Scholarship: how to tell good advice from bad advice?

by ChrisHallquist1 min read29th Jun 201234 comments

18

Scholarship & Learning
Personal Blog

Lukeprog has done some great posts talking about how scholarship has benefited him personally and how to do it. However, he hasn't talked much about how to tell the good from the bad, and I think this is an especially big problem with it comes to "how to"s, rather than standard academic subjects.

The thing is this: writing on academic subjects tends to be dominated by academics seeking to gain status from their work. Not all fields have a strong correlation between status and making actual discoveries, but many (perhaps most) do. So for academic subjects, the big challenge when it comes to making sure you're not reading nonsense is to try to evaluate the soundness of the field in question.

When it comes to seeking out "how to" advice, though, writing on the subjects seems to be dominated by people looking to make money selling advice. That means that if bad advice will sell better than good advice, they'll give the bad advice. Even if their writing includes legitimate advice, they have incentives to give a distorted picture of the subject they're writing on if it will sell better (for example, making whatever they're talking about sound easier than it really is). In some cases, advice books end up trying to nudge you towards buying some more expensive thing the author is selling.

So: when you're researching a "how to" topic, how do you tell the good advice from the bad advice?

34 comments, sorted by Highlighting new comments since Today at 4:40 PM
New Comment

When you receive advice, it must fall into one of these categories:

  1. The advice is worthless or even outright bad.

  2. The advice is valuable, but it's common knowledge, or at least can be obtained from public information in a straightforward way.

  3. The advice is valuable, and it's not common knowledge, nor can it be obtained from public information in any straightforward way.

Now, if you believe you're receiving advice that is in category (3), you must ask yourself what makes you so special that you are privy to this information. This leads to the following heuristics:

  • If the advice is in category (2), it is likely to be good if the source is reputable. For example, a book or website about programming written by a reputable author is likely to give you good advice on how to program. (But note that while it certainly adds value in terms of a convenient and attractive presentation, such a source doesn't give any significant information that wouldn't be available from other public sources.)

  • If the advice purports to be in category (3), and yet the source of information is public (e.g. a book or website, even a non-free one), it's almost certainly bunk. The only exception is if the message is highly unpopular or counter-intuitive, and somehow you know that you have overcome biases that prevent most people from evaluating it correctly, which is very difficult and rare. For example, nearly any book that claims to bring special wisdom about investment, career, relationships, etc. is bunk.

  • If the advice comes from a person on a private occasion, then there are several steps that you need to do. First, does this person show clear indications of the relevant knowledge and competence? If not, it's likely bunk for obvious reasons. Second, is the advice in category (2) or (3)? If it's (2), it's probably good, though it still pays off to check against other sources of information. If it purports to be (3), then you need to do the third, and most difficult evaluation: does this person have the motivation for an extraordinary degree of altruism towards you? If not, it's likely bunk, or otherwise they wouldn't grant you this privilege. If yes, for example if you're getting advice from your parents, then it is probably highly valuable.

One common failure mode is when people believe they're giving you advice of type (3), but in reality, their motivation for altruism towards you is weaker than their motivation for saying things that have high signaling value (and omitting things that have negative such value). This is one danger of socializing with people who are higher-status and more accomplished than you -- you'll be tempted to take their advice seriously, but in reality, even if they are giving it with good intentions, it's likely to be heavily censored and distorted so as to maximize its signaling value.

(This is exacerbated by the fact that good no-nonsense advice on topics that involve any aspects of human social behavior, both personal and professional, tends to sound crass, disreputable, cynical, or worse.)

There's a type of advice I've observed which I'm having trouble categorising into the above: advice which is valuable, but which has a prerequisite level of competence or understanding before you can use it.

I've been swing dancing for about four and a half years, and I've taken a lot of classes and workshops in it. There are several common pieces of advice that get thrown around by teachers: "keep your feet under you", "dance into the floor", "engage your core"... they're generally referring to how something feels when you're doing it, which has a reasonable margin for subjectivity, so some people will hear advice like that and think "oh, yeah, that makes sense", while others will hear it and think "well where else would my feet be? Over me?"

