All of nebulous's Comments + Replies

I was confused about Solomonoff induction a while ago. Since code from any part of whatever program is running could produce whatever string is observed, why would shorter programs be more likely to have produced the observed string? My understanding of the answer I received was that, since the Turing machine would produce its output linearly starting from the beginning of the program, a program with extra code before the piece that produced the observed string would have produced a different string. This made sense at the time, but since then I've thought... (read more)

In order for the prior probabilities you assign to programs to be well-formed, they must limit to 0 as the length of the program goes to infinity. Otherwise the probabilities won't add up to 1. No matter what prior you choose, you'll always end up with a variant of Occam's Razor where programs beyond a particular length are always assigned very low probabilities. You don't necessarily have to decrease proportionally to 2^(-length), instead of say 3^(-length) or 1/Ackermann(length), like the Solomonoff prior does. However, since each additional bit lets you identify a value from a space that's twice as big (i.e. the precision grows like 2^length), it seems like a natural choice to penalize by that much. (Personally, I prefer a prior that also lightly penalizes for running time. That removes computability issues and makes the induction approximable by dove-tailing.)
Solomonoff induction is a formalization of Occam's razor. You may find this article useful.

I'd wondered why no one used a time-turner the moment they knew a troll was loose. Even if Dumbledore had already used up his hours, another professor could've used some form of priority magical communication to call for aurors to travel six hours into the past, swiftly prepare to deal with a Hogwarts-attacking troll, and teleport to the site. Then I realized that Quirrell could prevent all attempts to stop the troll using time travel by exploiting the restriction against information traveling back more than six hours, i.e. by waiting until six hours after he wanted the attack to start, traveling back six hours, and initiating the attack.

I thought she mostly understood his sentence (though of course she hadn't known about ELIZA beforehand) and owned a few magical items that could talk to a limited extent.

Augh, right. I'd forgotten that was there.

Where are resources for finding an effective, context-appropriate exercise routine?

Career interest: Eventually founding an IT startup, as per recommendation by Carl Shulman. Motivation: Making lots of money to donate to effective charities. Background: My dad is a freelance (Windows) computer assembly and repair guy, and I picked up some troubleshooting and upkeep tricks from that, but nothing impressive. I also took a computer science class where I gained some ability in Java.

I love this goal!
If your goal is to found an IT startup, I'd recommend learning basic web development. I formerly used rails and, at the time I picked it up, the learning curve was about a month (just pick a highly rated book and work through). If not web, consider app development. If you know a bit of Java, Android would probably be the way to go. With either of these, you'll have a skill that allows you to single-handedly create a product. At the same time, start keeping a list of ideas you have for startups. Some will be big, others small. But start looking for opportunities. Particularly focus on those that fit with the skills you're learning (web or app). Potentially, that leaves you two months to start your first startup. Doesn't have to be great. Doesn't even have to be good. But knowing that you can take something from idea to product is extremely powerful. Because now, as you're learning, when you see an opportunity, you'll know how to take it. More, it will allow you to fit your studies into your ideas. In your algorithms class, you'll see techniques and realize how those could solve problems you've had with your existing ideas or spark all new ideas. And if you don't walk out of your first AI class with a long list of new possibilities, something went seriously wrong :). But everything you're learning will have a context which will be extremely powerful. All this time, keep creating. Any good entrepreneur goes through a training process of learning how to see opportunities and take them. You have four years of access to excellent technical resources, free labor (your peers), and no cost to failure (and learning how to handle those will be another step in your growth). If you go in with an ability to create (even a very basic ability), you will not only be able to make use of those opportunities, you'll get far more out of the process than you otherwise would. [also: I'd like to second the recommendations to establish an exercise habit]

A basic grasp of Java. I felt like there were other skills, but they're unremarkable in the circles in which I'll spend my time--above average vocabulary, general knowledge base, and dedication to studying for my school's environment, and Less Wrong memes.

