Posts

Long-chain correlation: lead paint and crime 2013-01-19T19:47:37.077Z · score: 13 (14 votes)

Comments

Comment by chrishibbert on Be secretly wrong · 2016-12-10T17:42:07.360Z · score: 0 (0 votes) · LW · GW

I'm all about epistemology. (my blog is at pancrit.org) But in order to engage in or start a conversation, it's important to take one of the things you place credence in and advocate for it. If you're wishy-washy, in many circumstances, people won't actually engage with your hypothesis, so you won't learn anything about it. Take a stand, even if you're on slippery ground.

Comment by chrishibbert on A Visualization of Nick Bostrom’s Superintelligence · 2014-07-26T16:35:42.749Z · score: 2 (2 votes) · LW · GW

To begin with, there are significant risks of medical complications—including infections, electrode displacement, hemorrhage, and cognitive decline—when implanting electrodes in the brain.

This is all going to change over time. (I don't know how quickly, but there is already work on trans-cranial methods that is showing promise.) If we can't get the bandwidth quickly enough, we can control infections, electrodes will get smaller and more adaptive.

enhancement is likely to be far more difficult than therapy.

Admittedly, therapy will come first. That also means that therapy will drive development of techniques that will also be helpful for enhancement. The boundary between the two is blurry, and therapies that shade into enhancement will definitely be developed before pure enhancement, and be easier to sell to end users. For example, for some people, treatment of ADHD spectrum disorders will definitely be therapeutic, while for others it be seen as attractive enhancements.

Not only can the human retina transmit data at an impressive rate of nearly 10 million bits per second, but it comes pre-packaged with a massive amount of dedicated wetware, the visual cortex, that is highly adapted to extracting meaning from this information torrent and to interfacing with other brain areas for further processing.70 Even if there were an easy way of pumping more information into our brains, the extra data inflow would do little to increase the rate at which we think and learn unless all the neural machinery necessary for making sense of the data were similarly upgraded.

The visual pathway is impressive, but it's very limited in the kinds of information it transmits. It's a poor way of encoding bulk text, for instance. Even questions and answers can be sent far more densely with a much narrower channel. A tool like Google Now that tries to anticipate areas of interest and pre-fetch data before questions arise to consciousness could provide a valuable backchannel, and it wouldn't need near the bandwidth, so ought to be doable with non-invasive trans-cranial techniques.

Comment by chrishibbert on Solomonoff Cartesianism · 2014-03-08T18:26:32.799Z · score: 0 (0 votes) · LW · GW

I'm confused by the framing of the Anvil problem. For humans, a lot of learning is learning from observing others, seeing their mistakes and their consequences. We can predict various events that will result in other's deaths based on previous observation of what happened to yet other people. If we're above a certain level of solipsism, we can extrapolate to ourselves.

Does the AIXI not have the ability to observe other agents? Is it correct to be a solipsist? Seems like a tough learning environment if you have to discover all consequences yourself.

It's still possible to extrapolate from stubbing your toe, burning your fingers on the stove, and mashing your thumb with a hammer. Is there some reason to expect that AIXI will start out its interactions with the world by picking up an anvil rather than playing with rocks and eggs?

Comment by chrishibbert on 2013 Less Wrong Census/Survey · 2013-11-30T19:56:01.198Z · score: 12 (12 votes) · LW · GW

I don't answer survey questions that ask about race, but if you met me you'd think of me as white male.

I'm more strongly libertarian (but less party affiliated) than the survey allowed me to express.

I have reasonably strong views about morality, but had to look up the terms "Deontology", "Consequentialism", and "Value Ethics" in order to decide that of these "consequentialism" probably matches my views better than the others.

Probabilities: 50,30,20,5,0,0,0,10,2,1,20,95.

On "What is the probability that significant global warming is occurring or will soon occur, and is primarily caused by human actions?", I had to parse several words very carefully, and ended up deciding to read "significant" as "measureable" rather than "consequential". For consequential, I would have given a smaller value.

I answered all the way to the end of the super bonus questions, and cooperated on the prize question.

