LESSWRONG
LW

Comment Permalink

From the inside it can be very tough to tell, but from the outside they're clearly they're wrong about them all being low probability.

if you learn you're wrong, if you can't learn how you're wrong and in which direction to update even after thinking about it

What do you mean, can you give some examples? Normally, if people locked themselves in a box of their own making, they can learn that the box is not really there.

The idea of making mistakes, not realizing, and then using that lack of realization as further evidence that I'm not overconfident is a trap I don't want to fall into.

That's a good point -- I agree that if you don't realize what opportunity costs you are incurring, your cost-benefit analysis might be wildly out of whack. But again, the issue is how do you reliably distinguish ex ante where you need to examine things very carefully and where you do not have to do this. I expect this distinguishing to be difficult.

"Actually thinking it through" is all well and good, but it basically boils down to "don't be stupid" and while that's excellent advice, it's not terribly specific. And "can you eat the loss?" is not helping much. For example, let's say one option is me going to China and doing a start-up there. My "internal model" says this is a stupid idea and I will fail badly. But the "loss" is not becoming a multimillionaire -- can I eat that? Well, on the one hand I can, of course, otherwise I wouldn't have a choice. On the other hand, would I be comfortable not becoming a multimillionaire? Um, let's say I would much prefer to become one :-) So should I spend sleepless nights contemplating moving to China?

jimmy9y00

I don't know about that. That clearly depends on the situation -- and while you probably have something in mind where this is true, I am not sure this is true in the general case. I am also not sure of how would you recognize this type of situation without going circular or starting to mumble about Scotsmen.

I mean about the whole group of things that any given person decides or would decide is "low probability". I see plenty of "p=0" cases being true, which is plenty to show that the group "p=0" as a whole is overconfident ... (read more)

See in context

8 Open thread, Nov. 02 - Nov. 08, 2015

by MrMind

2nd Nov 2015

1 min read

195

8

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.

Open Threads

Personal Blog

8

Open thread, Nov. 02 - Nov. 08, 2015

2MarsColony_in10years

1Good_Burning_Plastic

New Comment

195 comments, sorted by

top scoring

Click to highlight new comments since: Today at 8:31 AM

Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

[-]ChristianKl9y110

I'm in the process of reading Kuhn's "The Structure of Scientific Revolutions". It's interesting but it is quite dated. Is there a 21st cenutry book on the history and philosophy of science that you would recommend?

7Daniel_Burfoot9y

Maybe there are more up-to-date books, but it is hard for me to imagine any book having more insight per page than SSR. It is also incomparably well-written; even if you don't believe any of the philosophical claims, it is worthwhile simply as a lesson in how to write engagingly about scientific topics.

0ChristianKl9y

Yes, it's the book where I highlited the most things in my kindle. I don't want to discourage anybody reading it. At the same time I want to know what to read as a follow-up.

2Strangeattractor9y

It's not exactly on the same topic, but I liked The Nature of Technology by Brian Arthur. When I read it, at first it seemed like he was stating the obvious, but over time I've thought about it more, and realized that a lot of people don't understand what is in his book, and that his framework is more useful than many others.

1Douglas_Knight9y

I haven’t read it, and I’m not sure it’s really on the same topic, but a lot of people like the Golem (de) by Collins and Pinch (1993/1998). How can a work on the history and philosophy of science be outdated? I suppose new information could rewrite history, but I don’t think that happened. Philosophy is more likely to change, particularly as scientists respond to Kuhn, but largely, they didn’t.

0gwern9y

New information and representations and analysis of old information are both possible. I don't remember if Kuhn himself focused on the case of Galileo, but a lot of people took him to be a paradigmatic case (sorry!) and Feyerabend undermined a lot of that through close re-examination of primary sources, in support of his own particular philosophy of science.

0ChristianKl9y

Mainly 50 years of new history happened. People came up with concepts like "evidence-based medicine" and a bunch of concepts about how science is supposed to progress. After dealing a bit more with HPS (history and philosophy of science) I get the impression like logical positivism simple ignored the arguments made against it. The New Atheist crowd simply reject criticism of logical positivism as obstruce postmodernism but I never heard someone actually engage the kind of arguments that Kuhn makes. After I wrote the post I found a lectures series by Hakob Barseghyan. He makes a lot of sense and yet, for some reason HPS isn't in popular culture. I don't understand why HPS doesn't get taught in high schools.

1Lumifer9y

That's not a new concept. That's a straightforward application of the scientific method (and some common sense) to the area which stubbornly resisted and continues to resist it.

0ChristianKl9y

I think both Kuhn and Barseghyan would say that there isn't a single thing that's "the scientific method" and that believing in such a thing isn't defensible when you look at the history of science.

0username29y

http://plato.stanford.edu/entries/thomas-kuhn/#6.4

[-][anonymous]9y100

Excepts from ''Explaining and inducing savant skills: privileged access to lower level, less-processed information'' by Allan Snyder, available here

There are now several accounts of artificially induced savant-like skills, in drawing, proofreading, numerosity and false memory reduction, all by inhibiting the LATL with repetitive transcranial magnetic stimulation (rTMS; Snyder et al. 2003, 2006; Young et al. 2004; Gallate et al. 2009).
(i) Why does becoming more literal enhance numerosity?
We argue that the estimation of number by normal people is performed on information after it has been processed into meaningful patterns. The meaning we unconsciously assign to these patterns interferes with our accuracy of estimation, whereas savants, by virtue of being literal, have less interference.
This insight has an important generalization. The healthy brain makes hypotheses in order to extract meaning from the sensory input, hypotheses derived from prior experience (Gregory 1970, 2004; Snyder & Barlow 1988; Snyder et al. 2004). So judgements in general are likely to be performed on this hypothesized content, not on the actual raw sensory input. This suggests the possibility of artifici

... (read more)

4Kaj_Sotala9y

Closely related: When the world becomes ‘too real’: a Bayesian explanation of autistic perception

[-]gjm9y60

I just (in, say, the last couple of days) got something like 50 downvotes, presumably all on old comments since I don't see any sign of a lot of downvotes on recent comments of mine.

This is the kind of thing that has got people banned in the past; if any moderator happens to be reading this and considers it worth investigating, I'd be interested to know the result.

[-]NancyLebovitz9y100

I've contacted tech.

0gjm9y

Thanks!

0username29y

I think that you should have PMed moderators instead of posting about it in an open thread. You can get -50 karma, if, say, 5 people downvoted your posts to Main or meetup posts.

9gjm9y

You can indeed, but I think I've posted exactly once to Main, it was over five years ago hence probably getting very few votes either way now, and existing votes on the post in question suggest that it would be very unlikely that five people would independently decide to downvote it in one day. On the other hand, I've been hit by mass-downvoting more than once in the past, by a user who is widely (and with very good reason) believed to be active on LW with his (at least) third identity after the two previous ones were banned for downvote abuse. The advantage of posting about it is that if indeed someone (whether or not the particular individual just alluded to) is mass-downvoting me, then they might be doing it to other people too, and some of those other people might see my comment and mention that it's happening to them, which might be useful to moderators if they're contemplating any sort of action.

[-]Liron9y60

I made this joke site: https://flashcash.money

It's often rational to burn cash on positional goods like Rolexes and bottle service at clubs, but FlashCash.money takes that value proposition to the logical extreme.

0lmm9y

The problem is the site looks cheap. If I'm showing off how rich I am, I want something that looks elegant and refined.

1Lumifer9y

People who value elegance are refinement are NOT the target demographic for that website X-)

[-]SodaPopinski9y50

What is a computation? Intuitively some (say binary) states of the physical world are changed, voltage gates switched, rocks moved around (https://xkcd.com/505/), whatever.
Now, in general if these physical changes were done with some intention like in my CPU or the guy moving the rocks in the xkcd comic, then I think of this as a computation, and consequentially I would care for example about if the computation I performed simulated a conscious entity.

However, surely my or my computer's intention can't be what makes the physical state changes count as a ... (read more)

5Viliam9y

I think everything is a computation, and all computations happen... but somehow, some of those computations happen "more" and some of them happen "less". (Similarly how in quantum mechanics any particle can be anywhere, but some combinations of particles "exist"more", and some "exist less", so in real life we don't percieve literally everything, but some specific situations.) Without understanding the nature of this "more" and "less" it will not make much sense... and I don't really understand it.

0SodaPopinski9y

If there are really infinite instances of conscious computations, then I don't think it is unreasonable to believe that there exists no more/less measure and simply we have no reason at all to be surprised to be living in one type of simulation than another. I guess my interest with the question was if there is any way to not throw the baby out with the bathwater, by having a reasonable more restrictive notation of what a computation is.

0Viliam9y

I think having a measure is exactly the way to not throw the baby out with the bathwater. But I am not really an expert on this.

0Houshalter9y

I'm confused about your "interpretation". Lets say I throw together a bunch of random transistors. They compute a totally random function. What "encoding" can you possibly use to interpret this is a conscious mind? Lets just say we already know what consciousness is and what algorithm the human brain uses. Maybe it's something like current neural networks. How would you find a computation of a neural network inside a random circuit? I don't think you could. You'd need to find groups of logic gates which just happen to compute multiplication of two numbers. And other groups which computes addition. And another group which saves the state. And all of these groups would have to be connected in just the right way. I think conscious minds are a very specific kind of computation. That's very unlikely to form by random chance.