Here's my recommendation: build a web application that stores and retrieves data in a back-end database, from scratch. In the real world, you're not given exercises to test specific skills; you're given problems you have to solve. Sometimes you will have no idea how to solve them. Sometimes you'll end up solving them badly, and you'll learn from your error. Sometimes you will have to learn a whole new skill at a rudimentary level on a very short time scale. Sometimes you will be forced to steal something someone else has done with minimal idea of how it works. The real world is ugly like that. You are, of course, at liberty to not accept the challenge, but if you do, here are my recommended ground rules: * You have to put the entire stack together yourself. You build the server (if you're especially hardcore, you could construct a physical server yourself, but I don't think anyone would blame you for just running a VM installation), you configure it to serve web content, you build the database, you write the web application and you test and debug it. * No third-party built items. Wordpress is not allowed. Prefabricated web server VMs are not allowed. You are allowed to steal code snippets and config files, but you have to understand what they're doing. * You have to pick data of moderate complexity for it to store and retrieve. Something on the scale of a music database, with artists, albums, track titles, genre, year of publishing, etc. * You have to choose what technologies you use. If you want to run PHP on an IIS webserver, talking to a Progress database, no-one can tell you otherwise. * You're encouraged to ask other people for help, scour web forums, and learn from other people's examples. Stack Exchange is highly recommended. * If you have an idea for a different project of similar complexity while doing this, do that instead. * When you're not working, enjoy yourself. This is a common project for CS undergrads, so if you find yourself doing

I simply averaged the four numbers on those countries. I'll edit the post to have a weighted average by number of nets distributed. I don't know how to account for disproportionate early deaths in my calculations, since I don't have data on the typical lifespan of, for instance, a Zambian who survives childhood.

Just to be clear: I am not objecting that your numbers are imprecise. Your argument doesn't require precision. I have objected to incompatible levels of precision involved in the presentation: rough guesses and systematic errors of order 10% or more (some of them unavoidable) on the one hand and five-decimal figures on the other hand.

I refrained from rounding until the end so that if people were following my calculations starting from partway through they would arrive at the same answers. It wasn't really necessary, and now that you mention it it does raise questions about significant digits, so I'll round midway figures for display in the future.

Good point on the life expectancy being given for people currently born. I'll edit the post to use life expectancy figures from ten years ago.

If this is motivated by desire for trustworthiness, linking to the source of the life expectancy figures should have higher priority. (Also, how did you calculate the average? Weighed by overall population of those countries, population under ten, number of AMF clients, simply averaging the four numbers, ... ?) If you used the ten years old data, you would most probably obtain lower expected age at death for today's ten-years olds* than for today's newborns. But it should be higher. The problem is not that the life expectancy changes during the time. The problem is that the life expectancy at birth is average of life times of all people born, while the life expectancy at age of ten is average of life times* of all people that survived until this age. The latter is higher than the former, since all children who died before their tenth birthday (and who lower the former average) are excluded from the statistics. In e.g. Zambia, about 11% of children die before age of five, so you can imagine how this influences the discussed difference. (I was unable to quickly find data for this to illustrate the difference explicitly.) * Note that life expectancy at age x usually means the expected remaining time of life, not the expected age of death (which obviously is obtained from the former by adding x). (Edited.)

I had more ideas for Less Wrong posts, including an argument that donating to charity is more beneficial than paying for cryonics, but that assumes that the reader is altruistic. Since most people on Less Wrong are apparently not altruists, should I go ahead and post "For altruists, AMF > cryonics" here, or should I keep altruism-assuming arguments for some community more typically altruistic, like Felicifia?

I'd like to see it.
I'm not actually sure that's true. Though those of us who are were probably completely unimpressed with this article, as we've already been familiar with Givewell (for example) for years.
Make the posts, just abstract the language up one level, as you suggest.

