Solved Problems Repository

Follow-up to: Boring Advice Repository

Many practical problems in instrumental rationality appear to be wide open. Two I've been annoyed by recently are "what should I eat?" and "how should I exercise?" However, some appear to be more or less solved. For example, various mnemonic techniques like memory palaces, along with spaced repetition, seem to more or less solve the problem of memorization.

I would like people to use this thread to post other examples of solved problems in instrumental rationality. I'm pretty sure you all collectively know good examples; there's a comment I can't find from a user who said something like "taking a flattering photograph of yourself is a solved problem," and it's likely that there are other useful examples like this that aren't common knowledge. Err on the side of posting solutions which may not be universal but are still likely to be helpful to many people. 

(This thread is allowed to not be boring! Go wild!) 

270 comments, sorted by
magical algorithm
Highlighting new comments since Today at 1:26 AM
Select new highlight date
Moderation Guidelines: Norm Enforcing - I try to enforce particular rules (see below)expand_more

My only problem with this flow is that it waits far too long before Googling.

My only problem is the fucking annoying habit of software corporations to translate everything, not just the GUI which makes sense, but even stdout error messages. This means not only translating back to English but even guessing the original wording, then googling. The worst part is they seriously think they are helpful here, thinking even IT people who read console error messages not always read English. How the heck are they supposed to solve problems then? I don't think they still think people read documentation, do they?

Yeah, it's a tricky thing. I've actually been involved in translation projects for global software, and the closest I can come to an answer is that they don't really think anything at all... there's several different divisions involved, and each one has bits of the picture, and it all just chunks along without anyone thinking it through end-to-end.

Really, a lot of software development, and of organizational activity more generally, is like that.

All of that said... yeah, the "user googles the error message for instructions" use case is not one that gets taken nearly seriously enough. This is also why you get error messages in dialog boxes that don't support copy-and-paste. If it were, every error message would have a unique copy-able ID code .

It's a nice joke, but I don't think it's actually good advice. There is a lot of background knowledge about how most computer software works that goes into actually executing the steps of this or similar procedures, e.g.

  • knowledge defining the “looks related” relation
  • knowledge about which things are likely to be destructive enough to exclude from “pick one at random”
  • knowledge about what “it worked” consists of when the shortest path to the goal is more than one step

But you acquire that background knowledge faster when you follow the procedure.

My mother is retired, and sits paralyzed in front of the computer not knowing what button to press. I try to explain that you're unlikely to break anything, so just start looking around.

I initially tried giving my mother the "you're unlikely to break anything" advice as well, then reconsidered after she'd followed that advice and gotten malware on the computer.

"Boot into this live-CD and you're unlikely to break anything a reboot won't fix." (At least as long as you don't use webmail or similar persistent online accounts that can get hacked by malware you downloaded into RAM during the same session.)

I learned most of what I learned -by- breaking things.

For example, I learned how page files worked because American Online and Dungeon Keeper both tried to seize them for themselves, and if Dungeon Keeper was run, AOL wouldn't run subsequently without a reboot. Research on the issue turned up that disabling page filing would fix it, which led me to research page filing to see what disabling it would do.

I claim that you have a lot of background knowledge which gives your experimental actions a probability distribution much more like “unlikely to break anything” than hers.

Not really, and particularly not with the new managed computing environments (Android/Ipad) that don't give you root. You can install programs they pre screen, and run them. And she's not likely to install anything I hadn't suggested. Just not a lot to break.

When in doubt, (<-back), (Home), or reboot, in that order.

Is there any kind of widespread problems with a google Nexus getting pwned?

Can relate, the weirdest habit of non-computer-literate people is 1) not reading what is on the screen 2) not trying to interpret even really simple instructions on the screen. Is there any sort of a cognitive explanation why do we have to have conversations like this?

"The computer froze."

"Do you see a pop-up window with a message?"


"Is there anything written into it?"



(squint, lean closer) "Posting Date must not be empty in..."

"What do you think it means?"

"Ugh, fill out the Posting Date?"


Google never fails. The chart shall not allow it.

I didn't think of this until the recent munchkin post, but one solved problem is what financial instruments to invest in. The answer is index funds with low fees in a tax advantaged account. Vanguard has some good funds with fees as low as 0.1%. Unless you're a professional investor (and maybe not even then) your chance of beating index fund performance over the long term is tiny. Not quite winning the lottery tiny, but maybe winning a big stuffed animal at a rigged carnival game tiny.

The technical term for "low fees" is expense ratio.

your chance of beating index fund performance over the long term is tiny.

Isn't it more like 50%?

No; you are incurring extra fees due to your likely extra trading, diseconomies of scale, and uncompensated risk due to less diversification.