Good instructors find a way to give you the bodily experience without giving you misinterpretable verbal advice. "Engage your core" can mean different things to different people. There's a reasonable amount of crossover between the swing dance community and the circus skills community, and when an aerial acrobat hears "engage your core", they tense up like they're about to be thrown, which is not what's meant in the dance context. But if someone says "imagine you've got a sword stuck point-first in your navel, and you have to keep that sword horizontal", that's quite a specific piece of advice which directly addresses what is meant.

Sometimes you're just not ready for a piece of advice. Earlier this year at a weekend-long event, a well-respected and reputable instructor said something in a workshop. I can't remember his exact wording, but in my notes for the class I wrote "the tension and compression you feel in your hands is a consequence of your bodies moving, not a cause". I thought about this over the rest of the weekend and it completely blew my mind and made me revise huge chunks of how I thought the dance mechanically worked. If I'd heard it two years ago, my response to it would probably have been a lot less productive.

Sometimes you'll be working on something, and you'll notice a physical experience you'd previously not paid attention to, and a piece of advice you heard in a class months or even years ago will suddenly make sense. This doesn't necessarily just happen once with any given piece of advice. The simple and seemingly-obvious imperative of "dance to the music" is one that you can get a lot of repeated use out of at different levels of experience. When you're starting out, it means "just move in time to the music, OK?" A little later on, it means "fit the phrasing of your movements to the structure of the music". A little later still, it means "take inspiration from the features of the music. If the clarinet does something twiddly, maybe you can do something twiddly to complement it in your movements." At the moment, for me, it means something quite ridiculously art-wanky that I'd be mildly embarrassed about sharing, but in another six months I'll probably peel away another layer of "dance to the music", and it'll give me a whole extra take on the words.

I seem to have taken this opportunity to talk at length about swing dancing, (a perennial hazard), but this general idea of advice which sits on prerequisite knowledge or experience is one I see elsewhere. Less-Wrong-flavoured advice like "your strength as a rationalist is to be more confused by fiction than by reality" feels a lot like "keep your feet under you". "Go meta", much like "dance to the music", has different levels of subtletly and meaning. Less Wrong buzzword expressions sometimes feel like thing people are throwing around because they're available to do so, but when someone uses one to get to an important central point of an issue, when they nail it, it's genuinely illuminatory.

This category of advice is usually what I run in to when I'm evaluating things at all (people rarely trust me with secret wisdom, but they often trust me with merely advanced wisdom :))

As a metric, I'd suggest a variant on #3 works: is there some reason to suspect that YOU have a special level of competence that grants you unusual insight? And then ask equally why is it being shared? Is it because it's useful at all levels, or because your instructor trusts that YOU are clearly an advanced student, who can understand these things?

Or is it simply because it's a high-status platitude that will encourage people to start thinking for themselves, then credit the platitude for their success? Or perhaps it simply serves to keep you practicing, and practice tends to bring improvements! :)

Well, normally at dance workshops the advice is being shared because I've paid a not-inconsiderable sum of money for the privilege of being there, and have auditioned to make sure I'm in a group of dancers at a similar level to me :-)

Exercise for the reader: Apply this advise to itself.

Looks like a straightforward category-2 (logical extrapolation from publicly available information) with the added value of good presentation and doing the extrapolation work for us.

Hrm. There's a category 2.5 of "advice which is common knowledge to long-standing members of a small subculture but not to the public and that isn't written down." These are cases where the information isn't secret, but is primarily conveyed word-of-mouth, not in print, and where it's too specialized to be "conventional wisdom".

It might be that there's some piece of advice that most expert, say, patent lawyers would give you, but that wouldn't be in a standard book because the topic is too narrow or esoteric.

In my case, I've gotten valuable career advice from senior members of my profession. I don't think it was a unique boon to me and their advice parallels conventional wisdom, but with details that are specific to my narrow field.