Comment by chrishibbert on Wait vs Interrupt Culture · 2013-11-30T18:46:30.484Z · score: 1 (1 votes) · LW · GW

In my group at work, it's relatively common to chat "interruptible?" to someone who's sitting right next to you. You can keep working until they're free to take the interrupt, and they don't need to take the interrupt utill they're ready.

In f2f conversations, it's mostly an interrupt culture, but with some conventions about not breaking in in groups larger than 4 or so.

Comment by chrishibbert on Three ways CFAR has changed my view of rationality · 2013-09-14T18:39:30.313Z · score: 7 (7 votes) · LW · GW

I believe that emotions play a big part in thinking clearly, and understanding our emotions would be a helpful step. Would you mind saying more about the time you spend focused on emotions? Are you paying attention to your concrete current or past emotions (i.e. "this is how I'm feeling now", or "this is how I felt when he said X"), or more theoretical discussions "when someone is in fight-or-flight mode, they're more likely to Y than when they're feeling curiosity"?

You also mentioned exercises about exploiting emotional states; would you say more about what CFAR has learned about mindfully getting oneself in particular emotional states?

Comment by chrishibbert on How to Measure Anything · 2013-08-10T17:15:39.235Z · score: 1 (1 votes) · LW · GW

|New information can be gained that increases the expected work remaining despite additional valuable work having been done.

That's progress.

Comment by chrishibbert on The Least Convenient Possible World · 2013-03-03T06:32:36.787Z · score: -2 (1 votes) · LW · GW

When I've argued with people who called themselves utilitarian, they seemed to want to make trade-offs among immediately visible options. I'm not going to try to argue that I have population statistics, or know what the "proper" definition of a utilitarian is. Do you believe that some other terminology or behavior better characterizes those called "utilitarians"?

Comment by chrishibbert on Rationality Quotes October 2012 · 2012-10-06T18:06:16.072Z · score: 0 (0 votes) · LW · GW

Did Munroe add that? It's incorrect. There are lots of situations in which it's reasonable to calculate while throwing away an occasional factor of 2.2.

Comment by chrishibbert on Rationality Quotes October 2012 · 2012-10-06T18:04:06.895Z · score: 2 (6 votes) · LW · GW

downvoted. You're saying you don't know anything about the context provided by a story that is apparently of interest to (at least) several readers here, and you're proud of not sharing the context. Doesn't seem like something to crow about without first finding out if the content is frivolous.

Comment by chrishibbert on Who Wants To Start An Important Startup? · 2012-08-18T21:00:47.330Z · score: 1 (1 votes) · LW · GW

Atul Gawande has a new article on how the medical industry can learn from other businesses that use production methods to achieve consistent results. He mentions a couple of national start-ups that are trying to use consistent evidence-based practices, and continuous review of outcomes to make health care more reliable and consistent and do it at a profit.

Comment by chrishibbert on Be Happier · 2012-04-22T17:38:06.721Z · score: 1 (1 votes) · LW · GW

My significant other keeps a garden, and we have several productive fruit trees that we enjoy getting fruit from. Squirrels take a significant amount of fruit, and cats leave unwelcome surprises in the garden.

We trap squirrels and remove them to county parks. (We don't do anything about the cats.)

Marginally increasing the frequency of squirrels and cats is a negative externality for us. I'm glad you aren't feeding squirrels (or cats) near us.

Comment by chrishibbert on Teachable Rationality Skills · 2011-06-04T18:45:37.422Z · score: 13 (13 votes) · LW · GW

"and the wisdom to know the difference"

Comment by chrishibbert on Teachable Rationality Skills · 2011-06-04T18:37:59.689Z · score: 4 (4 votes) · LW · GW

For many more exercises exploring status behavior (both high and low), see Keith Johnstone's Impro. (Here's my review.) Johnstone's theory of improvisation (and acting in general) is that most of the weight of convincing the audience is carried by relative status distinctions among the actors. He provides a detailed set of exercises for exploring and understanding subtle and extreme differences so actors can be comfortable on stage projecting whatever distinction is called for.

Comment by chrishibbert on The Good News of Situationist Psychology · 2011-04-05T04:00:17.636Z · score: 15 (15 votes) · LW · GW

Without the follow-up report, this is hardly evidence that the theory works. I guess it counts as evidence that the theory is convincing.

