Hacker News Reader

Comment by Separo | Hacker News Reader

Well maybe in time ML will help break through the high energy physics roadblocks.

it already had, bottom-quark tagging has improved O(10)x in efficiency in the last decade without any new "physics" understanding, just from training with more low-level data and better ML arch (now using Transformers)

but we haven't found new physics with or without ML, making this prize a little weird.

Maybe it's a prize for hope and change that physics will be revolutionized by neural networks? Similar to how Obama got a Nobel Peace Prize in order to repudiate Bush's legacy in Iraq and Afghanistan. While Bush's legacy absolutely deserved to be repudiated, I don't think awarding a new president the Peace Prize was the best way to do it, especially because in the foreign policy realm, he ended up not so different from Bush.

Maybe read the announcement?

"With their breakthroughs, that stand on the foundations of physical science, they have showed a completely new way for us to use computers to aid and to guide us to tackle many of the challenges our society face. Simply put, thanks to their work Humanity now has a new item in its toolbox, which we can choose to use for good purposes. Machine learning based on ANNs is currently revolutionizing science, engineering and daily life. The field is already on its way to enable breakthroughs toward building a sustainable society, e.g. by helping to identify new functional materials. How deep learning by ANNs will be used in the future depends on how we humans choose to use these incredibly potent tools, already present in many aspects of our lives"

We used ML to discover the Higgs.

Whether or not the original Higgs discovery decay channels used ML, confirming that it was in fact the Higgs required measuring the decay to b-quarks, which has used ML since the LHC started taking data.

Over the lifetime of the LHC, he backgrounds got around 10x smaller for the same "efficiency" (fraction of true b-quarks tagged) if you want to be pedantic about the definitions. We've used NNs in b-tagging for decades now, so it was always possible to dial in a threshold for tagging that was e.g. 70% efficient.

Transformers gave us a factor of a few smaller backgrounds in the last few years though [1].

[1]: https://atlas.web.cern.ch/Atlas/GROUPS/PHYSICS/PLOTS/FTAG-20...

I sort of agree in principle but in practise they've always taken a broad view.

Kissinger was one of the most prominent disrupters of world peace in the postwar era but that didn't stop him winning the peace prize. Churchill won the literature prize for defeating Hitler. The blue led guys a few years back didn't do much except make a thing that would go on every single consumer gadget and disrupt my sleep but they won the physics prize.

Even when they get it right they often get it wrong. For example I believe Einstein supposedly won for "especially his work on the photoelectric effect" rather than relativity.

Einstein’s work on the photoelectric effect was incredibly important, and incredibly influential on other research at the time. He proposed that light was quantised - essentially the foundation of quantum mechanics.

It’s no exaggeration that Einstein’s work on the photoelectric effect was as important as special or general relativity, and it had the advantage of strong experimental verification by 1921.

The main reason that prize is remarkable is that Einstein himself hated quantum mechanics - but that doesn’t dispute the work’s importance.

The discovery of the photoelectric effect was certainly as important as relativity in terms of how much it affects society. But it was only an incremental advanced on top of Planck work on blackbody radiation.

I'm not saying that photoelectric effect didn't deserve a Nobel Prize. But relativity completely supplanted Newtonian Physics, and Einstein played a much greater role in this revolution than he did in that of Quantum Mechanics.

Also, I believe historical records have made it clear that relativity, even if it was still considered controversial in the '20s (and so not mentioned specifically), was indeed part of the reason he was awarded the prize.

Also, consider WHY it was still controversial, despite evidence piling up even for relativity. It was seen as such a fundamental shift away from common-sense understanding of the physical world that people refused to believe it, despite evidence.

Just like how many people to this day do not believe it's possilbe to build AI out of regular computers, as their intuition tells them that some magic vodoo needs to be there for "true" inteligence.

I would add to this that it had the advantage of something like 40 years of history as a field that was the basis for some of the biggest advances in instrumentation of that era.

>Einstein supposedly won for "especially his work on the photoelectric effect" rather than relativity.

just adding to this, this is because relativity wasn't experimentally verified (i.e. not sure if it's reality) at the time.

Also, the prize is about the greatest benefit to humankind according to Alfred Nobel, not the most impressive research. Arguably, the photoelectric effect fits that notion better than GR or any other of Einstein's research.

