Mirzakhani will always have a special place in my heart.
I was an undergraduate mathematics student when I discovered her work. It was a paper on closed geodesics and there was something special about her writing. Her approaches were simple an elegant -the kind that made you, as a reader, feel accomplished for understanding such a complex subject. It wasn't long until she was placed among other grand mathematicians that I looked up to.
A year later she died. I wasn't even aware of her health. It sucked to see an idol go so young. But it's incredible what she accomplished within her lifetime. She'll always be one of the greatest.
Long story short: Imagine you have an object, place an ant on the surface of the object, and then instruct the ant to walk in a straight line forever. Will the ant ever end up in the same place it started (with the same initial direction)? If so, then it has formed a closed geodesic. For some objects, the answer is obvious. For a perfect sphere, the answer is always yes. In fact, any sphere-like object (imagine warping/contorting a sphere without tearing or poking holes in it) will always have at least 3 such closed geodesics: https://en.wikipedia.org/wiki/Theorem_of_the_three_geodesics.
Miriam managed to construct an amazing formula that, when given the number of holes in an object, can give you the probability of forming a closed geodesic when starting from a random point in a random direction.
I know everyone's reaction (mine included) was: "No shit, people who drink low-fat milk probably live healthier lives in general"... But it looks like they really did account for other variables and found that milk-fat percentage was the only strong factor in play:
> "High-fat milk consumers may have lifestyles that are less healthy than low-fat milk drinkers. Since this possibility was recognized before the onset of the investigation, statistical adjustments were made for a dozen potential confounders. Statistical analyses determined that these variables had little influence on the milk fat and telomere relationship. Nevertheless, other variables could explain some of the relationship between milk fat intake and telomere length identified in the present investigation."
Okay, but let's stop and think about it for ten seconds: Of course drinking a slightly different brand of milk doesn't make you live 4.5 years longer; the headline claim is obvious nonsense. (And as I understand it, it's now thought that to the extent there is any difference, it's in favor of the high-fat milk.)
So then the conclusion seems to be that confounding factors have an extremely strong effect on the conclusion, even when the investigators have tried hard to screen them out.
Why? What went wrong with the attempts to screen out the confounding factors?
To be fair, it's a pretty large claim to just assume that something "went wrong with the attempts to screen out the confounding factors" and that the conclusion must be false.
It's a shocking conclusion, I'll give you that. And to be honest, I'm not convinced either. You might very well be right. But it's a strong claim to make agains a peer-reviewed journal publication.
> But it's a strong claim to make against a peer-reviewed journal publication.
Remember that peer-review only means that two or three persons had read the manuscript and found no obvious error and think it's inteligible and interesting. It doesn't mean that the reviewers had reproducer and checked all the details.
It is more trustworthy that a webpage in all-caps with white text over a black background, but it depends a lot of the journal. There are serious journals, and there are crappy journal that publish any rubbish if you pay them.
This is not may area, so I'm not sure. I looked at the other articles published in the journal and they look fine, but this is not my area. (Crackpot articles tend to aggregate, so looking at the other articles is sometimes useful.)
It's is very strange that an article about 5834 persons has only one author. Again, this is not may area, but I'd expect 5 or 6 authors. (The other articles in the journal have multiple authors.) It's not a smoking gun, but it's very strange.
In the article, the more interesting part is table 4 https://new.hindawi.com/journals/omcl/2019/1574021/tab4/ It looks fine. I didn't redo all the calculations, but it looks fine. (It may be a professional defect I have, but for me most articles are "Bla bla, bla bla, important table, bla bla.".) The table looks nice.
Also, they are measuring telomere length, not a self reported coefficient. I never trust self reported data. (Some of the covariant they use are difficult to measure like "percentage of total energy derived from saturated fat". How did they measure that? Anyway, I don't expect that to be a problem.)
It's important to wait until the study has been reproduced. (Exact reproductions are difficult to finance and publish, but you can make a twist. For example comparing the four combinations of normal vs cocoa milk and 1% vs 2% milk.)
@GP: Note that the (research) article does not claim that you live 4.5 years longer. They claim that the telomeres are reduced approximately 145 bases, that is somehow equivalent to 4.5 years of aging. Probably having 145 less bases in the telomeres increase some illness, but I doubt it affect too much the accident death rate.
I am inclined to think genetic signature matters. Is the lengthening of telemores a genetic thing? I am not sure if that’s even a right question to ask.
Isn't milk such a small part of most people's diets that it couldn't possibly have much of an effect? Especially for adults. Most people I know just put a bit in coffee and that's about it. I can't remember the last time I drank a glass of milk.
(Note I mean milk the drink, dairy is probably a big part of diets but that's not being studied here)
That was also one of my initial thoughts. But who knows. I'm sure milk consumption varies a lot in different geographies, cultural backgrounds, upbringings, etc...
I personally never drink milk. The only dairy product I regularly consume is butter. But I had a roommate who drank a glass of milk with dinner every day and occasionally had a bowl of cereal in the mornings. So It's at least a spectrum.
My immediate reaction as well. Seems like the perfect case of confounding. But after reading the paper, I saw this statement:
> "High-fat milk consumers may have lifestyles that are less healthy than low-fat milk drinkers. Since this possibility was recognized before the onset of the investigation, statistical adjustments were made for a dozen potential confounders. Statistical analyses determined that these variables had little influence on the milk fat and telomere relationship. Nevertheless, other variables could explain some of the relationship between milk fat intake and telomere length identified in the present investigation."
They didn't go into great detail as to what those confounders were, but it looks like they took that into account and isolated milk-fat percentage as well as they could. Source: https://new.hindawi.com/journals/omcl/2019/1574021/
Exactly. As I said, they didn't reveal much of the process that "accounted for confounding" but I just wanted to dismiss everyone's initial reaction that this is just an obvious case of confounding.
Perhaps it is. Perhaps it isn't. It just depends on how much you trust their methodology.
> It searches until it finds a seed, then goes back to the trail, maybe using the angle of the sunlight as a guide, to return to the nest, following the stream of outgoing foragers.
I remember reading that ants counted steps to find their way home. Perhaps I'm remembering incorrectly or maybe it was false?
Either way, cool article. Emergence is a cool property that shows up everywhere!
They say they're coming out with Android and iOS versions soon, so maybe take a look after that point to see how they've tweaked the models and if the error rates are a lot higher.
FWIW, we have already working versions in Android, iOS but didn't have time to open-source it with the current release. This is certainly in our future work.
The way I understand it (which isn't that well either tbh) is that most modern anti-cheat softwares run off of some Hardware ID (HWID). An amalgam of ids from your GPU, CPU, OS, Mac adress, and so on... that are somehow combined or hashed into some uniquely identifying ID. This is a bit harder to spoof than an IP address ban. And most are robust to gradual changes. Like Theseus' ship, if you change your GPU one day, OS the next, and so on... is it the same computer you should ban? Most HWIDs generated by games won't change until a certain number of those variables are messed around with.
Once again, I'm no expert but I think that's at least helpful.
This is called hardware banning and is not "super" common in the PC world, usually they just ban accounts. I know overwatch does this HW bans, but no VAC protected game does.