I'm not sure exactly what you mean by a lot of this ("common knowledge," "public," etc.) Some of these things, I suspect, will be a matter of degree.

Furthermore, legitimate information may be presented in a hyped-up manner for marketing purposes. For example, from what I can tell, Tim Ferriss' books do contain a substantial amount of accurate information, though often presented in a hyped-up manner that makes the things he talk about sound easier to do than they really are. So I think "almost certainly" is too strong for your second bullet point.

Does Tim Ferriss ever talk about finding Ferriss-style opportunities for yourself in addition to using what he's discovered?

Yes. In the Four Hour there the general advice of testing seeking out ideas that might produce big wins and testing them the Quantified Self way.

This is true, but a simplification. Specifically, it doesn't distinguish cases where good advice is mostly impersonal (e.g. how to invest money) from cases where the best advice will be highly personalized (e.g. diet).

In many complicated fields, like diet, good advice needs to be individual. Learning enough about the field to choose the right advice yourself may take years. And it's prohibitively expensive to find what works best for you by trying everything. At best, you'll stick with the first thing that works moderately well.

So I propose category (4): the advice is valuable, not because it relies on nonpublic information, but because matching the right advice to each person is complicated (though based on public info, such as medicine). People who study the field, master it, and then give personalized advice add real value. Most importantly, to trust the advice of such people, you don't need to assume an extraordinary degree of altruism on their part. Ordinary situations like paying an expert for counseling may be sufficiently trustworthy.

Learning enough about the field to choose the right advice yourself may take years.

In that case, how could the expert possibly know enough about the field to choose the right advise for someone they only know through at-best several hour long appointments?

That's a good point I hadn't thought of. Many fields probably won't be like what I described: one would need to know a lot both about the field and about the person who needs advice, to give personalized advice.

Still, I think in most fields good personalized advice requires many years of studying and working in the field, while a few weeks of studying the person who needs advice would be sufficient. There is a disparity, partially (wholly?) arising from the fact the expert is already experienced in the field when they start working with the client, and has also studied how to analyze clients' requirements.

Of course, like you say, professionals that most people are able to hire can only give them a few hours of their time at most; often much less, like the 10-30 minutes of a typical doctor's visit.

The best single rule I have found is useful for academic disciplines as well: "Everybody overstates the accuracy, applicability, universality, precision, or other some other aspects of claims they have made." Anybody who actually cares enough about something to go through the trouble of writing it up, unless they are being paid by the word (which as you note has its own problems), is going to be emotionally invested in their work and writings. So my best advice is to assume from the beginning that it is overstated, and consider a weaker version, or a charitable interpretation, of their claims.

Anybody who actually cares enough about something to go through the trouble of writing it up ... is going to be emotionally invested in their work and writings. So my best advice is to assume from the beginning that it is overstated, and consider a weaker version, or a charitable interpretation, of their claims.

As someone who cares a lot about accuracy, I feel that you are overstating this point. ("Most people" in place of "Anybody" seems more plausible.)

I read that as "take as a starting point, the assumption that the idea is overstated" - it's simply a base assumption, if you lack any further information. If you know you're dealing with someone who highly values accuracy, then you can rely on that information. But it's not a good default assumption to assume people value accuracy highly (except possibly in certain highly-specialized sub-cultures)

[-][anonymous]9y 5

when you're researching a "how to" topic, how do you tell the good advice from the bad advice?

Before advice is tested, it may not be possible to know if the advice is good or bad. So I look for qualities that make advice testable.

Bad advice will cost more when it fails than pay when it succeeds. Bad advice will explain too much and thus be hard to measure if it's working or not ('prayer works, so if I just pray a little harder... '). Bad advice will involve people who aren't consenting or informed about what I'm doing. Bad advice is often a package deal ('if you're a feminist, you have to be a vegan'). Bad advice is unaccountable ('no, it's different when I do it' / 'I was only following orders').