0SodaPopinski9y

Take the thermal noise generated in part of the circuit. By setting a threshold we can interpret it as a sequence 110101011 etc. Now if this list sequence was enormous we would eventually have a pixel by pixel description of any picture, letter by letter description of every book, state after state description of the tape on any Turing machine etc (basically a Library of Babel situation). Now of course we would need a crazy long sequence for this, but there is similar noise associated with the motion of every atom in the circuit, likewise the noise is far more complex if we don't truncate it to 0's and 1's, and finally there are many many many encodings of our resulting strings (does 110 represent the letter A, 0101 a blue pixel and so on). If I chose ahead of time the procedure of how the thermal noise fluctuates and I seed in two instances of noise I think of as representing 2 and 3, and after a while it outputs a thermal noise I think of as 5 then I am ok calling that a computation. But why should my naming of the noise and dictating how the system develops be required for computation to occur?

1Houshalter9y

Random sequences aren't really interesting. Even the digits of pi are believed to contain every possible sequence of integers. The hard part is finding where each sequence is located. The index is likely to be longer than the sequence itself! And a sequence of digits isn't computation. A recording of your neural activity isn't conscious. It's just a static object. But there is no computation happening there. It's just random noise. It's just as likely to output 5 as 6 or 3. There is no causal link between you inputting "2+3" and the output.

0SodaPopinski9y

I agree with your sentiment. I am hoping though that one can define formally what a computation is given a physical system. Perhaps you are on to something with the causal requirement, but I think this is hard to pin down precisely. The noise is still being caused by the previous state of the system, so how can we sensibly talk about cause in a physical system. It seems like we would be more interested in 'causes' associated to more agent-like objects like an engine than formless things like the previous state of a cloud of gas. Actually I think Caspar's article was trying to formalize something like this but I don't understand it that well: http://lesswrong.com/r/discussion/lw/msg/publication_on_formalizing_preference/

2Houshalter9y

Read Causal Universes first if you haven't. I think causality is the only requirement for "computation". Step A causes step B. A computation has happened. If A and B are independent, then there is no computation happening..

[-]Panorama9y40

Gene editing saves girl dying from leukaemia in world first

For the first time ever, a person’s life has been saved by gene editing.

...

Layla’s doctors got permission to use an experimental form of gene therapy using genetically engineered immune cells from a donor. Within a month these cells had killed off all the cancerous cells in her bone marrow.

[-][anonymous]9y130

Acute lymphoblastic leukemia and other blood tumors in which B-cells become malignant are extremely well-suited to this approach. You can cook up a T-cell that will react against B-cell specific proteins, inject it, and it will sense all the B cells around it and grow up to large numbers and kill all the B-cells and B-cell derived tumor cells in the patient's body. You can live without B-cells (with a hit to immune system strength) and they have some very cell-specific proteins. Going after B-cell malignancies with modified immune cells has been successfully done before.

I am loving the new twist though - rather than going through the process of extracting the patient's own T-cells and modifying them, they took a T-cell line they already had and destroyed its ability to ever respond to anything but the targeted antigen, meaning that a tissue-compatibility mismatch was irrelevant because it would never go after any foreign things it encountered other than the one coded target. The cells were apparently also modded to be resistant to chemotherapy drugs. The same cell line could be used in multiple people - though I'm sure that if any of the patient's own immune system remained at all the foreign T cells would eventually be killed off rather than becoming a permanent part of the immune system as sometimes happens when the cells come from the patient themselves.

[-]Panorama9y40

Zombie physics: 6 baffling results that just won't die

To celebrate Halloween, Nature brings you the undead results that physicists can neither prove — nor lay to rest.
When a scientific result seems to show something genuinely new, subsequent experiments are supposed to either confirm it — triggering a textbook rewrite — or show it to be a measurement anomaly or experimental blunder. But some findings seem to remain forever stuck in the middle ground between light and shadow. Even efforts to replicate these results — normally science’s equivalent of Valyr

... (read more)

[-]Panorama9y40

Laszlo Babai (University of Chicago): Graph Isomorphism in Quasipolynomial Time (Combinatorics and TCS seminar)

We outline an algorithm that solves the Graph Isomorphism (GI) problem and the related problems of String Isomorphism (SI) and Coset Intersection (CI) in quasipolynomial (exp(polylog n)) time.
The best previous bound for GI was \exp(\sqrt{n log n}), where n is the number of vertices (Luks, 1983). For SI and CI the best previous bound was similar, \exp(\sqrt{n}(log n)^c), where n is the size of the permutation domain (the speaker, 1983).

G. Ph... (read more)

[-]Lumifer9y40

An unusually clear discussion of the failings of the p-values and what you can (or can not) expect from them. The author seems to have a slight allergy to the Bayesian approach though he freely acknowledges that what he is using is, in fact, the Bayesian approach :-/

[-]passive_fist9y40

What are time-efficient ways of finding people with similar interests and skills to cooperate with?

7MrMind9y

I'll try to throw some suggestions at you, see what sticks: Online: * meetup.com * searching for a specific hashtag in Facebook and befriend the people that shows up * Kickstarter Offline * fliers with your contact info and the intended interests * a post on some wallboard in a crowded community (that is somehow related to the field of interest)

4Fluttershy9y

I've always been told something along the lines of "find a group based around a hobby that you like/ an interest that you have, and make friends through them," though I've been recently wondering if it's possible to guess in advance which groups might be more likely to contain the most potential close friends. Personally, I've had an easier time making friends with bronies and HPMoR readers in meatspace than I have had with making friends with people participating in, say, service organizations or chemistry club. The most obvious explanation here is that I have more in common with people in the first two of these groups than I do with people in the last two of these groups. Still, I'm nevertheless tempted to posit something about the fact that signaling membership as an HPMoR reader, or as a brony, is reasonably costly to some people-- and that this might serve to filter out a portion of the would-be members of these groups who I'd be less likely to be friends with.

0Stingray9y

Another possibility: the first group of people have more free time than the latter and spending some free time together is quite important for building friendships.

2ChristianKl9y

The most time-efficient way is to be open about who you are and what kind of people you are seeking so that other people can find you.

-1[anonymous]9y

Find a community that someone else created of people with similar interests and skills (eg, lesswrong if you're looking to cooperate with other people interested in rationality

[-]stoat9y40

I have a foggy memory of someone here (I think it was gwern) linking to an article about simulation interface design. It built up examples based on a bird's eye view of a car steering down a road. I haven't been able to find it, anyone know a link to the article?

[-]gwern9y150

Bret Victor's "Up and Down the Ladder of Abstraction"?

4stoat9y

Thanks a bunch that is the one!

[-]DanielDeRossi9y30

Does anyone have any good recommended reading on being social? Stuff like understanding social situations and how to respond appriopriately. Understanding personality types and how to engage with them. How to make people like you can how to keep people's attention. I'd really love to learn these skills , as I feel I am deficient at them.

7ChristianKl9y

The Charisma Myth by Olivia Fox Cabane is a book that helped a bunch of people on LW. I'm not sure that's a good approach to the issue. Don't label other people with a pesonality type and then try to engage with them based on the label.

7Vaniver9y

I think I would actually recommend this. If other people are deeply mysterious to you, then reading up on personality types and trying to recognize them in the wild is helpful training and theory. The trouble is twofold: 1. The theory will be incomplete, and only give you broad understanding. 2. The theory will be limiting, in that you will be more likely to notice observations that match the theory than observations that do not agree with the theory. You can ameliorate both troubles by learning multiple theories, and trying to hold them in your head / evaluate people along different ones simultaneously. (There's a longer conversation here, about how much learning should be system 1 vs. system 2, and how to tell what level of development you are in a skill, and so on, but that's probably enough for now.)

-2Tem429y

You might be able to do a bit better; learn a simple and catchy system like the True Colors personality spectrum (a simplified adaptation of the Myers-Briggs), and work on understanding why it works. (Or if you like, why it "works".) While you might guess someone's 'color' incorrectly, if you understand why everyone identifies at least a little with every color, you can start to use general, positive statements to identify what people like about themselves. It should be a productive exercise in understanding the average person's self concept.

-2ChristianKl9y

Success in social interaction is not about holding more things in your head but often about holding less things in your head. It's better to do exercises that raise physical presence as the one's suggested in "The Charisma Myth".

5Kaj_Sotala9y

In the sense that you'll want to be able to model people and their reactions automatically and without needing to spend effort on it or it hogging up all your working memory, true. But if you're not good at modeling people yet, it may be better to practice it consciously until it becomes automatic. These are not mutually exclusive.

1ChristianKl9y

Normal people don't model each other through putting each other in distinct mental categories (personality types) but via mirror neurons. Being judgemental of other people doesn't get better by doing it automatically. I don't get anything when I have an automatic thought that tells me "the person I'm interacting with is a ISTP" In social interaction "Get out of your head" is good advice for the average nerd. Judging another person as a ISTP rather keeps them in their head.

