Recent Updates Page 4 Toggle Comment Threads | Keyboard Shortcuts

  • 9:18 pm on September 19, 2015 Permalink | Reply  

    slatestarscratchpad shlevy Seen on my facebook timeline Just… 



    Seen on my facebook timeline. Just a reminder of how ridiculous you look when you talk about things you have no experience with or knowledge of.

    I’m not sure what’s got you so angry, but psychiatry is about 50% from column A, 50% from column B except with “drug companies” in place of “government”.

    This figure reflects my experience as a working scientist.

  • 6:14 pm on September 6, 2015 Permalink | Reply  

    Superintelligence, a review in many parts 










    To continue on my response to slatestarscratchpad‘s question about people who have read Bostrom and remain unconvinced, I thought I’d write a longer series of posts.  These same points were made by shlevy more succinctly in his goodreads review.  “that’s completely wrong and anyone with a modicum of familiarity with the field you’re talking about would know that”

    Part 1: Igon value problems.  

    In a now famous review of a Malcolm Gladwell book, Steven Pinker coined the phrase “igon value problem” to refer to Gladwell’s tendency to expound at length on various topics while also making very superficial errors.  i.e. when discussing statistics,etc Gladwell refers to igon values instead of eigenvalues

    Bostrom’s Superintelligence is loaded with these “igon value” problems.  Chapters 1 and 2 are particularly bad for this, as they talk a lot about past AI and current state of AI.  

    A few examples I recall off hand: Bostrom says genetic algorithms are stochastic hill climbers.  This isn’t true- the whole point of genetic algorithms is breeding/crossover in order to avoid getting stuck in local optima (like hill climbers do).  It wouldn’t be worth the work to recast a problem using a genetic algorithm, stochastic hill climbers are easy to write.  

    He says you can compare many types of algorithms because they are doing “maximum likelihood estimation” but most of the algorithms he lists can do more than maximum likelihood, and some of the algorithms he lists are non-parametric (decision trees).  

    Bostrom says that machine learning algorithms are making mathematically well specified trade offs from an “ideal Bayesian agent,” which I’ve expounded on at length on my tumblr blog.  There is no controlled approximation for an “ideal Bayesian agent.”  There is no mathematical sense of “how far from an ideal Bayesian agent” a model is.  

    These aren’t isolated mistakes, these igon value issues happen all over the place.  

    Now, on one hand, these mistakes aren’t huge and a lot of thrust is mostly correct if you squint at it a bit and ignore the details (which are misleading or just outright wrong). 

    But at the same time, the rest of the book is speculation that is only grounded in Bostrom’s understanding of AI.  Do I trust someone with an igon value understanding to reliably extrapolate from the state of the art today?  My answer is no, I don’t.    

    Ouch. Depending on how the rest of this series turns out, I may have to stop saying that while I’m down on MIRI, I have a lot of respect for Bostrom…

    None of those errors (assuming Bostrom is being represented accurately) is anywhere close to the Igon Value problem and the comparison just sounds partisan.

    Maybe Superintelligence is bad and Bostrom is dumb. But this post is not persuasive in that regard.

    Okay, I’m no longer on mobile and I have my copy of Superintelligence in front of me. Page 8 of the hardcover edition:

    In evolutionary models, a population of candidate solutions (which can be data structures or programs) is maintained, and new candidate solution are generated randomly by mutating or recombining variants in the existing population. Periodically, the population is pruned by applying a selection criterion (a fitness function) that allows only the better candidates to survive into the next generation.Iterated over thousands of generations, the average quality of the solutions in the candidate pool gradually increases.


    In practice, however, getting evolutionary methods to work well requires skill and ingenuity, particularly in devising a good representational format. Without an efficient way to encode candidate solutions (a genetic language that matches latent structure in the target domain), evolutionary search tends to meander endlessly in a vast search space or get stuck in a local optimum.

    So Bostrom clearly groks genetic algorithms. (Certainly it reflects my experience with them.)

    Okay and here’s the passage @su3su2u1 took issue with, on page 9:

    In fact, one of the major theoretical developments of the past twenty years has been a clearer realization of how superficially disparate techniques can be understood as special cases within a common mathematical framework.