How about index funds in countries that have an obviously larger growth potential instead of one's own country.

I don't mean to brag, but I'll bet none of you rich country folk have seen rates we've seen in Brazil in the last many years.

That's the Vanguard Emerging Markets Stock Index Fund and indeed I own some of that. However it's a relatively small part of my portfolio. The risk and variance are much higher, and the risk and variance of a single country index fund would be higher still. This fund dropped more than 75% during the 2008 crash and still hasn't recovered. The expense ratio ranges from about 0.2% to 0.33%, higher though not hugely so than domestic and total international index funds.

Most investors, especially those with smaller portfolios or who have a shorter time horizon, are probably better off with something like the Vanguard Total International Stock Index Fund which includes a developing markets component, but also includes Asia and Europe, or the Vanguard Total World Stock Index which adds the United States to the mix.

At least in the US, generally expense ratios on foreign index funds are higher than on domestic index funds. They're still a better deal than actively managed mutual funds.

one solved problem is what financial instruments to invest in.

That is not even close to true.

To start with, consider the variation in people's goals, expectations, and risk tolerance.

That only changes which index funds to invest in, particularly the mix of stocks vs bonds. Unless your goal is to lose money, the answer is still index funds.

Can you provide an unfolded argument as to why index funds are the right investment vehicles for everyone on this planet?

Index funds outperform all other investments over a reasonably long time horizon.

Even over a short time horizon while certain stocks and funds may outperform some index funds, it is not possible to pick what those stocks are in advance. While you may get lucky with an individual stock pick, you are more likely to get unlucky. The expected return on the index fund is higher.

Index funds outperform all other investments over a reasonably long time horizon.

Really? Surely you will be able to provide data to support this rather amazing claim. It certainly does not look true to me. is not possible

Why is that? How do you know what's possible and what's not?

Many intelligent, rational folks have written about this at great length with lots of charts and numbers. Instead of rehashing the same material, let me just point you at Eric Tyson's Personal Finance for Dummies. as a good starting point.

As to why it's not possible, the short answer is the weak efficient market hypothesis, which has been well proven by 100+ years of experience. I don't personally hold the stronger version--that is, I believe it may be possible to beat the market given insider information--but unless you have such insider information, the expected return of an index fund is greater than any other investment you can make in the same sector.

To bring this back on topic for LessWrong, keep in mind that when evaluating investor performance, you have to avoid survivorship bias. I.e. you need to consider the class of investors as a whole, not merely the ones who got lucky and shout their performance from the rooftops, but also the much larger class of investors who underperformed the market. This alone may be enough to account for the Peter Lynch's and Warren Buffet's of the world.

And of course we also need to account for hindsight bias. I.e. in 2013 investing in Magellan or Berkshire Hathaway circa 1983 is a no-brainer, but how could you pick those in 1983 amongst all the others that looked equally good at the time? Magellan's actually a really interesting case. When I first started paying attention to mutual funds about 15 years, it was the poster child for an actively managed fund that could outperform the indexes. However over the last 10 years or so, it's significantly underperformed the S&P 500. If I had bought it back then instead of the index fund I did buy, I'd have less money today.

Many intelligent, rational folks have written about this at great length with lots of charts and numbers.

Let me remind you what we are talking about. You said "Index funds outperform all other investments over a reasonably long time horizon." I believe this statement to be false. The general handwaving in the direction of "many intelligent, rational folks" does not look convincing.

Perhaps you could expand that statement a bit? What exactly do you call index fund, what are all other investments and what's a reasonably long time horizon? Does the concept of risk enter this claim at all?

let me just point you...

Can we move the discussion a couple of levels higher? I am reasonably well-acquainted with finance and not much interested in explanations for dummies.

the short answer is the weak efficient market hypothesis, which has been well proven by 100+ years of experience.

First, I do not believe EMH, even in the weak form, has been "proven". I also do not think it is correct (at least in a falsifiable form -- you can make it unfalsifiable easily enough by saying that the "long term" isn't here yet).

Second, weak EMH claims that it's not possible to predict future prices solely from past prices. Your claim -- "it is not possible to pick what those stocks are in advance" -- looks like semi-strong EMH and that one is pretty certain to be wrong.

Oh, and I am well aware of both survivorship bias and hindsight bias.

A few months ago I proposed writing a book that would answer all your questions in detail, but there wasn't enough interest to make it worth doing. There are many good books out there already that address these issues. I get that "Dummies" books are too low-status for you. I'll try to dig up some equally well-written, correct references that are higher status and explain things like "what an index fund is?" "what's a reasonably long time horizon?", and "what other investments are possible?" For starters, I've heard good things about Suze Orman's books, though I haven't read them myself.

