, perhaps it can be better done as a survey, with an additional option allowing people to state the 'intercon... (read more)

A lot of your participation here appear to be you trying to get people to do math/modeling work for you.
Quite so.