7Kaj_Sotala9y

So, let's take autistic vs. neurotypical people as an example. As a general (but not iron-clad) rule, autistic people tend to read less social connotations into the meanings of words. As a result, they are often less likely to take offense from things that a neurotypical person might read as insulting. And as a result of that, they're more likely to prefer the kind of communication that's more direct and to the point. In contrast, with more neurotypical people, exactly the same kind of communication might come across as cold and blunt. Knowing this lets me optimize my style of communication to the kind of person I'm talking with, more direct with autistic and more careful with neurotypical. Now of course there are some autistic people you need to communicate carefully with, and some neurotypical people who prefer direct and blunt communciation. But if I have a higher prior probability on someone preferring direct communication, that lets me make some cautious probes to measure their reaction to that style of communication. Probes which could have a negative expected utility if I put a higher prior probability on the person being easily offended by more direct language. This doesn't necessarily happen on a conscious level. Just having the background knowledge of neurotypical and autistic people differing on this dimension, helps me do this on a partially instinctive basis. I wasn't explicitly taught this thing about how autistic and neurotypical people differ. It was something that I picked up by experience, from interacting with both kinds of people. But for learning this, it was important to have some kind of a mental handle for hanging the differing experiences on. If I hadn't known that there was such a concept as an autistic person, I couldn't have noticed the correlation between autism and the preferred style of communication. Rather my experience would just have been "different people react totally differently to the same kind of words, and it's totally my

3Vaniver9y

Sounds like we do need to go into the longer conversation. I view most of these skills are something like the follows: at level 0, you have no clue what's going on; at level 1, you have a system 2 model of what's going on that's too slow / clumsy to operate successfully in real time; at level 2, you have a system 1 model of what's going on that's fast and good enough to operate successfully in real time. Most people go directly from level 0 to level 2, with some level 1 help. Most language speakers don't have an abstract grammatical model of their language in their heads, some constructions "just look weird" or don't come to mind, and they often can't articulate rules even if they use them correctly. For example, in English, why is something "harder" instead of "more hard"? Why is it "more difficult" instead of "difficulter"?* (This came to mind because my mother is teaching ESL classes and had been surprised that there was a simple underlying rule, which I could not successfully identify before the question was spoiled, even though for any particular word I could correctly determine whether 'more' or 'er' was appropriate.) But there are situations where it seems better to go through level 1. If you're teaching someone a second language, for example, they're much more likely to be able to make use of abstractly stated grammatical rules than children are. If someone has already been a child and yet not developed a 'normal' level of social intelligence, then the normal approach is inadequate, and we need to consider alternatives. When developing those alternatives, it's worth noting that the right approach for going from level 0 to level 1 (learning more grammatical rules into system 2) is different from the right approach for going from level 1 to level 2 (practicing the grammatical rules into system 1). So yes, someone who is at level 1 would not get much out of holding more things in their head, but someone who is at level 0 would. (To elaborate even further,

2ChristianKl9y

I think the problem is that you ignore the physiological effect of being in your head and how it makes people less likely to want to engage in social interaction with you. A problem that is about not being in contact with one's emotions is not helped by having concept with you can use to label the person with whom you are interacting. I don't think that it's useful to people into the bracket of caring about relationship closeness and people who care about factual accuracy. Depending on the context of the conversation the same person will focus on a different layer of the communicatoin. Schulz von Thun's model describe the issue well. You don't need to put people into categories for that.

0Douglas_Knight9y

I think that page oversimplifies the rule for constructing comparative forms. One-syllable adjectives definitely take suffixes and three-syllable adjectives take words, but two syllable adjectives are difficult. I think this page is largely correct. For two syllable adjectives, some terminal syllables (-y, -le) require suffixes and some (-ing, -ed, -ful, -less) require words. The rest are OK either way (quieter, more quiet).

04hodmt9y

This rule is incomplete. Most two-syllable adjectives ending in "y" can be converted to comparative form with "er". Some of these may be uncommon, but not all, and my spell checker agrees they are real words, in both British and American English. Eg. Angrier, heavier, cleverer, friendlier, happier, lazier, tidier, etc. And even three syllable words can take "er": bubblier, foolhardier, jitterier, slipperier, many words starting with "un".

6username29y

This handbook is written by a person with Asperger Syndrome and it's intended for other people with Asperger Syndrome, but it is very good even if you aren't autistic, because it spells out everything in detail and makes you explicitly think about all the rules and interpersonal skills.

[-]OrphanWilde9y30

Posting here to avoid introducing an irrelevant aside on one of the main [ETA: Discussion-main, not Main-main] threads, regarding the "retrocausality" of Newcomb-like problems.

Causality is always bidirectional. It is information which only goes in one direction. Once you dissolve that distinction, the question is one of information, which doesn't need to involve any strange loops at all; the behavior of Newcomb-like problems isn't produced by your actions changing history, but by information about what your action will be changing the future, o... (read more)

[-]NancyLebovitz9y20

It looks like letting math students flounder for a while before telling them how to solve a problem helps them learn.

[-]Panorama9y20

Is Economics Research Replicable? Sixty Published Papers from Thirteen Journals Say “Usually Not” by Andrew C. Chang and Phillip Li

We attempt to replicate 67 papers published in 13 well-regarded economics journals using author-provided replication files that include both data and code. Some journals in our sample require data and code replication files, and other journals do not require such files. Aside from 6 papers that use confidential data, we obtain data and code replication files for 29 of 35 papers (83%) that are required to provide such files as

... (read more)

[-]jsteinhardt9y110

Note that their implicit definition of "replicable" is very narrow --- under their procedure, one can fail to be "replicable" simply by failing to reply to an e-mail from the authors asking for code. This is somewhat of a word play, since typically "failure to replicate" means that one is unable to get the same results as the authors while following the same procedure. Based on their discussion at the end of section 3, it appears that (at most) 9 of the 30 "failed replications" are due to actually running the code and getting different results.

3Lumifer9y

Yes, there is a difference between "unable to replicate because we couldn't even attempt to replicate" (code and/or data are missing) and "unable to replicate because we tried and the results did not match". Either both or only the second case could be called "failure to replicate", depends on your preferred definition. Still, while the second case is clearly "bad science" -- it's either mistakes or fraud -- the first case is "not science" because science doesn't work by trusting the word of the researcher. A well-known example of the first case is cold fusion.

[-]SodaPopinski9y20

It is interesting to compare the Less Wrong and Wikipedia articles on Recursive self improvement: http://wiki.lesswrong.com/wiki/Recursive_self-improvement https://en.wikipedia.org/wiki/Recursive_self-improvement I still find the anti-foom arguments based on diminishing returns in the Wikipedia article to be compelling. Has there been any progress on modelling recursively self improving systems systems beyond what we can find in the foom-debate?

5Viliam9y

Quoting Wikipedia: Corporations are not superintelligences. Not in the narrow sense we use when we talk about AGI. The whole idea of superhuman artificial general intelligence is that the machine would be better at everything that humans can do. Also, if the machine wants to specialize, it doesn't have to deal with humans: it could just create more copies of itself (okay, it would need to buy the hardware, but that's it) and let different copies specialize at doing different things. Not sure if I understand it correctly, but this seems to me like an argument that "wealth = money, therefore the AI will trade with humans". If that is the intended meaning, I disagree. Money is just one way to get resources. You can also take them by force, steal them, convince people to donate to you, create them, discover them, et cetera. Just because the AI would use its intelligence to gather resources, it does not follow it will trade with humans. Why would the AGI still have human cognitive constraints? Just because we can't imagine otherwise? Why would it depend on other people to get things done? If it can get some people to build robots, the rest of the work can be done by those robots. Okay, maybe some of these arguments need much more thinking than I used now, but I wrote this to explain why the arguments in Wikipedia seem completely unimpressive to me. Most of them seem to be based on refusing the idea that anything, including any AI, could really be significantly smarter than humans. "The AI cannot do research in several fields at once, because humans can't. The AI will not have hands, and will therefore forever depend on humans. The only way an AI could get resources is to have a job, and then buy whatever it needs in the shop. This all ensures that AI is just another human, so other humans will have no problems to overpower it, if needed."

[-]MarsColony_in10years9y20

I've recently started using RSS feeds. Does anyone have LW-related feeds they'd recommend? Or for that matter, anything they'd recommend following which doesn't have an RSS feed?

Here's my short list so far, in case anyone else is interested:

Less Wrong Discussion
Less Wrong Main (ie promoted)
Slate Star Codex
Center for the Study of Existential Risk
Future of Life Institute [they have a RSS button, but it appears to be broken. They just updated their webpage, so I'll subscribe once there's something to subscribe to.]
Global Priorities Project
80,000

... (read more)

9Viliam9y

You may like some of the sources mentioned in "List of Blogs" on LW wiki.

1username29y

Overcoming bias sidebar has a lot of interesting blogs.

0Stingray9y

You should also subscribe to this subreddit.

0polymathwannabe9y

I used to use Thunderbird Portable, but my flash drive died and I lost thousands of saved articles. After Google Reader was discontinued, I switched to Feedly. This is my OPML. Note that I've included several Colombian media sources because that's where I live.

[-]cleonid9y20

From Omnilibrium:

0passive_fist9y

Evidence? This is LessWrong. Based on what I've read and the contents of the IPCC report, the match between models and climate change has been pretty good so far, actually.

3Lumifer9y

As you mentioned, this is LessWrong. So someone (like me) will go and look at your link. And find that it doesn't talk about forecasts, it is predominantly concerned with whether models can reproduce historically observed features of the climate. In other words, the issue is just trying to get a good fit to historical data. By the way, if you look at p.771 and around, you'll notice that the models have a lot of difficulties with the current "hiatus".