Some of what is left over from that winnowing process will be good advice.

"I conceive of nothing, in religion, science, or philosophy, that is more than the proper thing to wear, for a while." - Charles Fort, 'Lo!' (1932)

So: when you're researching a "how to" topic, how do you tell the good advice from the bad advice?

Pretty much the same as in the sound academic fields, differing only in the amount of woo, bullshit, and self-interest you'll have to assess. Compare sources, look at results, look at reviews, look at the evidence, ask whether it even makes sense, try things out yourself, etc.

In the area of self-help (as some comedian has remarked, the idea of getting self-help from a book is already something of a contradiction) you also have to contend with the basic therapeutic conflict: anyone selling a cure for a disease has given themselves an interest in never curing it, but providing only palliatives. This is especially true when the disease is an imaginary one and their first task is to persuade you that you have it, which is why so many self-help books begin by telling the reader what a schmuck he or she is.

Here is an extreme example of how not to do it. Conversation I once overheard between a middle-aged man and woman:

The woman: talks about how convenient she is finding her new microwave oven.

The man: "Oh, I've heard you shouldn't use those, they destroy the vibrations in the food."

The woman: "Oh dear, I'd better stop using it then."

Ok, the man's talking woo, but that's not the point here. Substituting a topic more easily able to pass the rationalist filters:

The woman: talks about how wonderful she finds her new bread-maker.

The man: "Oh, I've heard you shouldn't eat bread, it wasn't present in the ancestral evolutionary environment."

The woman: "Oh dear, I'd better stop using it then."

Whatever the facts about eating grains, the epistemic process of transmission here is completely corrupt. So, don't just read a book and nod along to the author's argument, thinking "How true this is, I must do this."

[-][anonymous]6y 2

as some comedian has remarked, the idea of getting self-help from a book is already something of a contradiction

I agree but not sure if for the same reasons. I think most of the time people know perfectly well what they should do differently, they just lack the willpower or motivation for it. A book here may inspire for a short while, if it is really well worded it can "pump" people for a while, but it will not last long. In the vast majority of the cases, people buy self-help book, read actual good advice in (generic good, not big insights, mostly the "get your sh.t together" type of good), nod, nod, and then do nothing.

Coaches, trainers probably work better. So do groups. I think the core idea of AA is that every meeting gives a jolt of motivation enough to last until the next meeting.

For starters, I'd say it would be best to take advice from people whose careers and accomplishments are to some extent a matter of public record. Then you can evaluate (a) whether they seem to have actually accomplished the things they're trying to teach you to accomplish, and (b) whether they seem to have accomplished those things via the procedure they're encouraging you to follow. If yes to both, then you might proceed further.

In that case, the problem of making good advice seem too easy might come down to a couple of things. First, you want to see a good step-by-step procedure where you can really understand each step and imagine exactly what you'd have to do to achieve it. Second, it would be a red flag if any of those steps seem to be "magic" steps such as "Have a brilliant, lucrative idea for a business."

The expert-at vs. expert-on distinction severely weakens this meta-advice. See also Unteachable Excellence.

For starters, I'd say it would be best to take advice from people whose careers and accomplishments are to some extent a matter of public record. Then you can evaluate (a) whether they seem to have actually accomplished the things they're trying to teach you to accomplish, and (b) whether they seem to have accomplished those things via the procedure they're encouraging you to follow. If yes to both, then you might proceed further.

If you are privy to the information necessary to evaluate (b), you can just look at it directly and skip listening to the advice altogether.

Yeah, but it might be useful to know what the person in question considers to have been the crucial aspects of their procedure, as opposed to merely ancillary aspects. This won't be failproof but will at least have better than chance odds of contributing something useful to the advice.

This won't be failproof but will at least have better than chance odds of contributing something useful to the advice.

That depends on whether this person is motivated by a real desire to benefit you, or a desire to sound in a way that has maximum signaling value. (Note that the latter can be the case even when people honestly believe they're doing the former, unless they have a special and extraordinary degree of altruism towards you, which is typically the case only for close family and friends.)

Advice about what to do if you have a brilliant idea for a business is still valuable for some people.

So: when you're researching a "how to" topic, how do you tell the good advice from the bad advice?

Find many credible sources of advice - preferably a combination of raw scientific studies and the applied practical advice of domain experts. Compare and contrast. Then try what remains and discard what doesn't work for you after a sufficient implementation period.

  1. Start with minimal knowledge. Everything you read seems plausible.
  2. Try stuff, keep track of your results. Form hypotheses, and seek new information being slightly more skeptical and discerning than before.
  3. Repeat 2 until 99% of what you read is crap.
  4. Follow the 1% advice, but you already know what you're doing by that point, anyway. Diminishing returns.

It's really hard to tell good from bad when you don't have the domain knowledge. You have to acquire a little bit by luck/accident, and then you start iterating.

This worked for me re: diet, meditation protocol, exercise protocol, time management, goal management, existential suffering, and much more. (Some are pretty steady state, others are still changing, and the steady state stuff may change if it stops working--which means my knowledge is incomplete, and more iteration is required.)

(A remark on how what you wrote sounds, even though the intended interpretation might've been entirely different.)

It's really hard to tell good from bad when you don't have the domain knowledge. You have to acquire a little bit by luck/accident, and then you start iterating.

This reads like a recipe for locking-in into your own randomly generated dogma. Personal anecdotes processed by bad understanding of statistics don't generally bring enlightenment.

This reads like a recipe for locking-in into your own randomly generated dogma.

I'm coming back to this years later, but, wait, what? How does one learn, then? How does one separate true stuff from false stuff without engaging with it, without wrestling with it, without trying to disprove it, without applying it?

When you don't know, you don't know what you don't know. How can you know, except by doing something intelligently accidental? Even if it's just doing the exercises at the end of the chapter?

You move in small reliable steps, making yourself stronger, so that eventually you can take bigger steps that you couldn't judge reliable before.

Ok. That's compatible with what I meant, even if it's not what I said. :)

I don't see that this method would necessarily result in you getting stuck dogmatically. If you refuse to acknowledge that your initial success was merely luck/accident, then it would seem easier to think "well, my method is perfect." If you realize it's a random dogma and that you could have gotten similar results from a dozen other dogmas, it encourages you to experiment and start narrowing in on the dogmas with the greatest returns.

It's also, in my experience, VERY difficult to measure success until you've had a taste of it. I was shocked the first time I did a successful exercise routine - I'd never known it could be so easy and enjoyable. Now I can quickly discard any dogma that says I'm supposed to suffer, at least until I hit a wall with my easy and enjoyable methods :)

While it isn't always possible to obtain, advice that makes specific predictions is typically good. For instance, for fitness I've tried both Starting Strength (for exercise) and Lean Gains (for diet). They both made strong predictions about what would happen to you after doing their program for one to two months (Starting Strength: get substantially stronger; Lean Gains: lose a substantial amount of weight). For me personally, the Starting Strength predictions were accurate, but the Lean Gains one were not. So, I continue to do Starting Strength, but not Lean Gains.

In general, even if strong predictions are not made for you, you can translate advice into implicitly making such strong predictions. (This is, unfortunately, harder for things like planning for the future, when waiting to see if the predictions pan out is very expensive.)

I would go seriously mainstream (as in mainstream scientists) for such advice. Accomplished mainstream scientists. The advice will be boring: enroll in a course, maybe online course, do not skip homework, do not skip problems, do not skip test. The kind of stuff you probably already know.

If you are looking at some much easier way to become awesome competent, I am afraid you're asking for bad advice.

The Luke sure thinks that what he did was scholarship and that it benefitted him, but you should look for people who did objectively benefit from their studies, in a way not involving some sort of selling of ideas to the uninformed public. You need to somehow exclude as source of the advice the people who just babble of some advanced subjects they have incredibly poor understanding of, unaware of own incompetence.