Dear reader, this will be a tad difficult, but I urge you to plow through as best you can, because this is becoming “a thing” in some areas. I stripped out most technicalities. It will not, or rather should not, come as any surprise that most statistical models are entirely
All wannabe modelers must be taught Fourier transform and then sent to lose a hundred bucks in the forex market – to understand once and for all that a perfectly fitting model may still have zero predictive ability.
Have you done any work (public writings I mean) on the malign influence of the novelty requirements in modern research supervision approvals and corresponding funding in the universities?
We've now got a system that focuses ever more closely on fabulously proliferating trivially significant (if at all?) hairs on the edges of the leaves while the main branches of science are ignored because "settled".
This is good for those making careers and patent profits out of the status quo but really bad for discovering errors or truths in the main branches.
Whereas the entire history of science is one of regular generational radical overturnings of the entire foundations. "The axe is laid at the root of the tree."
Ironically the insistence on novelty works to produce stasis, another name for Death.
Very, very helpful analysis. It’s rather amazing how often statisticians make up methods and use them, but which the statistical “community” at large is unaware of and which have not been independently validated. The results get reported but the methods remain uninvestigated.
Many (many) public health agencies are relying on questionable reporting of associations to make all kinds of policy recommendations, not even waiting for any demonstration that their recommendations will have a positive effect. I wish I could find partners for you to write an “antidote” paper but I’m afraid it wouldn’t be “novel” (just a service to mankind).
Imagine an obviously absurd claim, like eating cow meat is only deadly in the case that the political ideology of the cow is different from the political ideology of the human. A normal person who eats a communist cow will *have* greater statistical risk of dying than the person who eats a normal thinking cow.
I am not a scientist, so I can discard this claim simply because all cows share the same Viking ideology (they already have horns.) No money should be spent in any scientific study that includes an *obviously* false claim.
But scientists like to give a chance to obviously false claims. Sometimes, they get in love with them and are elevated to theories, which may end up dominating an entire field of enquiry. And money is never a problem.
I wonder why is it that scientists have exactly zero common sense. Is it necessary to renounce to even a smidgen of common sense before becoming a scientist? Is there some surgery that removes common sense from young scientists? Can we ban it?
So I went and looked through the hurricane paper. I figure if the results are as impressive as they imply, there would be a table with just names and death counts and a lay person like me could eyeball it, but I couldn't find such a thing.
As Descartes's Advocate, let me just state what a pleasure it is to see the best of the Logos's Advocates falling prey to the seductive trap laid by my Master In that Hell Devoid of Telos!
The trap?
The strawmen ("statisticians") with which my Master has populated the ranks of Specification Curve Analysts!
Oh, but it gets better! There is another Ring of Hell encompassing the Ring of Strawmen:
Denial of Algorithmic Information Theory's formalization of Ockham's Razor!
We shall not go into the depths of sophistry involved in this denial, hence the Algorithmic Information Criterion for CAUSAL model selection -- at least not here and now. See Hume's Guillotine and the NiNOR Complexity alternative to Kolmogorov Complexity for the debunking of that sophistry.
No, what stalks our haplessly heroic critic of Specification Curve Analysis is the emergence of a Postmodernist demon from pseudo-telos who, posing as a Specification Curve Analyst, does the following:
1) Looks at the standard dataset provided to all Specification Curve Analysts with their arbitrarily large number of "methods".
2) Constructs a computer program that spits out the standard dataset.
3) Challenges any other Specification Curve Analyst to come up with a smaller such computer program.
4) Places our haplessly heroic defender of Telos hence Logos in league with the Postmodernist pseud-Telos "statisticians" in setting forth the standard "gotchas" regarding Algorithmic Information Theory (hence Algorithmic Probability, Solomonoff Induction, etc.) -- all of which are ALREADY laid to rest by none other than this Descartes's Advocate:
That's not fair at all. Excel is a great tool. You can't blame its users -- who would have made the same mistakes some other way -- for what they do with it.
Maybe I´m being unfair. But I´m also old. So let´s spread the blame around, on some of that early 90´s stats software that started coming out with the Pentium processors (boy, I AM OLD! OLDER THAN BRIGGS!). Here in Brasil it was all PIRATE STUFF, so you had to reset the date-time on the PC before you could use it at the university lab. Ah, the good old days!
All wannabe modelers must be taught Fourier transform and then sent to lose a hundred bucks in the forex market – to understand once and for all that a perfectly fitting model may still have zero predictive ability.
Have you done any work (public writings I mean) on the malign influence of the novelty requirements in modern research supervision approvals and corresponding funding in the universities?