Comment by chrishibbert on Bayesians vs. Barbarians · 2011-04-02T17:44:54.098Z · score: 3 (3 votes) · LW · GW

My point wasn't just that I wouldn't make a good torturer. It seems to me that ordinary circumstances don't provide many opportunities for anyone to learn much about torture, (other than from fictional sources). I have little reason to believe that inexperienced torturers would be effective in the time-critical circumstances that seem necessary for any convincing justification of torture. You may believe it, but it's not convincing to me. So it would be hard to ethically produce trained torturers, and there's a dearth of evidence on the effectiveness of inexperienced torturers in the circumstances necessary to justify it.

Given that, I think it's better to take the stance that torture is always unethical. There are conceivable circumstances when it would be the only way to prevent a cataclysm, but they're neither common, nor easy to prepare for.

And I don't think I've said that it would be ethical, just that individuals would sometimes think it was necessary. I think we are all better off if they have to make that choice without any expectation that we will condone their actions. Otherwise, some will argue that it's useful to have a course of training in how to perform torture, which would encourage its use even though we don't have evidence of its usefulness. It seems difficult to produce evidence one way or another on the efficacy of torture without violating the spirit of the Nuremberg Code. I don't see an ethical way to add to the evidence.

You seem to believe that sufficient evidence exists. Can you point to any?

You wanted an explicit answer to your question. My response is that I would be unhappy that I didn't have effective tools for finding out the truth. But my unhappiness doesn't change the facts of the situation. There isn't always something useful that you can do. When I generalize over all the fictional evidence I've been exposed to, it's too likely that my evidence is wrong as to the identity of the suspect, or he doesn't have the info I want, or the bomb can't be disabled anyway. When I try to think of actual circumstances, I don't come up with examples in which time was short and the information produced was useful. I also can't imagine myself personally punching, pistol-whipping, pulling fingernails, waterboarding, etc, nor ordering the experienced torturer (who you want me to imagine is under my command) to do so.

Sorry to disappoint you, but I don't believe the arguments I've heard for effectiveness or morality of torture.

Comment by chrishibbert on Bayesians vs. Barbarians · 2011-03-27T20:40:12.783Z · score: 2 (2 votes) · LW · GW

Maybe my previous answer would have been cleaner if I had said "I don't think I can procure useful information by torturing someone when time is short." It's a relatively easy choice for me, since I doubt that even with proper tools, that I could appropriately gauge the level of pain to the necessary calibration in order to get detailed information in a few minutes or hours.

When I think about other people who might have more experience, it's hard to imagine someone who had repeatedly fallen into the situation where they were the right person to perform the torture so they had enough experience to both make the call, and effectively extract information. Do you want to argue that they could have gotten to that point without violating our sense of morality?

Since my question is "What should the law be?", not "is it ever conceivable that torture could be effective?" I still have to say that the law should forbid torture, and people should expect to be punished if they torture. There may be cases where you or I would agree that in that circumstance it was the necessary thing to do, but I still believe that the system should never condone it.

Comment by chrishibbert on Bayesians vs. Barbarians · 2011-03-27T20:26:20.428Z · score: 0 (0 votes) · LW · GW

I'm not completely convinced that all the people who were punished believed they were not doing what their superiors wanted. I understand that that's the way the adjudication came out, but that's what I would expect from a system that knows how to protect itself. But I'll admit I haven't paid close attention to any of the proceedings.

Is there any good, short, material laying out the evidence that none of the perpetrators heard anything to reinforce the mayhem from their superiors--non-coms etc. included? Your sentence "the people who went to jail went there for violating orders" leaves open the possibility that some of the illegal activity was done by people who thought they were following orders, or at least doing what their superiors wanted.

If you are right, then I'll agree that Abu Ghraib was orthogonal to the main point. But I'm not completely convinced, and it seems likely to me that it looks exactly like a relevant case to the Arab street. Whether or not there were explicit orders from the top of the institution, it looked to have been pervasive enough to have to count as policy at some level.