Your formulation of the weak efficient market hypothesis is incorrect. The weak EMH is not based solely on past prices, but on all publicly available information.

are too low-status for you

No, it's not an issue of status. They are too simplified and do not reflect reality to the degree that I consider necessary. It's like telling a competitive athlete "Oh, physical fitness is solved, just go to the gym".

I don't want references, anyway. You are claiming that investing is solved. You are saying that the only correct investment is into an index fund. Well then, tell me what's an index fund. Is it just any passive diversified investment? There are a great many indices, are all of them equally good? How diversified is diversified enough? Is "long term" a year? five year? ten? fifty? Should commodities be part of a well-diversified portfolio? should currency? real estate? sovereign bonds? volatility?

Your formulation of the weak efficient market hypothesis is incorrect. The weak EMH is not based solely on past prices, but on all publicly available information.

Let's see... Wikipedia: "In weak-form efficiency, future prices cannot be predicted by analyzing prices from the past. ... Share prices exhibit no serial dependencies, meaning that there are no "patterns" to asset prices. This implies that future price movements are determined entirely by information not contained in the price series. Hence, prices must follow a random walk."

Investopedia: "The weak-form EMH implies that the market is efficient, reflecting all market information. ... The semi-strong form EMH implies that the market is efficient, reflecting all publicly available information."

Your wikipedia quote is badly out of context. The article as a whole does not agree with you. A better summary of wikipedia's definition is given right at the start of the article:

There are three major versions of the hypothesis: "weak", "semi-strong", and "strong". The weak-form EMH claims that prices on traded assets (e.g., stocks, bonds, or property) already reflect all past publicly available information. The semi-strong-form EMH claims both that prices reflect all publicly available information and that prices instantly change to reflect new public information. The strong-form EMH additionally claims that prices instantly reflect even hidden or "insider" information.

Your investopedia quote is less out of context, but you still omit a key component of the semistrong hypothesis: that prices adjust quickly. Personally I neither agree nor disagree that prices adjust quickly. I think the wikipedia version is what most people in finance and economics understand as the weak efficient market hypothesis. However if you want to redefine these terms so only the semistrong hypothesis includes all publicly available information, then fine. I would agree with that part of the semistrong hypothesis.

Now on to your specific question. A few stinkers here and there aside, The Dummies Books are incredibly well written and an excellent introduction to most topics. The ones by Eric Tyson on investing and finance are definitely among the best, and I highly recommend them to you if your goal is to increase your wealth and income.

But if you really don't like the Dummies books, many other references are available. I consulted some folks who spend way more time and energy thinking about this than I do, and here are some sources they recommend:

Rick Ferri's columns at Forbes, specifically

Mutual Fund Managers Fall Short Again

Index Fund Portfolios Reign Superior

5 Lies About Index Funds

Bogle's Common Sense on Mutual Funds is "page after page of raw data from Vanguard Research and others. It's an onslaught of irrefutable evidence of the superiority of low-cost index funds"

The Coffeehouse Investor

The Elements of Investing

Passive Investing: The Evidence the Fund Management Industry Would Prefer You Not to See

These answer all your questions and more. I'm not going to write my own 700 page tome rehashing what's readily available elsewhere. You can dispute or ignore the evidence if you like, but you'll just be setting yourself up to be money pumped by hucksters in fancy suits who are ready, willing, and able to take your money.

This is pointless. We are speaking at different levels right past each other.

At least for people with the right personality profile (relatively high openness, above bottom decile of extroversion), CouchSurfing seems to have solved the problem of finding cheap (indeed, free) and comfortable short-term accommodation in a foreign city.

Is the problem of feeling terribly ashamed to accept accomodation without payment or compensation solved or this a test for openness / extroversion I just failed? :) There is nothing I hate more than getting favors and feeling obligated.

It might be that you haven't used CS enough to internalize the ethos governing host-guest relationships. I don't think CS hosts generally frame their decision to host someone as providing a favor to this person; rather, this is something they do because they genuinely enjoy it. Speaking for myself, I only expect my guests to be considerate (make little noise, be clean, etc., and show kindness in our interactions). As long as this minimal expectation is met, I take them to be under no obligation towards me.

I never dared trying it so yes, haven't used it "enough". Are they under no obligation to entertain the host, engaging in chat, small talk, not be morose and sullen, not be a killjoy? That alone would be quite a pressure. As it means they owe it to the host to be enjoyable.