0passive_fist9y

Exactly; that's what I said. EDIT: To clarify, I'm trying to say that past models have had a good fit to data so far, so it's reasonable to expect they will continue to perform. This is certainly evidence towards climate models being able to carry out predictions, and it's how science should be done. You make it sound as if they are just arbitrarily varying the parameters of the models to get a good fit. In reality, they are using model ensembles for various different emissions scenarios obtained from real-world data and seeing if the resulting predictions fall within a reasonable confidence interval of what was actually observed. The answer is: yes, they do. There are several ways of interpreting this; I'd be glad to have a discussion about it if you're interested.

2Lumifer9y

Why, then, are you talking about the models' fit when answering the question of whether the "climate models can predict weather changes over long term periods" (emphasis mine)? Not arbitrarily, of course, but "varying the parameters of the models" is the most common and a very general method of getting "a good fit". Do you mean something like cross-validation? I don't think they predict the future in this context.

2passive_fist9y

What other way is there? Building a time machine? How else can you estimate the suitability of models in making predictions than testing their past predictions on current data?

5Salemicus9y

One possible answer is to look at how the then-state-of-the-art models in (say) 1990, 1995, 2000, etc, predicted temperature changes going forwards. The answer, in point-of-fact, is that they consistently predicted a considerably greater temperature rise than actually took place, although the actual temperature rise is just about within the error bars of most models. Now, there are two plausible conclusions to this: * Those past mistakes have been appropriately corrected into today's models, so we don't need to worry too much about past failures. * This is like Paul Samuelson's economics textbook, which consistently (in editions published in the 50s, 60s, 70s and 80s) predicted that the Soviet Union would overtake the US economy in 25 years.

3passive_fist9y

It's not as simple as that. Most models give predictions that are conditional on input data to the models (real rate of CO2 production, etc.). To analyze the predictions from, say, a model developed in 1990, you need to feed the model input data from after 1990. Otherwise you get too wide an error margin in your prediction. True. As I said, this is definitely evidence towards the suitability of the models, and certainly seems to be counter to the claim that "there is no evidence that climate models are valuable in predicting future climate trends. That's definitely a possibility, but it's reasonable to think that the mathematics and science involved in the climate models stands on a firmer basis than economical analysis, and definitely a firmer basis than Samuelson's analysis.

4Lumifer9y

The usual plain-vanilla way is to use out-of-sample testing -- check the model on data that neither the model nor the researchers have seen before. It's common to set aside a portion of the data before starting the modeling process explicitly to serve as a final check after the model is done. In the cases where the stability of the underlying process in in doubt, it may be that there is no good way other than waiting for a while and testing the (fixed in place) model on new data as it comes in. The characteristics of the model's fit are not necessarily a good guide to the model's predictive capabilities. Overfitting is still depressingly common.

[-]Panorama9y10

NASA Study: Mass Gains of Antarctic Ice Sheet Greater than Losses

A new NASA study says that an increase in Antarctic snow accumulation that began 10,000 years ago is currently adding enough ice to the continent to outweigh the increased losses from its thinning glaciers.
The research challenges the conclusions of other studies, including the Intergovernmental Panel on Climate Change’s (IPCC) 2013 report, which says that Antarctica is overall losing land ice.
According to the new analysis of satellite data, the Antarctic ice sheet showed a net gain of 112 b

... (read more)

3Manfred9y

Interesting. Obviously if some place is still below freezing all year round (i.e. the bulk of East Antactica), global warming can easily increase ice mass due to increased snowfall. But I'd thought decrease in total ice mass was pretty well-established.

[-][anonymous]9y110

The amount lost in the Arctic is about a factor of 3 larger than the net gain in Antarctica, and West Antarctica as a subset of antarctica is losing ice on the net in a way that is likely to accelerate in the future. Also apparently Antarctica has been gaining ice on the net for 10,000 years according to the source, and it's a case of recent loss rate increases not yet balancing this normal gain rate.

Further quote from the article:

“If the losses of the Antarctic Peninsula and parts of West Antarctica continue to increase at the same rate they’ve been increasing for the last two decades, the losses will catch up with the long-term gain in East Antarctica in 20 or 30 years -- I don’t think there will be enough snowfall increase to offset these losses.”

3Manfred9y

A late follow-up. I read an article on the study, and it turns out that the difference from previous estimates (which basically all showed a decrease in antarctic ice mass) comes from an interesting place. Everyone agrees on the height change in East Antarctica. But the studies that got a net decrease assumed that the change in height was due to recently increased snowfall, in which case the extra height will have the density of snow. This new study that gets a net mass increase assumes that the change in height is actually part of a long-term mass rebound from the last minimum, and if that's true then the density profile of the Antarctic ice sheet should be roughly constant, and the extra height will have the density of ice, which is ~3x that of snow. I think the disagreement over this shows how big our error bars are.

1knb9y

This seems significant, but I'm not sure how to interpret it... Is it good news the ice sheet isn't shrinking or bad news that the sea level rise apparently came from other sources without us noticing?

0Lumifer9y

Which sea level rise?

4[anonymous]9y

http://www.worldviewofglobalwarming.org/images11/SeaLevelRiseRateChart2010.jpg This is the global mean. Rise measured at any given actual shoreline will be different and sometimes even falling, due to local geology altering elevations of land at not-dissimilar rates in some areas (especially areas where post-glacial rebound is still occurring) as well as thermal expansion being uneven.

1Good_Burning_Plastic9y

Was something weird happening in the 1920s or is it just an optical illusion due to the black lines?

0[anonymous]9y

I think you'd see similar anomalies in 1880 and 1985 stand out with similar lines.

-1Lumifer9y

Yes. But the sea levels have been rising continuously since the time of the last glacial maximum. 10,000 years ago they were rising at a rather more dramatic rate, too.

1[anonymous]9y

Yep! My favorite bit of what went on during the end of the last glaciation is the way that it happened unevenly, a sedate constant flow of water from ice to the oceans interrupted by centuries here and there where sea level rose by at least 2-5 centimeters a year. Presumably that's what happens once an ice sheet becomes unstable and pieces of them collapse quickly and nonlinearly.

0Lumifer9y

That was one of those "interesting times to live in"? Still it's peanuts compared to the mother of all floods :-)

[-]Ritalin9y00

This is the most terrifying comic SMBC has made yet How much of a point does Zach have, here? Can this be the shape of the future?

2Lumifer9y

Luddites are not new.

0knb9y

That scenario doesn't seem terrifying to me, though it's pretty vague. He says there are job losses and revolution is impossible but so what? Realistically in this scenario people just vote to raise taxes on capital owners and give themselves a paycheck. Machine labor is apparently extremely capable and near-free in this scenario so owning even a small amount of capital makes you effectively rich in absolute but not relative terms. I guess he's assuming democracy breaks in a way that is pro-capital owners somewhere along the way but that isn't actually stated.

0username29y

Probably not, because people who do not like new society can create a small closed society somewhere where population density is smaller and live off the land.

[-][anonymous]9y00

If you had to select just 5 mutually exclusive and collectively exhaustive variables to predict the the outcome of something you have expert knowledge (relative to say...me) about:

what is that situation?
what are the 5 things that best determine the outcome?
Please tell us about multiple things if you are an expert at multple things. No time for humility know, it is better that you are kind teacher than a modest mute.

If you can come up with a better way I could ask this, please point it out! It sounds clumsy, but the question has a rather technical ba... (read more)

4JoshuaZ9y

Do you have a citation for this? My understanding was that in many fields experts perform better than nonexperts. The main thing that experts share in common with non-experts is overconfidence about their predictions.

0[anonymous]9y

Yep, follow the summary here to the Expert Judgement textbook if you like. I was skeptical too before I followed up this claim and actually read the textbook.

[-][anonymous]9y00

Top 4 web resources about the application of machine learning to public health:

As taught coursework

As a field of research

As independent scientific movements

In one computationally intensive field of public health

Feel free to add any non significantly overlapping, high quality resources in the comments, or to comment otherwise.

[-][anonymous]9y00

This may be a silly question, if complex regional pain syndrome is 47 out of 50 on the McGill pain scale, where would 'torture' be?

1James_Miller9y

Reminds me of a joke where a kid in great pain is asked by a doctor to rate how much it hurts on a scale from 1 to 10 where 10 is the most pain he can imagine, and the kid says "1."

6Alejandro19y

That is from xkcd.

0[anonymous]9y

What specific method of torture? I'd assume that many methods are designed to get as high as possible, but there are others that are much lower and instead involve other negative sensations besides pain.

0SeekingEternity9y

Agreed. "Torture" as a concept doesn't describe any particular experience, so you can't put a specific pain level to it. Waterboarding puts somebody in fear for their life and evokes very well-ingrained terror triggers in our brain, but doesn't really involve pain (to the best of my knowledge). Branding somebody with a glowing metal rod would cause a large amount of pain, but I don't know how much - it probably depends in the size, location, and so on anyhow - and something very like this on a small scale this can be done as a medical operation to sterilize a wound or similar. Tearing off somebody's finger- and toenails is said to be an effective torture, and I can believe it, but it can also happen fairly painlessly in the ordinary turn of events; I once lost a toenail and didn't even notice until something touched where it should have been (though I'd been exercising, which suppresses pain to a degree). If you want to know how painful it is to, say, endure the rack, I can only say I hope nobody alive today knows. Same if you want to know the pain level where an average person loses the ability to effectively defy a questioner, or anything like that...