Comment by chrishibbert on Procedural Knowledge Gaps · 2011-02-13T19:56:08.585Z · score: 1 (1 votes) · LW · GW

I've been investing in stocks (occasionally) and mutual funds (consistently) for about thirty years, and I endorse Vaniver's advice heartily. I think overall, I'm up on stocks, due to doing most of my stock investing in cyclical stocks that I can buy and sell repeatedly over the course of many years. This has worked for me with both SGI and Cypress, which I repeatedly bought at low prices and sold at high prices. If you try this and find that you're not buying low and selling high, then you should stick to mutual funds and a buy-and-hold strategy. I've dabbled in other stocks where I thought I knew something and could time it, but few of those have turned out well. Happily, I knew I was dabbling, and kept the amounts low, so I got a valuable less for a relatively low price.

Mostly, I invest in mutual funds. I have subscribed to a newsletter that specializes in rating No Load funds (there are a couple). This gives me a monthly opportunity to review the performance of the funds I'm invested in, so I can tell when they stop being in the top performers and roll my money over to a different investment.

I record the monthly performance of each of my investments in a spreadsheet (used to be a paper notebook). The newsletter tells me which quintile the performance is in compared to the fund's peers. I highlight 1st and 2nd quintile in green, and 5th quintile in red. When the number of reds gets to be high compared to the greens, I look for a different fund with better recent performance. The commercials always say "past performance is no guarantee of future returns", but it's the only indication you can use. Most of the time performance is consistent over periods of a few years, so you have to look back a year or so when evaluating, and monitor continuing performance in a consistent way.

This whole process takes far more attention than most people are willing to put into it (a few hours a month on an on-going basis, and several hours every six months or so when choosing new investents), and few investors do even as well as the rate of growth of the broad market. That's why investing in the S&P 500 or an even broader market index is a good idea. If you put your money in a broad index and let it sit, you'll do better than 3/4 of investors.

Vanguard is only one decent brokerage. I personally use Schwab, but there are several others with reasonable prices.

Comment by chrishibbert on Frugality and working from finite data · 2010-09-05T18:37:52.467Z · score: 1 (1 votes) · LW · GW

For instance: you can keep getting new data on economics, but there's no way anyone's going to let you do an experiment.

This is somewhat true of macroeconomics, but manifestly untrue of microeconomics. Economists are constantly doing experiments to learn more about how incentives and settings affect behavior. And the results are being applied in the real world, sometimes in environments where alternative hypotheses can be compared.

And even in macroeconomics, work like that explained in Freakonomics shows how people can compare historical data from polities that chose different policies and learn from the different outcomes. So even if no individual scientist will be allowed to conduct a controlled experiment on the macroeconomy, there are enough competing theories that politicians are constantly following different policies, and providing data that sheds light on the consequences of different choices.

Comment by chrishibbert on Frugality and working from finite data · 2010-09-05T18:32:08.198Z · score: 0 (0 votes) · LW · GW

Over the course of your natural lifetime, your past light-cone will extend by about 100 years. Since it already envelopes almost 14 billion years, you won't get much new information relative to what you already know.

You are forgetting the impact of improving science. In fact, most of what we know about the 14 billion year light cone has been added to our knowledge in the last few hundred years due to improved instruments and improved theories. As theories improve, we build better instruments and reinterpret data we collected earlier. As I explained in a recent comment, suggesting new tests for distinguishing between states of the universe is an important part of the progress of science.

You are right about the growth rate of the accessible light cone, but we will continue to improve the amount of information we extract from it over time until our models are perfect.

Comment by chrishibbert on Kevin T. Kelly's Ockham Efficiency Theorem · 2010-08-23T06:29:11.269Z · score: 2 (2 votes) · LW · GW

The game of Science vs. Nature is more complicated than that, and it's the interesting structure that allows scientists to make predictions that are better than "what we've seen so far is everything there is." In particular, the interesting things in both Chemistry and particle Physics is that scientists were able to find regularities in the data (the Periodic Table is one example) that led them to predict missing particles. Once they knew what properties to look for, they usually were able to find them. When a theory predicts particles that aren't revealed when the right apparatus is constructed, we drop the theory.

But in the meantime, you'd have a more interesting game (and closer to Zendo) if Nature gave you a way of classifying objects. In Zendo, there is only one dimension. Something is a match (I forget Zendo's terminology) or it isn't. [In the real world, one of the possible moves is inventing a new test. "If I hold the object up to the light, what do I see?"] Some new tests turn out not to reveal anything useful, while others give you a whole new way of categorizing all the things you thought you knew about before.

