Hacker News Reader

Comment by kragen | Hacker News Reader

Maybe what we need is cheaper physics apparatus so you don't need a physics professorship or a big lab to advance the state of human knowledge. You aren't going to scale down the LHC to fit under your bed when you aren't using it, but you could surely fit an XRF, ICP-AES, AFM, and maybe an optical bench in there. For a while I worked at a satellite company whose first cleanroom was made only a few years back by covering the concrete walls with polyethylene film, and a class-10 clean bench that you stick your hands into is within the reach of lots of people. Lots of amateurs have built fusors, but mostly they aren't doing the work necessary to measure reliable, reproducible results, in part because Vixra doesn't offer any incentive to do so. Radio amateurs are one shining exception here, even if most of them are just using store-bought equipment these days.

View on HN

You'll also need to pay for the people doing the research to buy houses, have cars, retirement savings, raise families, and have a decent quality of life too.

Add that to a bare bones optics lab with a few staff and you will be running about $1 million startup costs for benches, lasers, optics, closed loop cryogencics, He4, interferometers, etc, plus half a million a year salary costs for 3-5 people in a low cost of living area.

It would also need job security to be competitive with going into something like data science, making the startup costs a careers worth of funding otherwise it would be insanity to choose.

Source: PhD in quantum optics, no longer do science.

> You'll also need to pay for the people doing the research to buy houses, have cars, retirement savings, raise families, and have a decent quality of life too.

Newton and Cavendish didn't have any of those, except that Cavendish had retirement savings. So while not having houses, cars, retirement savings, the ability to raise a family, and a decent quality of life might be a reason for not achieving more than Newton or Cavendish, it's not a reason for achieving less.

(I suspect that raising a family is actually counterproductive. I've seen an awful lot of promising researchers of both genders stop publishing after their first baby.)

I don't have any of those things, but while I'm no Newton or Cavendish, I can't attribute the difference to my lack of a car. Most of the people in my country don't have any of those except for houses; most of the people here who own houses built them with their own hands rather than buying them.

It's true that if you have to choose between going into data science and owning your own house and car, or doing quantum optics while living in poverty, the former is a lot better for you. But for most of us it's not an either-or choice:

1. Why not both? Lagrange did a significant amount of his work while subsisting on a day job teaching ballistics to gunners. Vipul Ved Prakash wrote Vipul's Razor by working one month out of the year in Delhi, then spending the other 11 months up in the mountains working on whatever he wanted to. Sidis deliberately didn't do anything others would consider useful, surviving on a series of menial jobs, but if he'd turned his formidable intellect on the problems of automatic computation or chemistry instead of collecting streetcar transfers, possibly he would have made significant progress. I've been living on US$6k a year, so a single year of a US$250k salary and stock package at Google or Fecebutt would allow me to survive for 40 years.

2. For most of us, it's neither. Most people don't have a data-science job available, or cars, or job security, or retirement savings. Given the choice between spending your spare time on watching Westworld and setting up experiments with Kerr cells and third-harmonic-generation crystals, what could possibly make the former a better choice? Though I'm one to talk! Here I am wasting my time commenting on the orange website, arguing about physics with people (who aren't you) whose understanding of physics evidently comes from WGBH Boston.

So, while I agree that everyone should live in material abundance, I don't agree with your apparent conclusion that more abundance of non-experimental-apparatus material goods would boost the world's research productivity. Many more people can buy houses, buy cars, have retirement savings, and raise families than 50 or 100 years ago, while the speed of scientific advancement has increased only modestly.

Theoretical physics just needs pencil and paper. If you do the more computation heavy parts of theoretical physics then add a computer as well.

Yes, but historically great advances in theoretical physics have usually followed great advances in experimental apparatus.

Like Einstein or Newton? Theoretical physics tends to be way ahead of experimental physics, it took 50 years from higgs boson existing as a theory to finding traces of the higgs boson in the LHC etc.

Physics is done using the true scientific method. First you make a theory, then you make experiments to test the theory. Theoretical physicists made a theory and proposed some possible experiments to test it. Then 50 years later data was found in those experiments to match the theory. What we lack today to further physics isn't experiments, we lack theories that are well formed enough that we can perform experiments to test them.

There's no Einstein without the Ultraviolet Catastrophe.

The Ultraviolet Catastrophe was discovered immediately after theoretical physicists derived the radiation law, the limit was theory and not experiments there as well. And the most important part here is that another physicist had already derived an alternative radiation law at the time that fit perfectly with the observed deviation and had a good explanation for it: Quantum Physics.

Edit: The problem with physics today is exactly like back then, we have no predictions to test. If someone comes up with a new theory that joins quantum physics with gravity in a way that is consistent with all past experiments, then we can test that. But there is no such theory today, nobody has figured out a way that the domains can work together.