[-][anonymous]9y-20

Identifying cognitive distortions

Psychologists someone's hand out a worksheet with the names of cognitive distortions. Ie: mentalising. It is usually encumbrance upon psychiatric patients to pick up on trends for these distortions themselves from what their psychologists tells them.

For instance, my psychologists might intuit that something is paranoid while I don't. That certainly kills that one belief after a bit of reflection, but my mind remains generally predisposed to paranoid thinking in the next moments of conscious ideation too. They will never cat... (read more)

[-][anonymous]9y-20

There are 4 computer memory designers. One is a major phone manufacturer, one is a major computer manufacturer and one is a major car manufacturer. The fourth, crucial, makes computer memory for their competitors. I reckon this is best if only example of a stable large scale unregulated efficiency monopoly.

I wish I was clever enough to understand computer memory, get in on the open hard ware movement and capitalise on Project Aras forthcoming release early next year! Imagine, an AppStore for hardware, sponsored by Google for android!

7ChristianKl9y

No. Producing memory cheaply needs volumne. Nothing that the open-hardware movement does changes the fact that you need scale to cheaply produce memory chips. I'm also not sure whether you can produce memory chips without violating patents from those company so you need additional millions of capital for that.

[-]Osho9y-20

I'm looking for an academic/science/technology enthusiast to cofound my new website with me.

Website is already built and nearing launch. I am looking for someone who can serve as an adviser and contribute directly to site content (I will explain the content in more detail to serious candidates). This will require knowing a lot about science, and a lot about how scientists think. He/she needs to be interested in new technologies and current events. From international affairs, economic policy, politics and ethical issues to cryptocurrency, biotech, physics a... (read more)

8ChristianKl9y

I would guess that you currently haven't said enough about your project to make anybody have serious interest to be part of your project.

6ChristianKl9y

Using the nick Osho is quite strange for a website about science. What's your background? What is the website supposed to be about? How does it differ from what is out there at the moment?

4Lumifer9y

Your post is all about what you want. What are you offering in exchange?

0Osho9y

I said there would be equity.

6ChristianKl9y

Equity is only useful if one consider a business to be a good idea. If you want someone to join and be payed in equity you need to convince them that working on your project with you makes sense.

0[anonymous]9y

Using the nick Osho is quite strange for a website about science. What's your background? What is the website supposed to be about? How does it differ from what is out there at the moment?

[-][anonymous]9y-20

I want to learn to play the Dataridoo. Is Swirl the right choice if I want to graduate into a career in machine learning, but failed at learning Python, and managed to learn Stata? Also I'm shit at concentrating, and it's the only learning platform that doesn't confuse me with all the features. I kid you not I took months to figure out how to use LessWrong. I may be stupid, but I am dedicated. Once I find the best platform for me, I'll stick to it. But good recommendations now may save me months later.

[-]gjm9y130

Pardon my candour, but if you are "shit at concentrating", are readily confused by things, and found Python too difficult to learn, then you might want to consider whether machine learning is a good choice of career.

(I gravely doubt that you are in fact stupid, and given sufficient dedication I expect you could do it, but it seems like a lot of pain. Are you sure it's worth it?)

-3[anonymous]9y

I am fairly confident it will be worth it. I'm not good at a whole lot of other things too. But, big data is said to be a profitable career trajectory in the long term. Most if not everything I do for my career is fairly painful, but I try to enjoy it as I do it and I'm grateful for the experience. The question I ask myself is: what could I be doing instead? And I honestly don't have a lot of better things to be doing anyway, haha.

-1ChristianKl9y

How strong are your math skills? Did you have IQ testing done?

0IlyaShpitser9y

Mathematicians don't give a shit about IQ. When was the last time you heard Terry Tao talk about IQ? Writing papers >>>>> psychometrics.

[-]Lumifer9y150

Mathematicians don't give a shit about IQ.

That's because they are an exclusive high-IQ club to start with.

Take someone who scored in 300s on his SAT -- would you recommend him to try to become a mathematician?

-1IlyaShpitser9y

What do you suppose Ramanujan's IQ was? Do you think Hardy cared?

[-]Lumifer9y110

What do you suppose Ramanujan's IQ was?

No one has any idea since he lived before IQ tests.

Do you think Hardy cared?

That's like saying a basketball coach doesn't care about the player's height, he only cares how high can he jump.

5IlyaShpitser9y

IQ is not like height. Height is a fairly objective physical measurement that is directly relevant for basketball because of the game setup. IQ is the result of a projection of an extremely high dimensional space into a single number that is not directly relevant for mathematics (people do mathematics in an extremely heterogeneous way). Erdos, Ramanujan, and Groethendieck were all top notch and were all very very different from each other. Erdos I think couldn't tie his shoes. Ramanujan was an Indian peasant. Groethendieck wasn't exactly a high functioning individual. A better analogy would be if a basketball coach cared about "hit points" (determined by whatever methods doctors use, slatestar would know more).

[-]Vaniver9y100

Ramanujan was an Indian peasant.

Ramanujan was a Brahmin. "Peasant' isn't quite appropriate.

2Lumifer9y

"Peasant" is also technically wrong since neither he nor his parents tilled the soil (his father was a clerk).

9Lumifer9y

How is it not "directly relevant"? What do you think the average IQ of mathematicians is, do you imagine it's anywhere close to the population average? Being able (or not) to tie one's shoes or being an Indian peasant are NOT indicators of IQ. Not being socially successful is not an indicator of low IQ either. I understand your point that genius mathematicians are really, really weird people. But I see no contradiction there, it's perfectly possible to have high IQ and be really weird.

7IlyaShpitser9y

My point isn't just that they are really weird, but that people think about mathematics in an extremely heterogeneous way, and reducing human brains to one number as some sort of "math hit points" is silly for that reason. You are just ignoring most relevant information. What made Erdos good and what made Ramanujan good were weird complicated facts about their brains (I expect Ramanujan's IQ wouldn't be very different from his cohort in India, e.g. likely not super high, but for some reason he just "saw" natural numbers). This does not make Ramanujan or his cohort stupid. I just don't think "smart" and "stupid" is what IQ measures in any interesting way. Telling people to go take an IQ test as a way to selecting themselves out from trying math is an especially toxic practice (especially if this advice is not coming from a mathematician). ---------------------------------------- I prove theorems for a living, and I say: ignore the haters, just read about math that interests you, try your hand at following and constructing arguments, etc. Math is hard (for everyone), don't worry about it. It's fun, too.

[-]Lumifer9y110

people think about mathematics in an extremely heterogeneous way

Sure, that's true.

reducing human brains to one number as some sort of "math hit points" is silly

And I agree. But consider the setup: we have a person who doesn't quite know what he wants to do and who has shown no signs of possessing any "supernatural" math abilities. Could he turn out to be another Groethendieck? Well, sure, it's possible, but we are talking about the base population rate here, the chances are, let's say, not very high.

Now, it so happens that most math professionals have high IQ. That's not a coincidence, of course -- if your brain is insufficiently weird to see math "directly", you have to rely on the same dimensions of performance (working memory, etc.) which are reflected in the IQ score.

Trying out a profession has costs, sometimes considerable. You can't try everything on the off chance that it might work out -- you want to focus on the areas where you expect to do well. And someone with an IQ of 130 has much, MUCH better chances of becoming a mathematician than someone with the IQ of 80.

1[anonymous]9y

Aaaanyway I haven't had a crude IQ test done but I've had a tailored subset of psychometric tests including subscales from the WAIS from which IQ is derived which indicate my maths skills are above average. The same tests indicated my concentration skills are below average....

6Vaniver9y

Hmm. I haven't put much thought into what professions are a good fit for someone with concentration as a comparative disadvantage. I would suspect that research, and mathematics research in particular, is a bad bet. Much of a day will be spent just thinking about ideas, and being able to think about the same idea all day long is necessary to reach the end of long and complicated chains of reasoning. The difference between Newton and his contemporaries, for example, seems to have mostly been superior concentration ability on Newton's part, not considerably higher intelligence. But people use machine learning many other places; you might be able to work as an industrial data scientist or analyst. It's not clear to me whether low concentration ability would be a smaller or larger handicap there.

1Lumifer9y

Sports commentator :-D

0[anonymous]9y

Ooh my main strengths are vocabulary and verbal abstract reasoning which are apparently greater than 3 standard deviations away from several populations means. Could you reassess possible career paths that might suit those strengths? Other weaknesses include social cognition and memory. The neuropsychologists reckons the memory thing may be due to lapses in concentration though. I'm highly skeptical of my verbal abstract reasoning results aince whenever I've done job psychometric tests or related tests when I was in school my verbal abstract reasoning can range very broadly including into the below average group albeit in sometimes competitive populations. I am very confident in my vocabulary though. It's probably the strongest in anyone I've ever seen except spelling b competition kids on TV, assuming they actually know what the words they're familiar with denote and connote in practice. I'm not so sure it's useful since among regular people people don't get what I'm saying when I slip in technical words often. I reckon it would be good in cross disciplinary technical communication. Don't know an example of that other than systems engineering but I'm no engineer and engineering isn't the broadest category. Politicians span multiple portfolios but I get totally stressed dealing with multiple issues or assignments at once. :S thanks for your assistance everyone once again I thought becoming an intelligence analyst would be a good choice. Military intelligence analysts in Australia may do shift work which isn't good for one's health.