    In a similar manner, genetic algorithms can be viewed as performing stochastic hill-climbing, which is again a subset of a wider class of algorithms for optimization.

    So I think the original criticism is even less fair than I’d thought. Disappointing.

    So the first section looks like a textbook description of a genetic algorithm. 

    The second section, the one you are quoting, is just flat out wrong.  Stochastic hill climbing is a local search, genetic algorithms are more broad than local search, and everything is a subset of “algorithms for optimization” because it’s so vague as to be undefined.  

    It’s that sort of “igon value” wrong where if you squint and move past it it isn’t going to break the book.  Which is exactly my criticism. 

    “evolutionary search tends to meander endlessly in a vast search space or get stuck in a local optimum.“

    Do you disagree?

    I’m really just baffled here because I’ve pretty much always heard and read GAs discussed as stochastic hill-climber variants, e.g. as an alternative to simulated annealing.

    The typical failure mode I tend to see is that they converge too quickly on non-optima (to just arbitrary points).  

    Where are you seeing GA discussed as hill-climbers?  That’s just not what they are.  They are usually discussed as alternatives to hill climbers for situations where you expect the hill climber to get stuck.  The whole point of mixing (breeding and mutation) is to perform a non-local search- you hope to bust the population out of the local optima.  The typical use case of genetic algorithms is when we expect lots of local optima where hill climbers will get stuck.  

    The typical comparison is between genetic algorithms and simulated annealing because they are both commonly used on the same sorts of problems (global optimization in situations where there are lots of local optima). 

    Evolutionary search is not the same as local search, they aren’t doing the same thing.  

    But basically they are. The programmer’s choice of representational format is where the work happens (as Bostrom points out). Evolutionary search is local search with a population.

    Wikipedia also classifies genetic algorithms as stochastic optimizers, FWIW.

    I do not think there is a wikipedia page which takes as a given the Igon Value spelling. Your comparison remains unfair.

    Stochastic optimizers are NOT stochastic hill climbers.  Any optimizer with a random variable is a stochastic optimizer.  You could make a stochastic hill climber by tying in a random variable.  Now it’s a hill climber that is a subclass of stochastic optimizer.  

    Simulated annealing will also be in there, probably under metaheuristic (or maybe randomized search).  Genetic algorithms will be in a similar category.  Is simulated annealing also a hill climber?  Are we going to abuse terminology that much?  

    Hill climber = local search.  LOCAL search.  Not evolutionary search.  Words mean things. 

    Genetic algorithms are like hill climbers in the same way bananas are like apples.  They are both fruit! So I wasn’t wrong when I said that the banana was an apple. 

    Simulated annealing is much closer to genetic algorithms in use case than genetic algorithms are to hill climbers. See literally any book on optimization alogirhtms.  See wikipeidas genetic algorithm and hill climber page. 

    If anything, Bostrom is guilty of not understanding that “hill climber” is a specific term of art. “Iff it uses a local gradient approximation.”

  • moridinamael 3:42 pm on September 4, 2015 Permalink | Reply  

    Trump Dynamics – Lessons from Media Criticism 

    I’m greatly enjoying the current popularity media adoration phenomenon of the Trump presidential candidacy.

    Many smart people have offered theses explaining his rise. Perhaps my favorite aspect of this story is that absolutely nobody predicted it. This highlights that everybody’s models are severely insufficient, and that’s very interesting.

    I can’t go back in time and predict it, but I do have a very simple theory that explains all the data.

    See, I have recently been co-running a movie/book criticism blog, and as a consequence I’ve learned a lot about how Media operates. For example, a sober and objective review of a new film might draw clicks from the demographic of people who are thinking about going to see that movie, and people who have seen it and want to know what other people think of it. This is potentially a large demographic – the entire moviegoing public theoretically falls in this demographic.

