Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

This reminds me of Goodhart's law:

"When a measure becomes a target, it ceases to be a good measure."

https://en.wikipedia.org/wiki/Goodhart%27s_law



I don't think this is a textbox example of Goodhart's law.

A healthy amount of sleep is a great target and measure, just like staying within a healthy weight range and lots of other biometrics.

I think Goodhart's law applies to people gaming the system or juking the stats to inflate a measured number causing the number to lose meaning. It doesn't matter what you do, getting enough sleep is important.


I see your point, but this ignores anxiety, sleep quality and training yourself to answer your body signals.

First, targeting a metric may create a counter productive stress that can make you sleep less.

Second, the device, and the process, may induce a sleep of lesser quality.

Last, always targeting numbers doesn't force you to listen to your body, an ability that our society is losing more and more. You don't need "8h of sleep". You need to sleep at the time and duration that your feelings indicate to you. Failing to do that result in plenty of problems, including oversleeping, sleeping at bad hours, neglecting diet related to sleep, ignoring light, not adapting to punctual overloads, etc.

Now, yes, Goodhart's law is more about cheating a system. But it's kinda what you do: you try to bypass your natural way of calibrating with sleep and only see a metric. We do that for a lot of things: food, sport, work, etc. And because of that, we behave erratically.


In a good example of Goodhart's law, there's a measure that tells you something useful, but it becomes less useful when you set a target. The three points you make argue almost as much against measuring at all as they do against setting a target.

Also, your arguments center on a metric being used as a substitute for making positive changes, rather than being used together alongside positive changes. This can happen with any measure/target. For example, if you set a target of 0 cigarettes smoked per day, sure, it's possible to just stress about the target and end up smoking more. That doesn't mean the target is bad, it just means it needs to be accompanied by action to make it useful.


You need to sleep at the time and duration that your feelings indicate to you

It's oversimplification. Those signals are dependent on light, among other things. And unless you live with no electricity they may fool you.


> And unless you live with no electricity they may fool you.

I live in Scotland - during the winter we get 8-ish hours of daylight, during the summer we get more like 17,(to the point that it doesn't actually get properly dark at night). Routine is far more important here than natural light.


Absolutely, but establishing routine means overwriting your natural signals with acquired habit, not following them.


Yeah, sleep is actually a good fundamental goal in itself, not a gameable proxy for something else.


How so? Can't think of a thing that's more a means to an end than sleep - you have to sleep to be able to live your life! Barring those who can call up lucid dreams at will[0], there's zero point in optimizing for sleep as a terminal goal. I mean, if more sleep was a good fundamental goal, then suicide would be an optimal strategy - we get all the sleep in the world when we're dead.

--

[0] - somehow they exist and they don't use it as a personal holodeck to escape reality, which tells me lucid dreaming must be overrated.


-- lucid dreaming

I've had over 100 lucid dreams, although none recently. My crowning achievement was summoning Socrates that would teach me things using the socratic method (only asking questions). The thing that blows my mind today is I was able to take away real value from these conversations even though I supplied both the questions and the answers.

Why didn't I use it as a personal holodeck to escape reality? Reality is a bigger, more interesting, and more exciting place than inside my own head.


> The thing that blows my mind today is I was able to take away real value from these conversations even though I supplied both the questions and the answers.

Well it worked for Socrates in that way too - he just happeed to be awake :)


> The thing that blows my mind today is I was able to take away real value from these conversations even though I supplied both the questions and the answers.

Not surprising - whenever we think, we supply both inputs and outputs ourselves :).

If I had that level of vividness of dreams and I could make them lucid, I'd definitely want to sleep more than I do now, to push experiments such as yours to sleep time. I mean, from what I read about lucid dreams, in best cases you have enough control over it that you could essentially use it as a sandbox prototyping environment!


> If I had that level of vividness of dreams and I could make them lucid

Lucid dreaming is actually a skill anyone can learn.

Edit: Now that I think about it, the population is probably less than anyone. Anyone that can get a goods night rest can learn how to lucid dream, and anyone that routinely enters REM sleep has a decent shot at it.


I've been able to lucid dream on a few occasions. What I really want is the ability to vividly imagine constructions while awake. Like design an A/C motor in my head without paper. Not to that degree but those fuzzy concepts that you loosely feel the connections of, I'd like to be able to 'draw it out' in my mind and use those visuals to work out other aspects.