Besides that, Einstein received the prize in 1921, whereas the Eddington experiment in 1919 generally counts as the first experimental verification of GR.

> Arguably, the photoelectric effect fits that notion better than GR or any other of Einstein's research

Today we could argue about it due to the importance of solar panels, but that was hard to forecast in 1921. Also, without GR there would be no GPS so it's not like it doesn't bring benefits to humanity.

Einstein laid the foundation of quantum mechanics with his description of the photoelectric effect, so you could add transistors, lasers, LEDs, CCD sensors and more to the list. Although I agree that it's doubtful that most of this could have been foreseen then.

Surely they would have just noticed a discrepancy in timing and added a few circles-upon-circles to effectively fix it up? Is deeply grokking relativity necessary for GPS to work?

On the other hand, it would be impossible to make those adjustments without someone coming up with GR :-)

More to the point, photoemission spectroscopy has been a workhorse tool for understanding the electronic properties of materials for quite a long time now (though perhaps not yet in 1921).

Not supposedly.

"for his services to Theoretical Physics, and especially for his discovery of the law of the photoelectric effect"

https://www.nobelprize.org/prizes/lists/all-nobel-prizes-in-...

Nobel prizes are generally awarded for verifiable observations but, also require real world applications.

Einstein won the physics prize on the photoelectric effect due to having real world applications and observable and if GPS actually existed while he was arrived (yes I know this is a stretch) he would have gotten it for relativity.

Blue LEDs allows you to access more of the color spectrum for LEDs in general and they were not easy to make.

For this year it does feel like a very large leaning into practical applications instead of physics though. Did we run out of interesting physics in the last year?

Relativity had just as much real world relevance in the 1920's as the Higgs boson has today....

I would add to this list Bernard Russel who won the Nobel in literature for being a public intellectual.

I should take up being a public intellectual, instead of a public nuisance.

Most people on the Internet and in a certain orange forum might consider it seriously. (I do think about it myself.)

Bertrand :)

The Nobel peace prize was a mistake. Peace is not a science, and you can't objectively measure how much anyone has helped peace, especially not before a few decades has passed.

So I agree that the peace prize committee has made some bad choices, but they do have an impossible job.

Oh come on, blue LEDs were a feat of physics and chemistry mastery.

I'm sure they are but they drive me nuts. If I ever become filthy rich and in doing so sell my soul and become a bad person, one of my priorities will doubtless be to have the blue led inventors hunted down remorselessly.[1]

[1] Note to future law-enforcement: I am honestly kidding. I wouldn't hurt a fly, officer.

A black sharpie over the offending led indicators will fix that. Now you can enjoy your sleep uninterrupted by dreams of manhunts and mephistophelian bargains.

Veritasium has a great video on the difficult physics of the blue led. Highly recommend if you think it didn't deserve the prize.

How about a prize for the Monte Carlo simulation methods needed as input to these models?

ML absolutely has helped astrophysics in sorting the massive amount of observation data to make new discovery.

So if someone will invent say a new keyboard layout which will improve median data input rate by 10% and will be used by astrophysicists then it will be worthy of the astrophysics prize? Or better yet - in your example the main driver for the ML is Nvidia, should be award Jensen a prize in astrophysics? Or in any other field where ML is deployed? In my opinion we should separate efforts of people making tools, from the efforts of people doing research using said tools.

No. Because the keyboard while faster is not instrumental in finding new stars. ML is instrumental in finding new stars and new planets.

So do telescopes. Has anyone every won a Nobel for a telescope?

People have won it over new microscope designs and techniques.. possibly telescopes, too.. but I'm less familiar with that and not somewhere where it's convenient to look it up.

In 1986 and 2014, science (electron optics, nanoscopy/nanolasers) came first. Then the microscopy. Even 2017 won for 3D microscopy. What Nobel-worthy physics does thing do?

Most disciplines in CS have done that one way or another.

One of the very early successful applications of ML was using neural network and other models in particle identification systems in particle physics experiments.

If I may be so bold, a breakthrough will require new experimental techniques, and we aren't likely to get those from ML.

Reply on news.ycombinator.com