0Vaniver9y

Sales is standard advice for people with high verbal ability, and there's plenty of sales jobs for technical subjects that do not require direct technical ability. (Someone sells MRI machines to hospitals, and they aren't an engineer.) There's a fairly large industry in machine learning enterprise solutions, where all you would need is the ability to tell apart Spark and Impala and R and Hive and Hadoop, not necessarily the ability to use any of them competently. Two issues: 'social cognition' is rather important, and there will be multiple issues or assignments at once. I think most other verbal fields are in a bad way and have declining prospects. Verbal + abstract reasoning has historically screamed law, but going to law school now is a terrible mistake. Similarly, journalism has very poor options that I suspect will continue to get worse.

0Lumifer9y

Yes -- sales at the corporate level is mostly about gladhanding and networking. People who can't seamlessly insinuate themselves into the local old-boy network will do poorly. Well... going to some law school has been a terrible mistake for years by now. On the other hand, if you can get into a top-tier law school (and there about half a dozen of those in the US), I would hesitate to call it a mistake.

0Vaniver9y

Yes, there are still top tier law firms, and you have a chance of getting hired by one if you go to a top tier law school. My point is more that even conditioned on knowing that you would survive law school and make it into a top tier law firm, it's not obvious to me that law is the best path to take: options in other industries may be far more valuable. (Consider claims about how doctors only get rich in real estate, or compare physics PhDs in academia and quantitative trading, or Peter Thiel narrowly missing out on a Supreme Court clerkship and founding a company instead.)

0Lumifer9y

It all depends, of course. Each path has its risks and its rewards. However, if -- and that's a huge if -- you can get admitted to a top-tier law school, get hired by Biglaw, and spend a few years in, say, a white-shoe NYC law firm, that doesn't sound like horrible fate to me (subject to the sensitivities of your soul, naturally).

0Viliam9y

Philosophy? Journalism?

0IlyaShpitser9y

Philosophy grad students tend to get very high combined analytic/verbal scores on tests.

-5VoiceOfRa9y

0username29y

In NFL they do care about intelligence tests.

5ChristianKl9y

I don't claim that they do. Clarity speaks of himself as stupid and the fact that he failed to learn python is indication of that. If his IQ is <100, I think that would be a valid ground on which to advice him against seeking a career in machine learning. That's exactly the purpose for which IQ test were designed.

0Viliam9y

This is only a weak evidence for non-high IQ. I know a few people who had bad opinion about their IQ, and when I convinced them to take the test, they scored above 130. It's because they believed the stereotype of "high IQ = math prodigy", and they happened to be average at math simply because they focused their lives on something else.

-1ChristianKl9y

I haven't implied that it's strong evidence, for me the available evidence was enough to raise the question. The answer to that question matters for whether or not to tell him not to seek a career in machine learning. I do think that for this purpose the testing that tells him that he's above average in math might be enough.

-1IlyaShpitser9y

I think it would be useful to taboo "stupid." It is not a useful word.

1ChristianKl9y

Tabooing "stupid" is what asking for IQ is about and why I asked about IQ in this context.

0IlyaShpitser9y

Except you are not tabooing anything then, you are just substituing "low IQ" for "stupid." The point of tabooing stupid is to get binary classification out of an inherently complicated multidimensional problem. The request of tabooing in general is a request for more cognitive work.

3ChristianKl9y

Scoring low on a specific test is something more complex than a label. Changing a vague term with a operationalised term is something that often makes sense for tabooing. I think you confuse cognitive work with explicitely describing cognitive work. When it comes to speaking about negative features of other people it's worthwhile not to say every negative thing that can be said publically.

6ChristianKl9y

What do you mean with "failed at learning Python"?

2[anonymous]9y

I tried to learn it on my own first, and didn't really pick up on anything. I tried to learn it at university then, and failed that course. As more and more helpful resources came online, I tried learning from them, and didn't end up learning it. I think my brain works very differently to most people. There are some things which simply require a kind of functioning I really don't have. It seems languages are the frontier of that - where I have it in me to learn exceptions, whereas I can't learn most. It does seem to generalise to languages - I couldn't even become bilingual even though my parents speak another language and continuously tried to teach it to me, then sent me through school for it. At one point I learned how to read in this one other language and latter forget - can't read that stuff at all now, which is kinda odd. Anyway, I've managed to learn Stata. And R is for statistical programming like Stata. So, I suspect I could learn it. Though, Stata is more GUI-like and you can't do machine learning with it.

0ChristianKl9y

The fact that made Stata easier to learn is that it's GUI like. R isn't. I see no reason to believe that learning R on a level where you can do machine learning with it is easier than learning python. Python has much better documentation than R. It has functions that are much more reasonably named.

-2MrMind9y

I don't really think this is the case, since you are using correctly English, which is far more complicated than Python. Just to be clear: complicated = has more rules and is more ambiguous.

1Kaj_Sotala9y

Formal languages require quite a different kind of thinking than natural ones do. It's not just a matter of comparing their complexity.

0MrMind9y

Unfortunately, there's only one study about the neuropsychology of programming language, but it does contradict your assertion. Or at least, if it requires a different thinking, that thinking is done with the same area used for natural language.

-2[anonymous]9y

That's really not how primary v.s secondary language acquisition works. Also k. complexity isn't the same as cognitive complexity.

0MrMind9y

I know of only one study on the neurobiology of programming language comprehension. It stacks evidence in favor of the theory that the brain uses the same areas of the brain associated with natural language processing (BA 6/44). On the other hand, studies in bilingual aphasia shows conflicting evidence: some patients lose/recover only one of the language following a brain lesion, while others shows modifications at both languages at the same time. So, if you think you have neurological deficiencies regarding the acquisition of Python, I think (wild speculation ahead) that you should show other signs of impairment in the acquisition/use of primary/secondary language. For example, were you able to learn Mata? Regarding K complexity, the difference in cognitive load is exactly my point: if for you manipulating something that has low complexity has higher complexity, it means that something is wrong in the way you learned it.

0[anonymous]9y

*Great research. Thanks for looking at the evidence, I didn't know those things and I'll try to take (admittently, a very poor and unbacked up claim on my part that I'm sorry for) your approach in the future. *I have yet to try learning Mata -I'm unclear of its applications. But, I've shown decent skill in the basic neuropsychological components of second language aquisition from military intelligence analysis testing. On the other hand I've been fairly bad at learning languages at school. May just have been the classroom format though! Didn't think of it that way. Wow! Edit: It's just hit me how complex this phrase is: I can't even conceive of what level of abstraction to place 'the way I learned a given thing' between the sandwhiches of cognitive and k complexity... In fact that may be because it's incommesurable within the domain of discourse of computational complexity

[-]Jurily9y-20

So, apparently NLP is pseudoscience, and now I'm confused. Does anyone actually claim

Richard Bandler hasn't demonstrated even a single verifiable, undisputable result with his methods, and he's been fabricating things like this for decades?
his methods don't lead to his results in a way that matches his predictions?
the creator of NLP is not qualified to decide whether or not his methods are NLP?

If there are no claims to any of the above, what exactly was discredited?

7ChristianKl9y

There's research that indicates that the NLP Fast Phobia Cure produce effects but there no research that it's better than other CBT techniques. I consider basic claims by Bandler about rapport as nowadays accepted by psychology as mimicry of bodylanguage. As far as I see nobody cited Bandler for that and mainstream psychology developed their ideas about mimicry separately decades later. The idea that there are eye accessing cues that are the same in every person that NLP taught in it's early days has been shown to be false in methodically bad studies and it's not taught anymore by Bandler and good modern NLP trainers. You will however still find articles on the internet proclaiming the theory to be true as claimed in the early days of NLP. In Bandler latest book he mostly talks about applying an idea about strengthing emotions that you want by spinning them in your body and disassociating negative emotions. I'm not aware about good published research around those mechanisms. Another significant claim of Bandler is that he can cure schizophrenics. I don't know his approach with schizophrenics works and as far as I know there no research investigating it. NLP trainers after Bandler are not in the habit of using language with the goal of saying things that are objective true, but focus on saying things that they believe will produce positive change in the person they are talking with. Bandler is not open about what he beliefs he's doing when he's training NLP trainers. Science itself rests on people openly stating what they believe. Bandler does tell people at the end of his NLP trainer programs that there no such thing as NLP, so the issue of whether he decides whether or not his methods are NLP is not straightforward. NLP works very differently with epistomological questions. It has a different approach to the question of how you teach a person skills to be a good therapist than mainstream psychology.

0Jurily9y

I'm aware that Strugeon's law is in full effect within the NLP community, my questions were specifically about Bandler and his results. I fail to see how anything you said has an impact on the observation that Andy did not need to return to the mental institute. Unless you dispute at least that single claim, the lack of research is better explained with the hypothesis that the researchers failed to understand the topic well enough to account for enough variables, like how Bandler almost always teaches NLP in the context of hypnosis. If whatever Bandler does is producing verifiable results, shouldn't it be at least an explicit goal of science to find out why it works for him, as opposed to whether it works if you throw an NLP manual at an undergrad? Shouldn't it be a goal of science to find out how he came up with his techniques, and how to do that better than him?

