Open questions from Gwern

This is all Gwern, I won’t add another layer of indentation:

Open Questions

Some questions which are not necessarily important, but do puzzle me or where I find standard answers to be unsatisfying (along the lines of Patrick Collison’s list & Alex Guzey; see also my list of project ideas):

  • What is personal productivity and why does it vary from day to day so strikingly, and yet not correlate with environmental variables like weather or sleep quality nor appear as the usual kind of latent variable in my factor analyses? Is it something much weirder than the usual kind of latent variable, like a set of zero-sum measurements drawing on a generic pool of energy or mana?
  • Does listening to music while working serve as a distraction, or motivation?
  • What, algorithmicly, are mathematicians doing when they do math which explains how their proofs can usually be wrong but their results usually right?Is it equivalent to a kind of tree search like MCTS or something else? They wouldn’t seem to be doing a literal tree search because then there would almost never be mistakes in the proof (as the built-up tree of theorems only explores valid inferential steps), but if they’re not, then how are they handling logical uncertainty? Are they doing something like MCTS’s random playouts where lemmas are not proven but simply heuristically given a truth value to shortcut exploration and the heuristic is accurate enough to usually guess correctly and this is why the proofs are wrong but the results are right?
  • Why did Jean Calment live so many more years than other centenarians, breaking all records and setting a life expectancy record which decades later has not just not been broken, but not even approached? Which is extraordinary considering that she smoked, medicine has continuously advanced, the global population has increased, life expectancy in general has increased, and the Gompertz curve implies that, with mortality rates approaching 50%, centenarians should die like flies and ever closer in age to each other and not have occasional enormous permanent 3 year gaps between the record setter (Calment) and everyone since then.
  • Why do humans, pets, and even lab animals of many species kept in controlled lab conditions on standardized diets appear to be increasingly obese over the 20th century? What could possibly explain all of them simultaneously becoming obese?
  • What happened to the famous genome sequencing cost curve after late 2012, which stopped price decreases, damaged genetics, and delayed the advent of whole-genome sequencing by perhaps a decade? Was it really just the Illuminati’s fault?
  • Why do humans have such a large mutation load on common genetic variants? Common SNPs make up a large fraction of variance, even for traits which must be fitness-affecting. Culture or technology slow evolution doesn’t wash when human fitness differentials are so large and so many people died young or as infants, and how did the many deleterious variants get pushed up to such high frequencies in the first place?
  • Why does the immune system so often surface as a genetic correlation or tissue enrichment in GWASes for many things not generally believed to be infectious? Are we missing an enormous range of infections directly causing bad things (or indirectly through autoimmune mechanisms), or the immune system just sort of like intelligence in being a general health trait?
  • Why does catnip response vary so much across countries in domestic cats, and also across feline species, with no apparent phylogenetic or environmental pattern? It is so heritable in domestic cats that a genetic reason is plausible, but if it’s adaptive, what is it doing when catnip doesn’t exist in the ranges of most tested cats, and if it’s neutral why can so many closely-different different animals respond to it in different ways?

TC again: There is much more at the link.  If you don’t know Gwern, you should know Gwern.


There are too many questions. Gwern is a loose cannon. At least I have answers to my questions.

Questions without answers are important. They're underrated compared to questions with answers.

But how do they rate versus questions with correct answers?

"What is personal productivity and why does it vary from day to day so strikingly, and yet not correlate with environmental variables like weather or sleep quality"

As a blogger, I have a pretty easily available measures of my daily productivity: numbers of posts, original words written, visits, pages viewed, comments, etc.

The most obvious factor that correlates with my personal productivity each day is the number of hours of sleep I had enjoyed over the previous 2 to 3 nights.

I’ve found you can go one night without adequate sleep, as long as you eat well the next day. Two nights without sleep, and productivity falls enormously.

Sleep and productivity might still correlate, just in a stepwise function.

While this is a good point, there is still a huge amount of variation this doesn't explain. Before having kids, I used to get 7-8 hours of sleep nearly every night, but still my productivity would vary wildly.

Why did Jean Calment [1875-1997) live so many more years than other centenarians?

Why did Vincent Van Gogh (1853-1890), who knew Calment, live only 37 years?

On average, Calment and Van Gogh lived almost 80 years.

You need more rest my fiend.


This is beyond fubar, son.

These are the kinds of questions we would ask in late night dorm room discussions. Well, not in a dorm room for me since I never lived in one, but these are the kinds of questions we would ask when the inspiration came upon us, the inspiration usually triggered by mind enhancers. I suspect many of Gwern's questions can be answered by migration, which mixed the gene pool. I suppose the really big question is whether the mixing of the gene pool will advance humanity or destroy it. On a personal level, I consider my own maternal gene pool, with the advantage of my mother having been an only and her (my) ancestors all having been born in the 19th century in a world (and gene pool) that was much smaller. Why did my ancestors (seven great uncles and aunts) accomplish so much, both academically and in their careers without the advantage of wealth? Serendipity? Is the negative reaction to gene editing because it will likely be used to clean up the gene pool after all these years of mixing? Maybe laws restricting cousin marriage will be relaxed after cleaning up the gene pool.