In this context, Occam's razor is a rule about inventing rules to explain complex behavior, not rules about how many things there are. Your objective is to explain a pile of evidence, and you get to make up whatever story you like, but in the end, your story has to guide you in predicting something that hasn't happened yet. If you can make better predictions than other scientists, you can have as complicated a rule as you like. If you and they make the same predictions, then the observers get to break the tie by deciding which rule is simpler.

Comment by chrishibbert on Diseased thinking: dissolving questions about disease · 2010-06-05T23:43:20.337Z · score: 0 (0 votes) · LW · GW

I don't believe much in penance. (The dictionary I checked said "self punishment as a sign of repentance". I don't think either aspect is valuable.) It's not related to the question of how we should treat people when they have conditions that are often under voluntary control.

We should convince them that (assuming they agree that it would be better to not have the condition) their best approach is to accept that the condition is at least partially under voluntary control, that control always appears hard, and therefore to change their lifestyle so as to address the problem. If they agree that the condition is a problem, and they find a magic bullet to solve the problem, then no penance is required. If there's no magic bullet, then they can try to change their lifestyle, but there is no need to for them to punish themselves for not understanding the situation before.

Comment by chrishibbert on Ureshiku Naritai · 2010-04-11T17:44:08.937Z · score: 0 (2 votes) · LW · GW

Was it Yoda who said "There is no try, there is only do"? The point is Alicorn's point about making it a top priority. You may have meant to be this positive, but you didn't sound this positive.

Comment by chrishibbert on The ABC's of Luminosity · 2010-03-27T18:55:19.591Z · score: 1 (1 votes) · LW · GW

I love "Codd help you". Brilliant!

Comment by chrishibbert on Let There Be Light · 2010-03-27T18:05:43.025Z · score: 3 (3 votes) · LW · GW

"The Cult of Statistical Significance" suggests that we're looking for tests that display power rather than significance.

Comment by chrishibbert on Undiscriminating Skepticism · 2010-03-16T03:20:39.428Z · score: 0 (0 votes) · LW · GW

the temperature change is not uniform everywhere

But it's non-uniform enough that some people are observing warming and some are observing cooling. So it seems clear from a perspective that accepts the terms of the claim that all purely local observations are uninformative.

second, the effects of such changes on weather may be noticeable in ways other than simple warming (e.g. more extreme weather events).

Tracking extreme weather events from a local perspective seems likely to give even less reliable results than looking for trends in your local climate.

If you accept the terms of the debate, you have to hope for non-biased global observations that are properly normed against a long baseline in order to make any decisions about what weather evidence counts for or against the positions. At this point, I'm having a hard time finding any non-biased observations.

Comment by chrishibbert on Conversation Halters · 2010-02-21T18:07:44.379Z · score: 1 (1 votes) · LW · GW

Isn't this an opportunity to allow them a line of retreat?

Comment by chrishibbert on You're Entitled to Arguments, But Not (That Particular) Proof · 2010-02-15T19:19:16.167Z · score: 3 (3 votes) · LW · GW

Part of my problem with arguing about AGW is that it has gotten to the point that it's not a science question, it's a political question at this point. So I can be reasonably sure that any "scientific evidence" that will be announced will come from one faction or another, and will have been carefully vetted by the policy board to ensure that it hews to the party line. (Whichever party it comes from. All sides are equally to blame as far as I can tell.)

In this kind of environment, it's hard to take any evidence at face value. Both (all) sides accuse the others of double-counting evidence, hiding unflattering data points, and shading results and simulations.

The only thing that makes any sense in this context is to compare historical projections to the world. Some AGW proponents seem to have over-predicted doom, so I heavily discount doom projections. It's not obvious that the worldwide climate is warmer than the (very) long term trend would indicate. There seems to be obfuscation about polar melting. It seems obvious that sea levels rising is miniscule to date. Climate change doesn't make any specific predictions AFAICT that have been upheld.

There are probably other rules of thumb that ought to be useful in this context, but that's all that comes to mind at the moment.