6jimmy9y

YES! Personally, I wouldn't take Bandler very seriously because of the whole "narcissistic liar" thing and the fact that the one intervention of his I saw was thoroughly lacking in displayed skill (and noteworthy result), but yes, you should look at the experts, not at the undergrads handed a manual designed by the researcher who isn't an expert himself. It's much better to study "effectiveness of this expert", not "effectiveness of this technique". I'd just rather see someone like Steve Andreas studied. I know from personal experience that even people with good intentions will strawman the shit out of you if you talk about this kind of thing because there's so much behind it that they just aren't gonna get. Ironically enough, Milton Erickson, the guy who Bandler modeled NLP after, allegedly had this exact complaint about NLP ("Bandler and Grinder think they have me in a nut shell, but all they have is a nutshell." )

0ChristianKl9y

A while ago I would have agreed, today I'm not sure whether that would go somewhere. I think you need researchers with both scientific skills and which actual abilities. Part of the reason why I respect Danis Bois so much is that after he was successful at teaching bodywork he went and worked through the proper academic way because he found the spiritual community to dogmatic. He got a real PHD and then a professorship. For hypnosis it likely would have to be similar. Someone who went deep into it. Who lives in the mental world of hypnosis and does 90%+ of his day to day communication in that mode but who then feels bad about the unscientific attitude of his community. A person who then starts a scientific career might really bring the field forward.

0jimmy9y

Yeah, I see the distinction you're getting at and completely agree. I was referring more to showing "hey, this can't be nonsense since somehow this guy actually gets results even though I have no idea what he's doing", which is an important step on its own, even if it's not scientific evidence behind individual teachable things.

0ChristianKl9y

Look at the state of pyschology today. They tried to replicate 100 findings. A third checked out. A third nearly checked out and another third didn't check out at all. If you are a psychologists at the moment and get embarrased as a result, you want to move in a direction where more results replicate. Studying highly performing people like Steve Andreas could very well not help with that goal.

0jimmy9y

Right. To me, that looks like a slightly different angle on the same thing. If you want to nail down some things so you can say "hey look, we know some things", then studying high performing people wouldn't be the way to go. If, on the other hand, you're pretty okay with saying "hey look, of course we don't know anything, that's why we're still in exploration mode, but look at all this cool shit we're sifting through!", then it starts to look a lot more appealing. It certainly doesn't surprise me that this kind of research isn't being done, and I can empathize with that embarrassment and wanting to have something nailed down to show the nay sayers. I also find it rather unfortunate. It strikes me as eating the marshmallow. Personally, I'd rather fast for a few days then drag back a moose.

0Lumifer9y

That, actually, depends on whether this cool shit is a stable pattern or just transient noise. Looking at cool-shit noise is fine as an aesthetic experience, but I wouldn't call it science (or "exploration mode" either). And, of course, there is the issue of intellectual honesty: saying "we found this weird thing that looks curious" is different from saying "we have conclusively demonstrated a statistically significant at the 0.0X level result". I suspect you'll go off chasing butterflies and will never get anywhere, if we're getting into hunter-gatherer metaphors.

0jimmy9y

That's a terrible aesthetic experience. Your sense of aesthetics is supposed to do something That's a very reasonable thing to suspect. It's a less reasonable thing to take as given, especially considering the size of the prize and the ease of asking a hunter "ever killed anything?".

0Lumifer9y

LOL. Besides the whole going-meta-on-aesthetics thing, wouldn't that depend on how cool the shit it? The hunter will proudly show you his collection of butterflies, all nicely pinned and displayed in proper boxes. Proper boxes are very important, dontcha know? I have a feeling we have different images in mind. You have a vision of intrepid explorers deep in the jungle, too busy collecting specimens and fighting off piranhas and anacondas to suitably process all they see -- the solid scientific work can wait until they return to the lab and can properly sort and classify all they brought back. I see a medieval guild of piece workers, producing things. Some things are OK, some not really, but you must produce the pieces, otherwise you'll starve and never make it from apprentice to the master. It would be, of course, very nice to craft a masterpiece, but if you can't a steady flow of adequate (as determined by your peers who are not exactly unbiased judges) pieces will be sufficient and the more the better.

0jimmy9y

The point is that how "cool" something is is supposed to track the potential value there. In practice it doesn't always (carbon fiber decals are a thing), but that just means they're doing it wrong. I'd find that very strange, but could happen. And if so, you can confirm your suspicion that they weren't getting anything interesting done. Still seems worth asking to me. It seems like you see me as implicitly asking "why do you guys keep making pieces instead of going on an adventure!?!?!" and answering with "you see epic adventure, but what they see is the necessity of making their pieces. If they didn't have to get their pieces made, and if there actually was epic adventure to have, of course they'd do that instead. It's that they don't agree with you". I agree. That's why they do what they do - 'twas never a mystery to me. I see room for that and epic and lucrative anaconda fighting adventures. Or for fools chasing that fantasy and running off into the jungle to starve. Or all three and more. I have a couple points here even before getting into what happens when you quit and seek adventure. 1) "you must produce the pieces". Really? You sure sure? What number do you put on that confidence? How you think you know? Often people get caught up running from what seems like a "must" only for it to turn out to be not mission critical. Literal hunger makes for a perfect example. When people fast for a few days for the first time, it often really changes the way they think about the hunger signal. It's no longer "You must eat" and instead becomes more of just a suggestion. 2) "I'm not convinced adventuring is worth it". Of course not. You haven't done your research. And from your mindset - if you really must produce the pieces, then you didn't need to. If I offer you a chance of a million dollars or a sure $500, but the mob is gonna kill you if you don't pay off your $500 debt, there's little point in asking what the chance is if you already know it isn't "all but gu

-1Lumifer9y

Nope. How useful something is is supposed to track the potential value. If I were to go meta, I'd say that "cool" implies a particular kind of signaling to a specific social sub-group. There isn't much "potential value" other than the value of the signal itself. Still nope. Most people don't want to go on a real adventure -- it's too risky, dangerous, uncomfortable. Most people -- by far -- prefer the predictable job of producing the pieces so that they can pay the mortgage on their suburban McMansion. In the case of academia, going for broke usually results in your being broke (and tenure-less) while a steady production of published papers gives you quite good chances of remaining in academia. Maybe not in the Ivies, but surely there is a college in South Dakota that wants you as a professor :-/ If you want tenure, yes. If you don't want tenure, you can do whatever you want. Sure. The answer is a shrug and if you want a verbalization, it will go along the lines of "Nobody knows". There is no way for all of them to "have much much more". Whether you think the trade-off is acceptable depends, among other things, on your risk tolerance, but in any case the mode -- the most likely outcome -- is still of you losing.

0jimmy9y

From here it looks like you aren't addressing what I'm actually saying and instead are responding to arguments you think I must be trying to get at. Are you sure you're being sufficiently careful and charitable in your reading of my comments?

0Lumifer9y

Sufficiently? X-D Clearly not.

1jimmy9y

Heh, okay. I'll try again from another angle. To be clear I do see the whole "intrepid explorers" thing pretty much exactly how you said it. I went that way myself and I'm super glad I did. It has been fun and had large payoff for me. At the same time though, I realize that this is not how everyone sees it. I realize that a lot of the payoffs I've gotten can be interpreted other ways or not believed. I realize that other people want other things. I realize that I am in a sense lucky to not only get anything out of it, but to even be able to afford trying. And I realize why many people wouldn't even consider the possibility. Given that, it'd be pretty stupid to run around saying "drop what you're doing and go on an adventure!" (or anything like it) as if it weren't that from their perspective not only is "adventure" almost certainly going to lead nowhere, but they must make the pieces. As if "adventure" actually is a good idea for them - for most people, all things considered, it probably isn't. My point is entirely on the meta level. It's not even about this topic in particular. I frequently see people rounding "this is impossible within my current models" to "this is impossible". Pointing this out is rarely a "woah!" moment for people, because people generally realize that they could be wrong and at some point you have to act on your models. If you've looked and don't see any errors it doesn't mean none exist, but knowing that errors might exist doesn't exactly tell you where to look or what to do differently. What I think people don't realize is how important it is to think through how you're making that decision - and what actually determines whether they round something off to impossible or not. I don't think people take seriously the idea that taking negligible in-model probabilities seriously will pay off on net - since they've never seen it happen and it seems like a negligible probability thing. And who knows, maybe it won't pay off for them. Maybe I'm

1Lumifer9y

Yes, I agree that people sometimes construct a box for themselves and then become terribly fearful of stepping outside this box (="this is impossible"). This does lead to them either not considering at all the out-of-the-box options or assigning, um, unreasonable probabilities to what might happen once you step out. The problem, I feel, is that there is no generally-useful advice that can be given. Sometimes your box is genuinely constricting and you'd do much better by getting out. But sometimes the box is really the best place (at least at the moment) and getting out just means you become lunch. Or you wander in the desert hoping for a vision but getting a heatstroke instead. You say but, well, should they? My "in-model probabilities" tell me that I'm not going to become rich by playing the lottery. Should I take the lottery idea seriously? Negligible probabilities are often (but not always) negligible for a good reason. Sure. But things have costs. If the costs (in time, effort, money, opportunity) are high enough, you don't care whether it's epsilon or a true zero, the proposal fails the cost-benefit test anyway.