Which is extraordinary considering that she smoked ... could smoking be much more dangerous for some people than for others?

medicine has continuously advanced ... I dare say, but it's Golden Age is well past.

the global population has increased ... but much of that increase has happened in Shitholia.

life expectancy in general has increased ... true, but there are indications - which might of course prove fleeting - that in some developed countries the increase is stopping.

and the Gompertz curve implies that ... if the model doesn't accord with the facts the model is wrong.

AAAAAArgh. Its Golden Age.

Please correct the centenarian's name: it was Jeanne Calment (female), not Jean Calment (male).

My personal productivity varies inversely with how interesting TC's blog posts are. The more interesting, the less time I devote to getting things done.

Instead, I waste time "learning" and other crap like that.... (grin)

What, algorithmicly, are mathematicians doing when they do math which explains how their proofs can usually be wrong but their results usually right?

That one always got me.

Me too. It seems to be a generalizable phenomenon of insight. In some sense they seem to really be “seeing” the answer. Sight metaphors for thought are compelling for a reason. It may not be simply that the algorithmic processes are hidden, it may actually be that there is something real and they are apprehending it. Maybe the mind is simply adequated to reality?

On obese critters, including us, when did grain farmers start using herbicides as a dessicant?

I am not sure that is it, but if you want a change affecting "humans, pets, and even lab animals of many species kept in controlled lab conditions on standardized diets" there it is.

My first thought went to all the additional estrogen in the water from birth control.

Yeah, we could generalize it to all the chemicals that have crept into "standardized diets."

How about infections?

A lot of interesting stuff, but the items often would benefit from Gwern doing more research. One gets the impression he does a little research but then stops, and so he writes "no one has looked into this"-type sentences, when in fact many people have.

To give just one example, his piece on "sacs" implies it's a new idea, but it is in fact a long-time SF trope going back to Arthur Conan Doyle and continuing until (at least) Ian Banks.

Still, I admit I have bookmarked the "Notes" site.

I can't vouch for his sacs piece, but gwern is an incredibly diligent researcher and experimenter. I'd call him in outlier in that regard even among the rationalist-autodidact types.

I've only read part of Doyle or Banks's corpuses, but I don't recall any evolved hydrogen-sac organisms. Can you be more specific?

One more.
If Pluto was a dog then what was Goofy?

Dear god, that might be my new favourite site for demonstrating the importance of typography.

Is that an insult or a compliment? Hard to tell. The page seems perfectly readable to me, a lot more so than a lot of other websites these days.

The site doesn't look "modern"--i.e. it doesn't contain loads and loads of unnecessary whitespace, so you can actually see quite a bit of content on the screen at once.

I don't see anything problematic (or particularly noteworthy) about the fonts.

The wikipedia page on genetic correlations confused me. There is a list of of genetic correlations "both positive and negative," which includes traits with traits and diseases with traits. Can someone explain why an entry like "borderline personality disorder/sexual-emotional-physical abuse in childhood" made it onto the list?

Incidentally I went to the linked site because Tyler's sentence "Why does the immune system so often surface as a genetic correlation or tissue enrichment in GWASes for many things not generally believed to be infectious?" didn't precisely make sense to me as written, but I never did find that one.

Sorry, that was not Tyler's rewording, and there was nothing more at the link. I guess he was talking about schizophrenia, autism, stuff like that.

Odd fellow. I can relate to his confusion about past standards of beauty, having often puzzled over portraits of "noted beauties" with weak chins or mean little mouths, or sepia-toned photos of plain-faced famous seductresses, or even pictures of starlets like Betty Grable; re the fried hair, I recently saw a Doris Day movie, with Frank Sinatra, in which she was quite charming, but I wondered why the studio made her look like a big platinum Gerber baby, complete with curl ... photos of her in late middle age, with normal length hair or a wig, actually better show her undeniable good looks. Or I thought I could relate, but I see that he loves anime, so I guess our similar reactions were a coincidence.

Perhaps it's something about the way the question is worded, but I just don't understand how something that happened after late 2012 could have "delayed the advent of whole-genome sequencing by perhaps a decade" when this service was -- as the link explains -- available a full three years earlier...

Oh wow, this guy is cool. They mystique makes him cooler.

The page has a nice layout. It is interesting and bold to put all one's diverse and contrasting opinions on one page. It would be like me telling you Trump sucks, Baden-Powell is a nice hike, and a Big Green Egg (or suitable knock-off) is a great barbecue, all on the same page. With GPS way-points and recipes.