Yup, me too. Apparently the capability to form vivid mental images is not a human universal. See this informal poll: https://twitter.com/backus/status/1091203973246111744?lang=e.... Unfortunately, I fall into the 27% that choose (1), i.e. they see nothing.

I really, really, really wish vividness of mental images would be trainable. I so want it.


I find it hard to believe that 39% see a red star or that 57% see a pink or red star. After minutes of trying, I can vaguely fool myself into thinking I can see (2) a faint black on black outline of a star as if you traced it out the way you'd draw it with crossing lines--I was mentally tracing out that pattern repeatedly.

Color is rare. I mostly don't even notice in most dreams except a very few where it's central, like the one time I saw a colorful dreamcatcher-like crystal mobile-thing after waking from a dream and enjoying still having it from my dream, only to wake up and realize that they were both dreams. Doh!

I keep believing I can learn to visualize based on the fact that seeing is hallucinating. We take samples of light and imagine that we're actually seeing the world when really we're guessing that's what it is and synthesizing what we see. A great example of this was when I came back from a tropical vacation and back in the city a leaf blew past my feet. I swear I clearly saw a small lizard (like I'd testify that I saw a glimpse of one) as I had been seeing for the past few weeks. I saw the leaf on the second glance. So if I can conjure a lizard, I can draw some cylinders and wires eventually.


Most of the time I can't visualize either while awake but sometimes I can and my dreaming is almost always in color (I've had dreams where part of it turned grayscale and that was quite distinctive). I also over decades of sleep issues noticed that I often visualize unintentionally right before I fall asleep and sometimes attempting to do so can help me get to sleep. Some people seem to think that NMDA receptors might be involved somehow. I think in my case images do sometimes appear very briefly even when I can't visualize. On the flip side, I don't think even people who visualize well are usually able to hold a detailed image in mind for a long time and examine different parts of it individually (however I could be wrong about this, it is just my impression; I think people can usually recall the image at least a few times). It might be that my brain is just prioritizing a different representation.

My internal audio reproduction seems to be quite good (which would be great if I could remember more music, or at least whole songs, to play on my internal jukebox :/ - sometimes I do remember more when dreaming). I think recall/visualization/etc. is affected by limited working memory capacity and how working and long term memory interact; I suspect that is often the major limitation when awake also and would still apply to some extent when asleep (although things can clearly work somewhat differently when asleep).

I do believe that a majority of people can visualize fairly easily since it seems like most people can describe the appearance of someone they just met when not looking at them, which I am unable to do (most days). Yes there are non visual tricks to do that, but it does seem like many people do that type of thing effortlessly and also for things they are less likely to have previously learned tricks about. When I do know someone better I seem to be better able to recognize their distinctive style of movement than most people.

Dreaming is strange... lately I haven't been wanting to be aware of my dreams at all since I often end up with events looping while I am increasingly paralyzed (and of course everyone else want me to do something). I've noticed this seems more likely to happen on days I take baclofen (as a help stay asleep sleep aid), but it happens at other times also and happened occasionaly before I ever took baclofen.


It is a good fundamental goal not because it itself is the end but because there is ample evidence that a very large number of quality of life measures have a dependency on sleep.


You'd have those QoL measures as fundamental goals then, not sleep.


Sleep is a fundamental need for the human body, on par with food and (ok, maybe lower than) oxygen.

Maybe you're blessed with effortlessly getting enough of it, which in case congratulations. A lot of the people around you are not.

One thing I'm not saying is you should maximize sleep. Just get enough, and get good quality. Maybe that's our misunderstanding.


I think Goodhart's law applies to people gaming the system or juking the stats to inflate a measured number

Indeed, it is. From TFA: "In the case study on orthosomnia, researchers found that patients had been spending excessive time in bed to try to increase their sleep numbers, which may have made their insomnia worse." (emphasis mine)


Your argument would make sense if sleep processes were being measured via EEG or similar means, but virtually all "sleep trackers" are just movement trackers; they're absolutely measuring a proxy that can be gamed.


> virtually all "sleep trackers" are just movement trackers

This has been false since March 2016 when Fitbit introduced sleep stage analysis based on heart rate variability. I work for Fitbit but don't speak for Fitbit. Check my comment history for a link to a help article with details.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: