[westoncb]

> Throughout the book, the author cites a scientific study and article on nearly every page, so you know that these are real hard facts.

This is almost discouraging to me. If spending too much time on HN has taught me anything, it's that citing a study (especially medical, nutrition, and psychology papers, among other fields I'm probably missing), is not nearly enough of a guarantee that some assertion should be called a 'hard fact'.

I tend to depend more on the force of overall coherence in an argument, as well as its fidelity with my own experiences. Of course this is a problematic heuristic in its own right, since it by definition limits your ability to believe counter-intuitive but true results—but I'm not sure what's better.

I think with enough reproductions of results in varied contexts, unintentional reappearance of results in overlapping or unrelated studies, applications being developed on top of results (treatments, engineering applications, other fields of study)—then you can start considering the result to be 'factual'.

But from the above description, the impression I get of the book is that it's probably one of these taking a bunch of new studies which can be read a certain way to push a hypothesis the author is in favor of, but it'll take a decade or more before we have a good sense of whether the whole thing was BS (or at least badly exaggerated) or not.

[b_b]

I definitely get where you are coming from, given the unfortunately common practice of misconstruing scientific studies to make exclamatory headlines.

I should add then, that the author definitely backs up the studies (which are multiple and often independently discovered) with reasoning about why certain things are evolutionary advantageous and how they fit in to societal phenomenons. I hope you will give this book a chance :)

[westoncb]

Thanks b_b, that does increase my interest and I'll probably look a bit closer now :)

[guiambros]

disclaimer: I've commented so many times on this book before [1][2][3][4] that I may sound like the author's PR person (spoiler: I am not; just truly enjoyed the book).

I second parent's comment that for me this was one of those perspective-altering books in recent years. Even now, months after I finished, I still have several thoughts going through my head: why schools start so early, the societal impact of sleep deprivation, why doctors complete ignore the importance of sleep for your health [5], what happens when you drink alcohol (particularly at night), how Alzheimer's works, lucid dreams, and more.

I had already read about most of it in piecemeal over the years, but having everything pieced together for me, in a cohesive and fluid narrative, really helped for things to "click" for me.

Of course, if you want just the TL;DR, the gist is "Yeah, you really should sleep 7 to 8h per night, no joke". But if you want to understand the WHY, then it's a great read.

[1] https://news.ycombinator.com/item?id=17381548

[2] https://news.ycombinator.com/item?id=17446932

[3] https://news.ycombinator.com/item?id=17520658

[4] https://news.ycombinator.com/item?id=17589509

[5] https://news.ycombinator.com/item?id=17531171

[Ntrails]

> why schools start so early

Because they need to fit in around the schedules of working parents. I'm loathe to accept that this is a source of problems given the likely issue is actually the time students go to sleep, how long they sleep for, and what time they wake up.

The solution does not obviously need to be changing school start times rather than student behaviour

(seriously though, I'd love to hear the counter, as there probably is one)

[KevanM]

I've just started reading this, but the counter is that at the age the students are their bodies start to produce chemicals that mean their sleep pattern tends towards late nights (evolutionary advantageous as young fitter ones of the tribe are active at the most dangerous time of day) and consequently are sleepier if forced to wake at a time where their body is still chemically calling for sleep.

[daveed]

If the topic of the book isn't something you know well, how well do you think you can judge the coherence of the argument? Is coherence measured in some presented mechanistic model? Is it measured against your prior conception of how the body works (gained from school or other historical reading)? What if we were wrong before?

I guess I'm asking this because I'd believe that this attitude is not good when heeded by the general population. My belief there is that in that population, the better essayist wins over the average expert.

Probably cosigning by other experts would help. As well as links and discussions to opposing studies.

[westoncb]

> If the topic of the book isn't something you know well, how well do you think you can judge the coherence of the argument?

Hopefully by the time you have read the book, you have picked up a set of ideas from the field under the discussion and have an understanding of how they relate to one another. So I'm referring to coherence within that set of ideas and relations (which I'll refer to as a 'framework'). To be more specific I'd define 'coherence' as mutual support and lack of contradiction ('mutual support' is like... one point gains strength because of the content of other points.

The probability that the framework is strongly coherent while also in reality being false decreases with the size of the framework. In other words, if the author has made a large number of claims, which are all coherent with one another, the probability that they are true (in the sense of corresponding to something in objective reality) is relatively high. The alternative is that the author went through enormous efforts to create a large, coherent framework which doesn't correspond to something in objective reality. It's just not easy to do that (requires lots of imagination and lots of attention to detail), and it's fairly easy to detect. So either the framework matches something in reality, or the author is a very imaginative and meticulous con artist.

I think the main risk is that the authors trick themselves too, not that it's a fully intentional scam. So it seems like a decent heuristic.

[psyklic]

I prefer references so I have more information available to assess the validity myself. If an author makes assertions without evidence to back them up, then they are much less credible.

Having many recent studies available to back up a hypothesis at least informs us that it is credible, given our knowledge at the time.

[westoncb]

Sure, all things being equal, it's better to have the references than not.

That said, as a practical matter, if the author cites 100 papers, and each paper takes half an hour to judge the reliability of it (assuming you have good command of statistics and scientific methodology, and some domain knowledge, and access the the required journals), we're talking on the order of 50 hours to validate the argument (of course this is a crazy rough estimate)—who is going to do that? You may as well just take the list of papers, try and validate them, and skip reading the book (and this is largely ignoring the fact that most likely your ability to validate is probably not that great in the first place).

Instead, for ~99.9% of readers, the argument presented will end up working solely as an authority argument.

Not sure what the solution is aside from waiting longer, then coming with maybe a handful of extremely reliable authoritative results that the reader can reasonably verify on their own. I'm not sure. I just know reading a book grounded in 'hard facts' because a study is cited on every page is not at all appealing to me.

[retsibsi]

There is a middle ground: spend a few hours verifying a small sample of the citations. This still takes more time, effort, and ability than most readers will have to spare, but it is feasible and would be enough to expose the worst offenders. Of course it can't guarantee that every study cited in the book is reliable, but it can give you a sense of how accurately the author characterises the studies and how carefully (s)he has chosen them. And if you choose your sample carefully, you can directly verify the claims that you care most about.

(Even better if this is done collaboratively, for a larger sample and less time spent per person.)

> for ~99.9% of readers, the argument presented will end up working solely as an authority argument

Agreed, but an author's authority can still be undermined or reinforced by expert opinion. In an ideal world where book reviews were written by qualified, careful people with time to do a thorough job, the existence of citations would be extremely important, even if hardly any ordinary readers looked at them. Obviously we don't live in that world, but sloppy scholarship is often exposed eventually.

[westoncb]

That's definitely a better, more realistic process than what I described.

> Agreed, but an author's authority can still be undermined or reinforced by expert opinion.

That's a good point. Actually, that's probably how I will try to validate this book (or others like it): I'll give the 'okay', once I see some unassociated other researchers confirming that the work is good, conclusions drawn are not a stretch etc.