    Yet this type of review sees relatively few clicks. The reasons for this are twofold: (1) there is too much competition. There are too many movie review blogs, and other sources of information, competing for the eyes of anybody looking for the scoop on any given movie. And (2), sober, objective reviews of popular movies are inherently non-viral. You may indeed find our blog and read a review; you may enjoy the review; it may influence your thinking on a movie. But you’re not going to Retweet it, or Like it. I know this because nobody Likes or Retweets our straightforward reviews of popular movies. There’s nothing in them that inherently makes you think, “Oh, people gotta see this.”

    What type of content sees a lot of clicks?

    My essay on Neal Stephenson (focusing on Seveneves but ranging over many of his works) is our most popular article. This is a long-winded and ranging essay where I throw out mildly controversial political thoughts and use the words “Libertarian” and “Reactionary” several times. This article was Retweeted beyond my capacity to easily track, and was linked by Vox Day and by Xenosystems.

    By being mildly controversial in a way that certain subgroups found particularly relevant, I obtained mild virality.

    Our second-most popular article is my review of Worm. I said nothing particularly controversial, but the Worm fandom is sizable and fanatical enough to read pretty much anything about Worm that comes along. So, what the Worm review has in common with the Seveneves review is not that it appeals to a wide demographic, but that it appeals specifically to a small demographic.

    Trump is viral as hell. Self-consciously so. Everything he says is Retweetable. Maybe because it’s unbelievably dumb, or cartoonishly bombastic, or shockingly confrontational, or just embarrassingly gauche – whatever the case, it’s always interesting enough for a Like/Share/Retweet. (The same concept of virality applies in a different way to non-social forms of media. If you’re CNN, you have to cover Trump because (a) FOX is covering Trump and (b) because you can’t just not cover it when he says and does these unbelievable things.) And it doesn’t matter that his behavior makes him “not a serious candidate” if the virality he cultivates overwhelms the establishment desire to silence him.

    The above paragraph is the crux of my thesis. Anything further would be adding spandrels.

  • 6:34 pm on September 3, 2015 Permalink | Reply  

    ozymandias271 liberals the problem with Kim Davis is… 



    the problem with Kim Davis is that she’s a homophobe. The problem is not that she is willing to go to jail rather than obey a law she finds immoral. Homophobia is wrong. Refusing to obey laws you find immoral is admirable. 

    Why talk about that when we can talk about the fact that she’s a hypocrite? Hypocrisy being the gravest imaginable sin.

  • moridinamael 5:46 pm on September 2, 2015 Permalink | Reply
    Tags: munchkin   

    Low-Hanging Fruit 

    Urban Dictionary defines Low-Hanging Fruit as: Targets or goals which are easily achievable and which do not require a lot of effort.

    I am always interested in collecting examples of low-hanging fruit that I haven’t heard of. Here are some of my favorites.



    Whether you are just learning to program or you are a hardened programming veteran, Stackoverflow is the website you will be using. It is supremely well organized in terms of specific programming questions, designed to reduce redundancy in those questions, and the best answers are upvoted by the community so you can have confidence in the best solution. Very high-value resource.

    I’ve used Mint to keep track of my financials for a while now and it’s been pretty helpful. It’s nice for a number of reasons to only have to look at one website to see all your accounts. You can also keep track of budgeting if your into that, and keep track of your spending history.

    Amazon Prime

    I mentioned this on the call, and I think everybody may already have this one?  But anyway, totally worth the cost. I will let this guy explain why. Basically … after you get in the habit of using it, you realize that you just don’t need to go to the store very often.



    I get cold easily and I hate to be cold. This year I discovered scarves and learned that simply keeping the neck area covered reduces my subjective sense of suffering due to cold substantially. There’s a scientific basis to this; there are sensitive areas of the body, notably the neck, that it really pays to keep covered.

    Macbook Air

    Not “low-hanging” perhaps because it is expensive, but this is the only computing product I’ve ever owned where I was not disappointed in some way after buying it. It is just exactly what I wanted it to be. I guess I should note here that I only use it for writing and coding and I do zero gaming.