I agree that existential risk is a higher priority. I used AMF in my example because its benefits are easy to accurately quantify.

Ah, thank you. Now I know more proper uses of the words "utilitarian" and "altruist", that should help me communicate.

Edit: Just read thomblake's comment. Now I'm back to using "utilitarian" to mean "altruistic value-maximizer".

It is obvious that the trade off is there--I thought that people weren't taking the option of helping people at their own expense because they didn't know that that option caused more benefit overall than the option of having fun at others' expense. The reason that people who know about efficient charity aren't helping people at their own expense is instead apparently an objection to utilitarianism in general. I had thought before posting that most people at Less Wrong were utilitarians.

As for your questions, I'm a high school student, so I want to spend my money on college to increase my chances of making much more money later in life so that I can donate more to efficient charities.

Some amateur etymology, for those enamored of distinctions: Most people here seem to be consequentialists, and consequentialism is merely the view that an action's goodness is defined by its outcome. Most are also 'utilitarians', seemingly meaning that they believe expected utility maximization is the correct metric for ranking outcomes, where 'utility' refers simply to the output of an agent's utility function. Probably most are also 'altruists', where altruism is the view that doing good things for others is good for oneself. This is a departure from how the terms are used in the ethical literature. There, 'utilitarianism' tends to refer to an agent-neutral consequentialist moral theory; that is, my survival doesn't matter any more than yours. Also, 'altruism' in ethics refers to a theory that looks just like utilitarianism but discounts the agent entirely; think Christian self-sacrifice. Our use of the term 'utility' has its roots in game theory, decision theory, and economics; our 'utilitarianism' seems to be an offshoot of that. Our use of the term 'altruism' seems to come from anthropology by way of evolutionary psychology.
Utilitarianism isn't a synonym for altruism. You can be a utilitarian and value your own happiness above that of others.

From the Thiel Fellowship website: "The ideal candidate has ideas that simply cannot wait. She or he wants to change the world and has already started to do it in some fashion. We want fellows who dream big and have clear plans [...]" I'm interested in what you have in mind for your Thiel grant--some singularity-promoting project? Have you already done some work in that direction?

I have a lot of ideas. Mainly education reform and media curation and community building, but also some ideas for the application of environmental technologies. But, currently lack business knowledge. Working on that right now (I'm reading the Personal MBA and Getting Things Done. 4 steps to epiphany is on my list. I'm hoping that if I got the scholarship, I would be directed to mentors and resources that would help me pick up the business information I lack). I will be making a post soon about my ideas in more depth shortly. They will be imbedded within one of those "how can I help?" style posts.
  1. 17, senior.
  2. Not when thinking about it that way, though recreational research in pop sci, apologetics, and epistemology have gotten me to the point where I'm much more knowledgeable in those areas than my average classmate.
  3. None in my family, and one in my school--the calculus and Bible teacher, very into apologetics. We argue for ~45 minutes every Tuesday thanks to my school's shuffling schedule.
  4. Highest paying career I can attain, don't know what yet. Whatever school I can enter that gets me to the career.
  5. After reading the winners of the grants and noting that they have absurd accomplishments, no.

Thank you. I had assumed that Less Wrong had no private messaging when the envelope icon near the top right corner of the interface took me to the reply to my first comment.

I've briefly tried to find a way to contact Shulman and failed. Is there a known way to contact him? Possibly useful information: I would prefer an IRC session over Skype. I've already followed the posted link, googled his name, googled his name with the word contact, looked at' s contact page, and googled his name while restricting the search to

Have you used Less Wrong's "Send message" functionality? (Edit: It's a button in the upper right corner of the page you get when you click someone's username.)

Is the linked website right about banking being the optimal career path for professional philanthropy or is there a more efficient method of moving resources to charities? I'm especially curious since I'll choose my degree soon.

You may wish to request a Skyped career counseling session from Carl Shulman, one of the experts behind High Impact Careers.