At least for me, getting up when the alarm clock rings used to be almost impossible and I kept staying in bed for 12 hours a day unless there was something that I absolutely had to get up in time for. Anders Sandberg's caffeine pill trick solved that problem for me, and it has worked for over five years now:

This week I have experimented with a new way of getting up in the morning. My problem is that Anders-Sleepy has different goals than Anders-Awake, and is quite adept at resetting the alarm clock. Now I, Anders-Awake, has found a way around this self:

I set my alarm to 6:00 and 8:00. At 6:00 I go up, take a 50mg caffeine pill, and go to bed again. Then I sleep and wake up rested and energetic around 8.

In my case the time for the pill to start working seems to be 1.5 hours. A dose of one pill ensures that I wake up (but still yawning) while two pills makes me start the day much more quickly. The added benefit is of course a regular sleep schedule.

I ran into the problem of a late night one of the days, where I remained awake until 3:30. In this case I adjusted the program slightly, taking the pill at 7:00 and sleeping to 8:30, this seemed to work and the rest of the day was efficient. I became tired earlier in the evening, which was fixed by going to sleep earlier.

Using caffeine to combat sleep inertia is not my idea; a study has shown that it works for naps. Music may also help.

What works for me in addition to my equivalent of the caffeine pill is:

  • An alarm that gradually gets louder. It's not even one of the traditional annoying tones, just kind of a gentle new-agey song. The key features seems to be the getting louder with time.

  • Putting my underwear, clothing, glasses, and the alarm at my desk, away from the bed, so that I have to get up to turn off the alarm and by then it's not that much extra trouble to also get dressed, and I tell myself that if I'm still tired after that I can sit down at my desk, which is still better than rolling back into bed.

For me, the problem of waking up when my alarm goes off has been totally solved by using an alarm (Alarm Clock Plus for Android) which requires me to do complex mental math before it will shut off.

This doesn't solve the related problem of getting out of bed when the alarm goes off; only waking up and not going back to sleep.

one thing that helps me with this is having set up a thing I do semi-automatically whenever I get up. Go to bathroom, check weight, go make coffee usually means I actually get up unless I'm crazy tired and don't actually get as far as coffee making

The problem of transferring large files over the internet has been solved by Dropbox or similar services.

A padded envelope containing a dozen 16GB SD cards, sent by Express Mail (next-day delivery) has a bandwidth of about one megabit-per-second.

Networks have actually gotten fast.

This is a classic demonstration of the difference between bandwith and latency. As in, if you have only 1MB to tansfer, with this method it still takes a day. High-bandwith, low latency, like ship freight (as opposed to low-bandwith, high-latency air freight).

I have been using Dropbox for the past couple of years, but the new BitTorrent Sync seems to be a superior alternative: it is more secure and has no size limits.

"How do I get stronger?" has been solved and the solution is Starting Strength.

Evidence: The set of my friends who are strong is exactly the set of my friends who do / have done Starting Strength or a close variant. Also, I used to lift with several competitive power lifters (including someone ranked top 100 nationally in the deadlift) and they unanimously advocated it.

These are relatively large N and effect size btw, i.e. I know at least 15 people who've done SS and they're out-benching the non-SS'ers by 20 pounds on the low end, and 100 pounds on the high end (I pick bench because it is the exercise most people are familiar with; the gap for other things like squat is more like 50 pounds on low end, 200 pounds on high end).

I wouldn't call SS the end-all, be-all solution for getting stronger, that would more closely be something like "progressive overload using compound exercises (or whatever you want to get stronger at) while under caloric surplus and having decent macro/micronutritional spreads. Also sleeping well and not having any other unusual health problems".

SS is a great program for beginners, but any other program that fits the above should work (like stronglifts). I also wouldn't recommend SS to intermediate or advanced lifters, when linear progression is no longer possible.

This is the impression I've gotten as a beginner to lifting.

My current problem is "how can I learn to do SS safely and cheaply?" My university has an appropriately equipped gym, but I can't really afford coaching.

No need. Lifting weights is not rocket science. Even if it were, you're smart, right? I'm sure your gym has a squat rack & mirrors -- that's all you really need.

Sure. I feel pretty comfortable with squats and benching already. I'm just a bit concerned about learning, say, powercleans without guidance, and going up on weight on them without assurance that I am, in fact, throwing the huge metal object into the air correctly.

There are numerous serious lifting forums on the web that will critique your beginner cleans (for free) out of the goodness their hearts. You just need to film it and upload the video. So do this with lighter weights and see what people say., make the adjustments and ask again. Also, definitely start with cleans--they are a lot safer than snatches and much easier to master.

Oh, that makes much more sense. Actually there are various versions of SS that swap out powercleans for technically easier exercises.