    Snap Ware

    You’re probably still using Tupperware or some shit to store food. This is a terrible mistake. Buy some Snap Ware immediately. It is better in every possible way. It is microwaveable, you can eat out of it, it is sturdy, the lid snaps firmly into place, and once so snapped, you can hold it upside down and the seal is so tight it will hold back liquids. Good shit right here. We got ours as a wedding gift (two years ago) and have used it hard, and it hasn’t showed any signs of wear. After glancing at the Amazon product reviews, I suspect people are accelerating wear on the lids by putting the lids in the microwave when the heat up their meals, so just don’t do that.


    Gas Up on Mondays

    Filling up your car with gas on Monday seems to statistically result in paying the least for gas in the long run. I threw this in because y’all expected me to say something related to oil & gas. There are apps you can use to track local gas prices but then you’re literally spending more time-value working to earn a few dollars back than you’re saving in fuel costs.

    Go to the Doctor for That Thing

    You know that thing that’s been bothering you for the last fifteen years?  Go see the doctor about it. It really isn’t that much of a hassle. You’re already paying for this service – that’s what insurance is for. It will turn out to not be serious, the doctor will have some cream or pill that will immediately fix it and all the psychic weight you don’t realize you’ve been carrying all this time will vanish immediately.

    Dry Cleaners

    For my first two jobs I washed and ironed all my own shirts and pants. This takes a lot of time and I frankly don’t do a very good job of it. Use a god damn dry cleaning service. It’s not that hard to set up and once you get into the routine of it, it’s just like any other thing. From a time-value perspective it’s a lot cheaper than ironing your own damn shirts.

    Things I Haven’t Tried But Sound Cool

    You know I’m fond of trying crazy things, but here are some things that seem relatively easy to do that I just haven’t done due to not having low-hanging-enough fruit (ha ha!) to attempt it.


    Modafinil is a drug you can take that will make you not need to sleep and doesn’t really seem to have any bad side effects. People report popping some of this stuff and staying up for forty hours – basically working all day, all night, all day the next day, and going to bed at their normal bed time the next night, and not crashing, not needing to sleep for twenty hours to recover, not dying or vomiting blood, etc. Basically it violates everything we’ve been taught to believe about how sleep works. I feel like if I took it, it would causes me to just drop dead due to my neurological fragility, so I’m not going to risk it, but I’m curious so, you know, knock yourselves out guys.

    Virtual Assistants

    You can get what is basically an online secretary in India for a really, really cheap hourly rate. Services like this have now existed for like ten years or more, and they’re still in business, and you can read reviews of the businesses online, and you would think if they were just stealing people’s identities left and right you would have heard about it, so one can assume it’s probably legit, right?  It seems really compelling to me, except every time I come to the point of actually doing it, I realize, what do I really need a secretary for?

    Your Turn

    I posted this to trigger you guys. These are the places my head goes to when I try to think of the types of things that make my life easier. If I didn’t know about these things, I would wish somebody would have told me about these things.

  • 1:59 pm on July 31, 2015 Permalink | Reply  

    asocratesgonemad Warning Sex gender opinion that I feel… 


    [Warning: Sex/gender opinion that I feel really really bad about having but do pretty much endorse]

    Keep reading

    My wife is closer to being an actual super-genius programmer/engineer than any men I know, maybe you just need to hang out in university/government labs and recalibrate your expectations of what’s plausible.

  • 2:27 am on July 31, 2015 Permalink | Reply  

    Dear white Vegans: This is Your Collection Agency Calling 


    This shit white vegans call “speciesism,” where human beings are supposedly “privileged” as a species over animals, doesn’t exist and here’s a reason why: there are nearly three times as many animal shelters in the US as there are shelters for battered women and their children. And this isn’t a new phenomenon. It is far less political, controversial, and socially inflammatory to house animals than it is to house abused women, which should immediately demonstrate structural priorities and “privileges” when it comes to these two “species.”

    When white vegans compare the consumption of certain animals to racism, they are drawing a figurative comparison between human beings (who “oppress” animals) and white people (who oppress POC). By that white logic, they are also drawing a figurative comparison between animals (“oppressed” by human beings) and POC (oppressed by white people). This means they are reproducing the same kind of racist arguments that equate white folks with human beings and POC with animals. Sitting on my privileged white ass eating a hamburger is not now and will never be the same as racism. 