We've now got a system that focuses ever more closely on fabulously proliferating trivially significant (if at all?) hairs on the edges of the leaves while the main branches of science are ignored because "settled".
This is good for those making careers and patent profits out of the status quo but really bad for discovering errors or truths in the main branches.
Whereas the entire history of science is one of regular generational radical overturnings of the entire foundations. "The axe is laid at the root of the tree."
Ironically the insistence on novelty works to produce stasis, another name for Death.
Very, very helpful analysis. It’s rather amazing how often statisticians make up methods and use them, but which the statistical “community” at large is unaware of and which have not been independently validated. The results get reported but the methods remain uninvestigated.
Many (many) public health agencies are relying on questionable reporting of associations to make all kinds of policy recommendations, not even waiting for any demonstration that their recommendations will have a positive effect. I wish I could find partners for you to write an “antidote” paper but I’m afraid it wouldn’t be “novel” (just a service to mankind).
We've written enough. But they never take.
Problem is, we recommend slowing down and checking. Nobody wants to slow down, nobody wants to check.
Got it. Let’s make a list of some references of papers already written that can be handed over to stastisically-illiterate folks.
Good idea. Will do.
Imagine an obviously absurd claim, like eating cow meat is only deadly in the case that the political ideology of the cow is different from the political ideology of the human. A normal person who eats a communist cow will *have* greater statistical risk of dying than the person who eats a normal thinking cow.
I am not a scientist, so I can discard this claim simply because all cows share the same Viking ideology (they already have horns.) No money should be spent in any scientific study that includes an *obviously* false claim.
But scientists like to give a chance to obviously false claims. Sometimes, they get in love with them and are elevated to theories, which may end up dominating an entire field of enquiry. And money is never a problem.
I wonder why is it that scientists have exactly zero common sense. Is it necessary to renounce to even a smidgen of common sense before becoming a scientist? Is there some surgery that removes common sense from young scientists? Can we ban it?
So I went and looked through the hurricane paper. I figure if the results are as impressive as they imply, there would be a table with just names and death counts and a lay person like me could eyeball it, but I couldn't find such a thing.
As Descartes's Advocate, let me just state what a pleasure it is to see the best of the Logos's Advocates falling prey to the seductive trap laid by my Master In that Hell Devoid of Telos!
The trap?
The strawmen ("statisticians") with which my Master has populated the ranks of Specification Curve Analysts!
Oh, but it gets better! There is another Ring of Hell encompassing the Ring of Strawmen:
Denial of Algorithmic Information Theory's formalization of Ockham's Razor!
We shall not go into the depths of sophistry involved in this denial, hence the Algorithmic Information Criterion for CAUSAL model selection -- at least not here and now. See Hume's Guillotine and the NiNOR Complexity alternative to Kolmogorov Complexity for the debunking of that sophistry.
No, what stalks our haplessly heroic critic of Specification Curve Analysis is the emergence of a Postmodernist demon from pseudo-telos who, posing as a Specification Curve Analyst, does the following:
1) Looks at the standard dataset provided to all Specification Curve Analysts with their arbitrarily large number of "methods".
2) Constructs a computer program that spits out the standard dataset.
3) Challenges any other Specification Curve Analyst to come up with a smaller such computer program.
4) Places our haplessly heroic defender of Telos hence Logos in league with the Postmodernist pseud-Telos "statisticians" in setting forth the standard "gotchas" regarding Algorithmic Information Theory (hence Algorithmic Probability, Solomonoff Induction, etc.) -- all of which are ALREADY laid to rest by none other than this Descartes's Advocate:
https://github.com/jabowery/HumesGuillotine
&
https://groups.google.com/g/ait0/c/D1wd2fV6Ax4/m/zJLCmh00BAAJ
Our haplessly heroic advocate of Telos will then find himself mired in yet another ring of Hell Devoid of Telos!
BWAAHAHAHAHAHAHA!!!!
PS: I'm a mole who has infiltrated that ring of the Hell Devoid of Telos. But don't tell the other demons.
"which must, by kingly decree," of patent offices
Good stuff, Briggs.
The effect of MICROSOFT EXCEL on science has been truly astounding.
That's not fair at all. Excel is a great tool. You can't blame its users -- who would have made the same mistakes some other way -- for what they do with it.
Maybe I´m being unfair. But I´m also old. So let´s spread the blame around, on some of that early 90´s stats software that started coming out with the Pentium processors (boy, I AM OLD! OLDER THAN BRIGGS!). Here in Brasil it was all PIRATE STUFF, so you had to reset the date-time on the PC before you could use it at the university lab. Ah, the good old days!
Eek! Your spell-checker substituted “eek” for “eke”!
No. My enemies did this. They are always doing things like this. Evil so-and-sos.