Comment by chrishibbert on Rationality Quotes: February 2010 · 2010-02-03T01:53:13.306Z · score: 4 (10 votes) · LW · GW

If we take in our hand any volume; of divinity or school metaphysics, for instance: let us ask, "Does it contain any abstract reasoning concerning quantity or number?" No. "Does it contain any experimental reasoning concerning matter of fact and existence?" No. Commit it then to the flames: for it can contain nothing but sophistry and illusion. --- David Hume

(quoted in Beyond AI by JoSH Hall)

Comment by chrishibbert on Open Thread: December 2009 · 2010-01-10T21:09:45.915Z · score: 2 (2 votes) · LW · GW

There's also a group of proponents of this style working on Caja at Google, including Mark Miller, the designer of E. And some people at HP.

Actually, all these people talk to one another regularly. They don't have a unified plan or a single goal, but they collaborate with one another frequently. I've left out several other people who are also trying to find ways to push in the same direction. Just enough names and references to give a hint. There are several mailing lists where these issues are discussed. If you're interested, this is probably the one to start with.

Comment by chrishibbert on On the Power of Intelligence and Rationality · 2009-12-31T18:29:04.631Z · score: -1 (1 votes) · LW · GW

You didn't state a point of view. I'm surprised that MatthewB was willing to guess at what side you were taking.

Comment by chrishibbert on You Be the Jury: Survey on a Current Event · 2009-12-12T18:35:54.918Z · score: 0 (0 votes) · LW · GW

I was aware of the case before, but hadn't looked into it in any detail. My reaction to the sites is that the site arguing innocence seems to be presenting facts and showing contradictions in the other side's arguments. I couldn't find any consistent argument on the other side. There were many scenarios, with inconsistent adherence to the facts, lots of innuendo and plausibility arguments for particular claims, but no coherent story.

  1. Your probability estimate that Amanda Knox is guilty. less than 30%
  2. Your probability estimate that Raffaele Sollecito is guilty. less than 30%
  3. Your probability estimate that Rudy Guede is guilty. more than 40%
  4. How much you think your opinion will turn out to coincide with mine. It seems likely. more than 75% overlap, if that means anything.
Comment by chrishibbert on Rationality Quotes November 2009 · 2009-12-05T06:44:16.538Z · score: 5 (5 votes) · LW · GW

I routinely use "a couple" and "a few" to indicate vague quantities. A few is bigger than a couple, but they overlap. I know that not everyone does this (my S.O., in particular, thinks I'm wrong) but I also know that I'm not nearly alone in this habit.

Yes, certainly, there are circumstances in which "a couple" means exactly two. If I'm talking about some friends, and refer to them as "a couple" rather than "a couple of people", you'd be justified to think I meant exactly two people with some relationship. But if I say "I'm going to read a couple more pages", I think you'd be making a mistake to be upset as long as it was between 1.5 and 4 pages. When I say "a few" it might range from 1.7 to 5 or 6 depending on whether we're talking about potatoes or french fries.

So, to my ears, it could be the 16th century or the mid-18th century, and giving the benefit of the doubt, it's a reasonable statement.

Comment by chrishibbert on Less Wrong Q&A with Eliezer Yudkowsky: Ask Your Questions · 2009-11-14T19:39:13.929Z · score: 2 (2 votes) · LW · GW

Do you disagree with Eliezer substantively? If so, can you summarize how much of his arguments you've analyzed, and where you reach different conclusions?

Comment by chrishibbert on Money pumping: the axiomatic approach · 2009-11-07T20:57:53.150Z · score: 0 (0 votes) · LW · GW

Agreed.

I work on prediction markets, so I see it all as bets, and am used to thinking that both participants in a purely financial trade can gain from it, even though many people on the outside of the deal see it as zero sum. Sometimes you increase your variance because you think it's worth increasing your expected return, other times you reduce your variation.

Comment by chrishibbert on Money pumping: the axiomatic approach · 2009-11-06T18:11:38.786Z · score: 3 (3 votes) · LW · GW

Maybe uncertainty makes you nervous, [...]. Then either I'm weakly money pumping you [...], or I'm objectively granting you the removal of your worry as a service. Most people at the time feel that I'm granting them a service, but afterwards they feel I money pumped them. Especially if I repeat it.