Happen to know any that also lack pullups? I can't do one, and likely won't be able to for a while (I am female). All the powerclean-free ones I've seen replace them with pullups.

(Also, I just found this on Google and am somewhat disturbed. Rippetoe claims you can learn to power clean without a coach, while this guy claims that Rippetoe doesn't even teach power cleans correctly:

On pullups - you can replace them with pull-up negatives. Jump up to the chin-above-bar position and lower yourself as slowly as you can, particularly towards the end of the movement. Let go of the bar and jump again. About the only thing this doesn't train, that the regular pullup will, is the starting-from-a-dead-hang motion.

(This has the added benefit of eventually being able to do pullups.)

I also advise that between jumping pullups, you return to a faux-deadhang position (even if this requires you to bend your knees to fit under the bar). Again, you will get tired of jumping pretty quickly and then will start to rely on your arms and back.

I recommend working up to pullups by using the cable pull-down machine and programming it as 3x5 the same way you do for the main lifts. People typically program pullups for reps because of the inconvenience of adding weight, but it's fundamentally the same as any other movement in terms of the expected effect at different rep ranges. On the cable machine you are limited to lifting at most your own body weight conveniently, but by the time you can do 3x5 at a weight near your body weight you should be able to perform at least one pullup. Also, you should know that at most gyms it's possible to add 5 lb plates to the machine by threading the attachment thing through the plate or by using supplemental plates that you can add to the top of the weight stack, if your gym has them. The cable pull-down is reputed to be slightly easier but I find the difficulty comparable as long as you focus on keeping your torso vertical and pretending you are lifting up to the bar rather than lifting the bar down to your neck (kind of silly but it works). Most people tend to lean back slightly which makes it a little easier and a little less like a pullup.

I program my chinups and pullups this way and I can do them weighted with 62.5lb and 52.5lb respectively, starting from only being able to do 1 or 2 at bodyweight. Also I have a powerlifting total >1100lb from Starting Strength then Wendler's 5/3/1 over the last two years, and highly recommend that track.

That is correct, Rippetoe teaches the powerclean based on olympic lifting technique from the 60's when the rules were different.

Do barbell rows instead of cleans. This won't cause any problems.

based on olympic lifting technique from the 60's when the rules were different.

Interesting; that's a different reason than the one that guy cited. He claimed that a better source for learning powercleans is actually an older one from the '70's by Bill Starr. He seems to think, and I get the impression, that Rippetoe's changes are actually newer, instead of a reversion to an older standard.

Could you elaborate on the differences between what he teaches and... what one should do instead?

The concept of a vertical pull originates from technique when brushing the thigh with the barbell was disqualifying of the lift. This lead to lifters keeping the bar further away from the body, and subsequently for rippetoe to teach the first pull as extremely similar to a deadlift. In contrast, modern technique shows a very distinct S shaped bar path on ascent. Rippetoe is disagreeing with every world record holder in the olympic lifts here, and notably has not trained anyone near close to competitive at even a national level. In contrast, Glenn Pendlay (who has trained the number 1 nationally ranked lifters in several weight divisions) and every other olympic coach teaches them very differently. Clean pulls are NOT deadlift pulls.

I was looking up the Marines' fitness requirements at some point randomly, and for females the pull-up requirement is apparently replaced with a flexed-arm hang (wiki, so you could maybe try doing that.

Flexed-arm hang won't work as a replacement for pull-ups in the context of this particular program; maintaining a load in a static position is very different from moving that load up and down.

As a former Marine, in addition to the difference pointed out by jsteinhardt, the flexed arm hang works your arms only, not your back. Pullups require an overhand grip, whereas the flexed arm hang (and chinup) focus on your biceps.

As someone who has been in this situation, pullup negatives have worked for me. I would also add regular jumping pullups, without the negatives. As you do more, your jumping abilities will decrease and youll start to rely more on your arms/back, thus building those muscles. Most gyms also have an assisted pullup machine.

Additionally, fitness roughly breaks into two broad categories - resistance and cardiovascular. Starting Strength covers resistance training, but the cardiovascular version of Starting Strength is Couch To 5K. It uses the same basic concept of progressive overload applied to running.

SS covers strength training. There is more visual resistance training, body building, which is more about looks than strength, and utilizes about 50% SS like composite exercises, 50% isolation and higher rep counts, 8-10.

This is also a solution to "I am underweight/too skinny, how do I get bigger?"

Is SS for looking good, or for practical strength? I know they correlate, but optimizing for one doesn't necessarily mean optimizing for the other.

You will not be looking good in the 4 months SS takes. Body composition is a long term project. SS and similar programs are to strengthen the substrate so you have something to build on.