My guesses

Personal productivity measures output, not input in a meaningful sense, there being many, many inputs necessary before output is possible. While time is meaningful in terms of cost, it is much less so in terms of where you need to be from where you are. There are two outputs, the measured output, and your changed state of knowledge which isn't measured.

Music works by supressing other distractions.

Mathematicians are path finding between known or likely truths in proofs. However these paths can be treacherous and have hidden assumptions or gaps that aren't seen, can't be filled, or need another path between them. The tree may be there, but we don't know it, only try to find it

The probablity distribution of outliers is logarithmic with no expectation.

No idea.

Generation of sequencing equipment. Existing equipment is used until cost effective to develop and replace existing equipment. Meanwhile investments are made in the next generation.

It takes a long time to eradicate bad genes, especially when the population is increasing rapidly and bad is still survivable and the leading causes of death were not primarily genetic. Life expectancy was only 39 in 1900 and that wasn't due to bad genes for the most part..

Environmental exposure seems likely.

No idea.

re:Jeanne Calment
optimistic answer:
Of all the people in her small age group, she was the only one, who still did something approaching exercise. And apparently she just never stopped moving. That alone wont make you immortal, but it gains you a couple of years, if you're in a cohort of the least mortal.
"Her exercises included flexing and extending the hands ("a distinguished woman must have beautiful hands"), then the legs, and her carers noted that she moved faster than the other residents, who were 30 years younger, despite her blindness."
Though, I have no idea if the other centenarians were physically active. But since old people usually aren't, I'm guessing it's unlikely.
Haven't heard of a centenarian crossfitter. Some of them swim maybe?

Well, I suppose that's actually a pessimistic answer.

"Why do humans, pets, and even lab animals of many species kept in controlled lab conditions on standardized diets appear to be increasingly obese over the 20th century? What could possibly explain all of them simultaneously becoming obese?"
[pure conjecture]
Because if an organism gets fed the same thing over and over again two things happen:

It gets better at absorbing more energy out of it.

An organism would prefer a varied diet over a non-standard diet, if available. The natural situation, where the diet is standardized is during Fall, when food gets scarcer and season-specialized plants are eaten. At the point, when the standardized diet isn't available at all, it'll be time for hibernation. So you'd better be fat for it.

The species in question don't hibernate, but hibernation is just the extreme end of a seasonal pattern, that most mammals exhibit.
[I remember some historical anecdote about French villagers being huddled together during Winter in the farm house and mostly not doing anything. Doing things cost energy. And that's scarce in subsistence farming. Especially after a bad harvest (which is probably like every other harvest).

No idea what mice eat during Fall, but all Mammals are able to trigger some version of hibernation, because it's a fairly hardcoded response, given the right circumstances.
Humans don't normally hibernate like a Grizzly bear, but sometimes they kinda do. If some common ancestor of all mammals had the hibernation pattern, it might not have been a fitness-neutral trait, so it never got bred out.
[/pure conjecture]

That can't be an answer because it doesn't explain why it changed over time. All those species evolved well before the 20th century, and if hibernation mechanisms caused immediate obesity in a lab setting, they obviously would all have been highly obese as soon as lab records started being kept, and there would have been no increase over time.

"These animals are typically fed ad libitum, so if weight increases are attributable to increases in food consumption (which is possible), it is difficult to understand why animals in controlled environments on diets of constant composition are consuming more food today than in past decades. "

Haven't we been told that the nutritional value of grains, fruits, and vegetables has declined over the past century (supposedly because of nutrient depletion in the soil)? Would this prompt us to seek more calories?

A bit of perspective on wrong proofs for the correct theorems. I believe a majority of mathematicians build a mental model for their proofs before writing them, and the writing of proof is an exercise in expressing that model in a suitable formal structure. Sometimes there will be problems with this translation process, much like using one word while intending for another. If the mental model is correct, this results in a proof with many "local" fixable problems, but no fundamental problems. There is a joke about a mathematician I know who always make sign mistakes, but always make even number of them so they cancel out in the end.

One must caution, though, sometime the inability to translate the mental model to a proof is an indication that the mental model is wrong. This also happens plenty of times.

Why do humans, pets, and even lab animals of many species kept in controlled lab conditions on standardized diets appear to be increasingly obese over the 20th century? What could possibly explain all of them simultaneously becoming obese?

Average globabl temperature has risen 0.8 degrees celsius over the last century. How much fat would be required to store all the energy no longer required for thermoregulation by the combined biomass of all endotherms on the planet?

Why do humans, pets, and even lab animals of many species kept in controlled lab conditions on standardized diets appear to be increasingly obese over the 20th century? What could possibly explain all of them simultaneously becoming obese?

“Every leaf and every grass blade on earth makes more and more sugars as CO2 levels keep rising,” Loladze said. “We are witnessing the greatest injection of carbohydrates into the biosphere in human history―[an] injection that dilutes other nutrients in our food supply.”

Comments for this post are closed