Does this mean we should start treating certain types of money pumping as payment for a service rather than something rational agents always avoid?

The name of the service is "insurance". This is a business in which customers repeatedly make bets that they wish they hadn't made in retrospect, but it still makes sense to make the bet ex ante.

Comment by chrishibbert on Your Most Valuable Skill · 2009-09-29T06:43:53.499Z · score: 6 (6 votes) · LW · GW

My most valuable skill I can think of is in the context of being a software developer. I've learned to be pretty good at extracting requirements from customers. I often say this is one of the more important skills a software developer can learn. It's important at all levels of the profession, and is really a gateway skill to performing at a high level.

The reason it's important and hard to learn is that most of the time, customers don't know what they want, or they have an idea in their head, but they're wrong about what will satisfy. Extreme Programming (XP) teaches one approach that lets you get by with less of this skill, and that's to build something simple that approaches what they say they want, and then keep adding stuff until they're happy with it. But if the customer is really confused (which happens more often than you might think) then you'll have started in the wrong direction, and the customer won't get less confused about what they want.

So it helps to know how to talk to the customer and find out what the real problem is, rather than just what they think the solution might be. It's also valuable to have good intuitions about features they're likely to want that they haven't asked for yet, so you can build in the hooks for those. But if you make too many guesses, you'll waste time building general support where it's not needed. So you have to calibrate your guesses as well.

Comment by chrishibbert on How to use SMILE to solve Bayes Nets · 2009-09-20T17:27:52.261Z · score: 1 (1 votes) · LW · GW

(Note: The tutorial 2 code is not correct, but the appendix tutorial 2 code is).

Did you submit a correction to the maintainers?

ETA: Here, too:

Note: The smilehelp document claims that to use the XML format, you need a separate library, but this is no longer correct.

Comment by chrishibbert on Reason as memetic immune disorder · 2009-09-20T17:22:12.694Z · score: 6 (6 votes) · LW · GW

voted up for backing away from the details of the metaphor rather than trying to justify them. Not always an easy choice.

Comment by chrishibbert on Decision theory: Why we need to reduce “could”, “would”, “should” · 2009-09-04T04:20:39.759Z · score: 3 (3 votes) · LW · GW

I think some concreteness might be useful here. When I write code (no pretense at AI here), I often write algorithms that take different actions depending on the circumstances. I can't recall a time when I collected possible steps, evaluated them, and executed the possibility with the highest utility. Instead I, as the programmer, attempt to divide the world into disjoint possibilities, write an evaluation procedure that will distinguish between them (if-then-else, or using OO I ensure that the right kind of object will be acting at the time), and design the code so that it will take a specific action that I expected would make sense for that context when that is the path chosen. There's little of "could" or "should" here.

On the other hand, when I walk into the kitchen thinking thoughts of dessert, I generate possibilities based on my recollection of what's in the fridge and the cupboards or sometimes based on a search of those locations. I then think about which will taste better, which I've had more recently, which is getting old and needs to be used up, and then pick one (without justifying the choice based on the evaluations.) There seems to be lots of CSA going on here, even though it seems like a simple, highly constrained problem area.

When human chess masters play, they retain more could-ness in their evaluations if they consider the possibility of not making the "optimal" move in order to psych out their opponents. I don't know whether the chess-playing automatons consider those possibilities. Without it, you could say they are constrained to make the move that leaves them in the best position according to their evaluation metric. So even though they do explicitly evaluate alternatives, they have a single metric for making a choice. The masters I just described have multiple metrics and a vague approach to combining them, but that's the essence of good game playing.

Bottom line? When I'm considering a big decision, I want to leave more variables open, to simulate more possible worlds and the consequences of my choices. When I'm on well-trodden ground, I hope for an optimized decision procedure that knows what to do and has simple rules that allow it to determine which pre-analyzed direction is the right one. The reason we want AIs to be open in this way is that we're hoping they have the breadth of awareness to tackle problems that they haven't explicitly been programmed for. I don't think you (the programmer) can leave out the could-ness unless you can enumerate the alternative actions and program in the relevant distinctions ahead of time.

Comment by chrishibbert on Rationality Quotes - September 2009 · 2009-09-04T03:28:18.344Z · score: 1 (1 votes) · LW · GW

It has to be "may your grandchildren live in interesting times", or the caster of the curse is as cursed as the recipient. sheesh!