    The cruelty, brutality, and abuse of factory farming is reprehensible. No one is denying that. But when white vegans compare eating meat to genocide, they ignore that this small privileged demographic (in comparison to the global population) is consuming excessive amounts of meat, which doesn’t even nearly constitute the same worldwide oppression and genocide POC experience. Claiming animals deserve the same civil/human rights as POC suggests that their experiences are identical, and once again equates people of color with animals. 

    Consider the racial and class privilege white vegans have when making these dietary decisions in the midst of numerous farmer’s markets, health food stores, or accessible concentrations of grocery stores period. Whether it’s the ability to afford organic produce or the individual free time to prepare vegan meals, privileged personal consumer choices are not a solution to the abuse of factory farming or the social inequalities of racial oppression. Vegans with white skin maintain considerable white privilege over human beings of color who contend with the institutional and environmental racism of being segregated into neighborhoods without grocery stores, farmer’s markets, or spare land for home vegetable gardens.

    As a side note: refusal to wear animal products like skins, furs, and/or leather does not morally free white vegans from wearing clothing made by oppressed people of color in sweat shops–even if that clothing comes from a thrift store.

    Vegans, and white folks in general, are not qualified to compare or equate racial oppression with other forms of oppression (real or invented), because we do not know what racial oppression is like from our own experience. When millions of domesticated animals eat better than millions of POC living in poverty, a white justice crusade of personal food politics on the behalves of “oppressed” animals becomes its own punchline. What’s next? Extraterrestrialism? Where aliens oppress human beings with the use of anal probes? Wow, wait a second… isn’t it mostly white rednecks who get anally probed? Did I just discover reverse racism in extraterrestrials? Alert the white liberal media, yall.


    Speciesism. Noun. Definition: This post.

  • 3:42 pm on June 19, 2015 Permalink | Reply  

    itsvondell at the end of All Yesterdays the… 


    at the end of All Yesterdays (the extremely good book about imagining and illustrating dinosaurs in complex speculative ways i was talking about yesterday) there’s a section where they prove the point about the fact that we need to be more open to imagining skin coverings and fat/cartilage deposits by illustrating modern-day animals as if a nonhuman paleontologist from millions of years in the future reconstructed them using the just-skin-stretched-over-the-skeleton-and-muscles method that unimaginative paleoartists use with dinosaurs

    with results like:



    and i love it so much because it absolutely unquestionably proves the point the book is making

  • 8:44 pm on June 15, 2015 Permalink | Reply  

    Anonymous I’ve actually been to a physics colloquium… 


    I’ve actually been to a physics colloquium about time crystals, one of the big ‘problems’ with having the crystal persist after the universe (according to the speaker) is that the darn lab that’s keeping it stable doesn’t exist anymore.

    Oh this is easy, you just encase the time crystal inside a cube of other time crystals, creating a Timecube.

    Trust me, I’ve read about this.

  • 4:24 pm on February 4, 2015 Permalink | Reply  

    Abortion rescue simulations 

    Somebody will eventually retrieve the lives of all aborted fetuses via rescue simulations.

    In the end, these rescued souls will be grateful at having been rescued, and queasy at the thought that they were robbed of full lives in the historical worldline. There will be very little to distinguish them from rescued copies of murdered five-year-olds, with respect to their attitudes and memories of the historical worldline.

    There is a difference between a rescued aborted person and a hypothetical person, e.g. a person who would have existed IF a certain act of intercourse had counterfactually resulted in a pregnancy. The difference is that the rescued abortion victim will feel like they were intentionally robbed of something in the historical worldline, whereas the rescued hypothetical person will not feel any sense of injustice, and will feel simple gratitude at the chance to exist.

    Whether injustice has been done depends a lot (entirely?) on whether somebody feels or counterfactually feels that injustice has been done.

    I hope these simple straightforward conclusions have clarified your thinking on the matter.

Compose new post
Next post/Next comment
Previous post/Previous comment
Show/Hide comments
Go to top
Go to login
Show/Hide help
shift + esc