1jimmy9y

Yes. From the inside it can be very tough to tell, but from the outside they're clearly they're wrong about them all being low probability. They don't check for potential problems with the model before trusting it without reservation, and that causes them to be wrong a lot. Even if your "might as well be 100%" is actually 97% - which is extremely generous, you'll be wrong about these things on a regular basis. It's a separate question of what - if anything - to do about it, but I'm not going to declare that I know there's nothing for me to do about it until I'm equally sure of that. I think one of the real big things that makes the answer feel like "no" is that even if you learn you're wrong, if you can't learn how you're wrong and in which direction to update even after thinking about it, then there's no real point in thinking about it. If you can't figure it out (or figure out that you can trust that you've figured it out) even when it's pointed out to you, then there's less point in listening. I think "I don't see what I can do here that would be helpful" often gets conflated with "it can't happen", and that's a mistake. The proper way to handle those doesn't involve actively calling them "zero". It involves calling them "not worth thinking about" and the like. There is nothing to be gained by writing false confidences in mental stone and much to be lost. Right. With the lottery, you have more than a vague intuitive "very low odds" of winning. You have a model that precisely describes the probability of winning and you have a vague intuitive but well backed "practically certain" odds of your model being correct. If I were to ask "how do you know that your odds are negligible?" you'd have an answer because you've already been there. If I were to ask you "well how do you know that your model of how the lottery works is right?" you could answer that too because you've been there too. You know how you know how the lottery works. Winning the lottery may be a very bi

0Lumifer9y

I don't know about that. That clearly depends on the situation -- and while you probably have something in mind where this is true, I am not sure this is true in the general case. I am also not sure of how would you recognize this type of situation without going circular or starting to mumble about Scotsmen. What do you mean, can you give some examples? Normally, if people locked themselves in a box of their own making, they can learn that the box is not really there. That's a good point -- I agree that if you don't realize what opportunity costs you are incurring, your cost-benefit analysis might be wildly out of whack. But again, the issue is how do you reliably distinguish ex ante where you need to examine things very carefully and where you do not have to do this. I expect this distinguishing to be difficult. "Actually thinking it through" is all well and good, but it basically boils down to "don't be stupid" and while that's excellent advice, it's not terribly specific. And "can you eat the loss?" is not helping much. For example, let's say one option is me going to China and doing a start-up there. My "internal model" says this is a stupid idea and I will fail badly. But the "loss" is not becoming a multimillionaire -- can I eat that? Well, on the one hand I can, of course, otherwise I wouldn't have a choice. On the other hand, would I be comfortable not becoming a multimillionaire? Um, let's say I would much prefer to become one :-) So should I spend sleepless nights contemplating moving to China?

0jimmy9y

I mean about the whole group of things that any given person decides or would decide is "low probability". I see plenty of "p=0" cases being true, which is plenty to show that the group "p=0" as a whole is overconfident - I'm not trying to narrow it down to a group where they're probably wrong, just overconfident. It's not that they can't learn that the box isn't really there, it's that even if they know it's not there they don't know how to climb out of it. There are a lot of things I know I might be wrong about (and care about) that I don't look into further. It's not that I think it's unlikely that there's anything for me to find, but that it's unlikely for me to find it in the next unit of effort. Even if someone is working with an obviously broken model with no attempts to better their model, it doesn't necessarily mean they haven't seriously considered the possibility that they're wrong. It might just mean that they don't know in which direction to update and are stuck working with a shitty model. Some things are like saying "check your shoelaces". Others are like saying "check your shoelaces" to a kid too young to know how to tie his own shoes. Heh. Yes, it is difficult and I expect that just comes with the territory. And yes, it kinda sorta just boils down to "don't be stupid". The funny thing is that when dealing with people who know me (and therefore get the affection and intent behind it) "don't be stupid" is often advice I give, and it gets the intended results. The specificity of "you're doing something stupid right now" is often enough. I'd much prefer to be a multimillionaire too, yet I'm comfortable with choosing not to pursue a startup in china because I am sufficiently confident that it is not the best thing for me to pursue right now - and I'm sufficiently confident that I wouldn't change my mind if I looked into it a little further. It's not that I don't care about millions of dollars, its that when multiplied by the intuitive chance that th

0ChristianKl9y

Science itself is about the search for finding knowledge and not about sifting through cool shit. I also consider it okay that our society has academic psychologists who attempt to build reliable knowledge. I think it's worthwhile to have different communities of people persuing different strategies of knowledge generation.

0jimmy9y

I don't disagree with any of the statements you made, and none of them are required to be false for my point to be valid. I'm kinda getting the impression that you aren't being very careful or charitable in your reading of my comments. Is that impression wrong?

0ChristianKl9y

I don't think the point of a post is to show how another person is wrong or to only say things where who I'm talking about is likely to disagree.

4ChristianKl9y

Given the current scientific framework you don't change a theory based on anecdotal evidence and single case studies. Especially when it comes to a person who's known to be at least partly lying about the anecdotes he tells. What do you mean with the phrase "explicit goal of science"? The goals that grand funding agencies set when they give out grants? To the extent that you think studying people with high abilities is good approach of advancing science, I wouldn't pick a person who's in the habit of lying and showmanship but a person who values epistemically true beliefs and who's open about what they think they are doing. I think the term pseudoscience doens't really apply for Bandler. For me the term means a person who's pretending to play with the rules of science but who doesn't. Bandler isn't playing with the rules or pretending to do so. That doesn't mean that he's wrong and what he teaches isn't effective but at the same time it doesn't bring his work into science. It's typical for New Atheists to reject everything that's not part of the scientific mosaic as useless discredited pseudoscience. I don't think that's useful way of looking at how the world works. If you want to go further into that direction of thought, a nice talk was recently shared on the Facebook LW group: Scientific Pluralism and the Mission of History and Philosophy of Science For full disclosure, I do have a decent amount of NLP training with Chris Mulzer who attended Bandlers trainer training program every year for a decade. I know multiple people who attended seminars with Bandler.

-2Jurily9y

Oh, I see the problem now. You're waiting for research to allow you to decide to do the research you're waiting for. When the scientific framework tells you there isn't enough research to reach a conclusion, doesn't it also tell you to do more research? Picking a research topic should not be as rigorous a process as the research itself. Even if all the anecdotal and single case studies are false, shouldn't you at least be interested in why so many people believe in it? NLP is not a religion, you pick it up as an adult. Even if the entire NLP/hypnosis/seduction/whatever industry is just a giant crackpot convention, they still demonstrate enough persuasion techniques to convince people it's real. Shouldn't you be swarming over that with the idea of eliminating your suicide rate?

2ChristianKl9y

What do you mean when you say "you"? I have more formal credentials with NLP then with academic psychology. I have multiple friends who makes their living in that industry. One of my best friends worked for a while as a salesperson for Bandlers seminars. I don't have friends who have as much friends who have degrees in academic psychology. I just understands both sides well enough to tell you about the situation we have at the moment.

3MrMind9y

NLP is arguably very difficult to analyze, because it's not a single body of coherent knowledge forming a model, rather than a mash-up of psychological techniques and some assertions about how the minds works. I think that when you can extrapolate something that is definitely an assertion about the mind or how some techniques improves the life of people who use it, then you can test it. And it's usually found to be false. There are however some assertions that turned out to be 'true' (that is, an experiment showed some effects), like the mirroring effect, and others that were borrowed from other fields or experiments. It's better not to be too hang-up about the pseudoscience label: just know that when you talk about NLP, you are entering in a field of not necessarily related beliefs which are mostly false.

-1knb9y

I think homeopaths and faith-healers could probably dredge up a few convincing-seeming anecdotes as well. The wikipedia article you linked to presents numerous meta-analyses in support of the claim that NLP is a pseudoscience. If you want to know what they think they've discredited, read them.

[-][anonymous]9y-20

Health is good. Intelligence is sometimes associated with health. Wikipedia has two articles on the relationship between intelligence and health. The other is here. So wouldn't you want to know more about your intelligence?

Clarity's index of sometimes subtle neurological conditions that impair cognitive functionl

Kleine–Levin syndrome

a rare sleep disorder characterized by persistent episodic hypersomnia and cognitive or mood changes. Many patients also experience hyperphagia, hypersexuality and other symptoms.
KLS can be diagnosed when there is confusion,

... (read more)

9ChristianKl9y

I don't think trying to diagnose themself via a bunch of descriptions of rare illnesses is a good road way to learn more about one's intelligence.

[-][anonymous]9y-40

i reckon the reason i have admired and.imitated entrepreneurship, innovation and initiative above their market value, social and financial, is that it seems to signal reaponsibility, in spite of circumstanes for which their solution may not be apparent for a particular problem. But there are a whole class of.other values and hehaviours that entail.reaponsibility and I reckon the quality of those particular aforementioned signals to imply responsibility has and is decaying as I mature. Duty, and consistency, are the traits a value them on par with now, things I had neglected before.

[+][anonymous]9y-140

Moderation Log