Comment by chrishibbert on Timeless Decision Theory and Meta-Circular Decision Theory · 2009-08-24T15:47:45.116Z · score: 0 (0 votes) · LW · GW

I use OmniGraffle for such things on a Mac. Many people seem happy with the drawing packages in their word processor or presentation program, though. The advantage of an object based editing program is that you can keep arrows connected as you drag things around.

Comment by chrishibbert on Timeless Decision Theory and Meta-Circular Decision Theory · 2009-08-24T05:21:14.284Z · score: 2 (2 votes) · LW · GW

Eli, you are doing an amazing good job of putting Pearl's calculus into a verbal form, but I can't help feeling that this would be clearer if you had a few graphs. Do you have tools that would let you draw the causal diagrams? Why not use them? Is it that the move from Pearl's causal calculus to TDT is hard to express in the graph notation? I still think, in that case, that the causal surgery part of the argument would be clearer in Pearl's notation.

Comment by chrishibbert on How inevitable was modern human civilization - data · 2009-08-24T01:04:03.575Z · score: 4 (4 votes) · LW · GW

There were wheels in the western hemisphere before the arrival of the europeans, but they were only used for toys. No one seemed to guess that they might be useful. But if they'd had more time....

ETA: perhaps I should cite Diamond here in case anyone wonders where this factoid comes from. It's from "Guns, Germs, and Steel".

Comment by chrishibbert on Ingredients of Timeless Decision Theory · 2009-08-20T17:44:54.596Z · score: 2 (2 votes) · LW · GW

okay, you're right they're in there, but Pearl uses those in the proofs, not the explanations, as I recall. I don't think you have to understand the proofs to get the idea.

If you find math oppressive, let me know if you try Pearl and find it too daunting. If that happens, I'll change the way I describe the book, I promise.

Comment by chrishibbert on Ingredients of Timeless Decision Theory · 2009-08-19T16:52:27.157Z · score: 0 (0 votes) · LW · GW

Some kinds of majoritarianism, certainly. The confusion is based on mistaking correlation of votes with commonality of interests. "If we can all agree to vote for proposition X, then it must be in our favor, right?"

Comment by chrishibbert on Ingredients of Timeless Decision Theory · 2009-08-19T16:47:26.928Z · score: 1 (1 votes) · LW · GW

CTDT vs. ETDT. Hmm, that's a tough one. First, CTDT allows "screening off" of causes, which makes a big difference.

I liked EY's formulation above: "TDT doesn't cooperate or defect depending directly on your decision, but cooperates or defects depending on how your decision depends on its decision." It's hard to collect evidence, I think, but reasoning about a causal graph gives you the ability to find out how latent decisions affect other outcomes.

So in this case, expected utility based reasoning leaves you in a posiiton where you make some decisions because they seem correlated with good outcomes, while the causal reasoning lets you sometimes see either that the actions and consequences are disconnected or that the causation runs in the opposite direction to what you desire.

ETA: EY's street crossing example is an example of causation running in the opposite direction.

Comment by chrishibbert on Ingredients of Timeless Decision Theory · 2009-08-19T05:52:07.534Z · score: 3 (3 votes) · LW · GW

This feels right to me. I can't implement it, and I'm not sure I could explain what Eli said, but I understand Pearl well enough (at an intuitive level) to say that it feels like the kind of additions Eli is talking about would clarify and reach the results he's talking about.

Read Pearl. It's not mathy, it's mostly words about graph manipulation.

If you're bothered by math, read Pearl anyway. He doesn't use equations or make you transform symbols. If you can think about information flows or reason visually, Pearl's calculus is for you. You'll understand what it means for something to be a cause or a possible cause or not a possible cause of something else in a deeper way than you do before Pearl.

If you're already comfortable with math, there's nothing hard about the theory, it's just using a different formalism than linear symbols to explain how events are connected causally.

Thanks Eli.

Comment by chrishibbert on Experiential Pica · 2009-08-18T16:30:44.498Z · score: 2 (2 votes) · LW · GW

The "fresh" in California cuisine is about "recently alive", though "novel" is often part of the experience as well.