In https://news.ycombinator.com/item?id=29144119 I listed a lot of the big unsolved problems in physics, though many would argue that the question of how consciousness arises isn't part of physics. In https://news.ycombinator.com/item?id=29153894 I also listed a lot of recent advances, many of which came from experimental physics. Some of those problems are susceptible to experimental investigation even with the theories we have.

But, aside from these big problems, many smaller problems are susceptible to experimental investigation. You may not create an Einstein-style paradigm shift by detecting CNO-fusion neutrinos from the sun, observing Higgs decay, improving quantum-dot solar cells, fabricating nanotube rope whose strength approaches that of a single nanotube, understanding the lubricity of BAM well enough to design more similar materials, deriving useful energy from the fission of lithium, making a metamaterial with a higher Young's modulus than diamond with negative-elasticity inclusions, constructing logic gates out of fluid vortices, building a usable hypersonic plasmoid pistol (one that doesn't require explosive flux compression pumping!), building transistors that function at 800 degrees, finding a SHS route to cubic boron nitride, making a 50-tesla magnetic field in the lab, finding a way to construct quantum-dot solar panels that's cheap and scalable enough to undercut poly-Si, confining a particle of gold in a stable minimum of the Casimir potential, or finding a way to fabricate high-quality optics apparatus rapidly out of aluminum foil; but the obstacles to these problems are (or were) mostly not that we don't have any useful hypotheses to test.

No. No no no. It’s extremely standard to measure a material that’s known to be “weird” and not know what you’re going to see.

"The fasted way to get the right answer is to post the wrong one on the internet"

Actually, Kepler came before Newton. Kepler's laws of planetary motion were derived from observations of others. They also follow straightforwardly from Newton's laws of motion. In particular, Kepler's second law is equivalent to conservation of angular momentum. Kepler's third law is equivalent to "Newton's universal gravitation + conservation of angular momentum". And Kepler's first is a consequence of Newtonian kinematics. That is to say, Newton generalized Kepler's laws.

The order is:

1. Astronomical observations (Brahe et. al.) 2. Kepler's laws 3. Newton's laws + universal gravitation.

Similarly with Einstein: His theoretical treatment of Brownian motion was based on observations by... Robert Brown half a century ealier. The photoelectric effect (for which Einstein was awarded the Nobel Prize) was an extension of theoretical work by Plank, who started theorizing to explain observations made by Hertz.

Special Relativity resolved a conflict between E&M and Mechanics, but it was really needed to explain why the Michelson-Morley experiment couldn't find a difference in the speed of light, despite increasingly-sophisticated apparatus (which was an unsolved paradox for a quarter of a century before SR was invented).

And, while not on your list, quantum theory had many experimental inspirations. The Millikan oil drop experiment, for one. And spectral lines in stellar observations, for another.

Yes, like Einstein or Newton.

Einstein's work on relativity (01905) was inspired by the Michelson-Morley experiment (01887), less than 20 years before, and its many improved replications. His work on the photoelectric effect was inspired by Hertz's experimental discovery of it (also 01887), followed by numerous further experiments which clarified the nature of the effect. Although Brownian motion had been observed, in some sense, since Lucretius (00060 BCE), Brown's 01827 observations under a microscope less than a century before were crucial to Einstein's theorizing about it.

Newton's work on orbital mechanics, which gave rise to understanding of universal gravitation (published 01687 but finished years earlier), derived from Kepler's laws of planetary motion (01621, say) and his published tables of planetary observations (01627), the Tabulae Rudolphinae. Not coincidentally, Kepler is also known for his dramatic improvements in the tele-scope, but much of the improvement in the Tabulae was actually due to the meticulous work done at the pre-telescope observatory of his predecessor Brahe, a huge stone structure.

Certainly the traffic between theoretical physics and experimental physics is not entirely a one-way flow from experiment to theory; that would lead only to the sort of overfitting we find in Ptolemy. But neither is it, as you paint it, entirely a one-way flow from theory to experiment.

It's probably true that we aren't going to resolve the problem of quantum gravity, dark matter, or consciousness with experiments, because our theories aren't good enough to design the experiments yet. But turbulence, magnetohydrodynamics, and especially quantum computers are eminently subject to experiment.

(Although I disagree with your comment, it certainly seems to be made in good faith, so I deplore the knuckle-draggers who are downvoting it.)

I'm curious about your reasons for prefixing years with an additional 0 in all your posts.

This notation is encouraged by the Long Now foundation which promotes long term thinking in various ways.

It’s absolutely the other way around. High Tc superconductivity, quantum hall, Geiger-Marsden, etc…

Hmm, are you saying that high Tc superconductivity, the Geiger-Marsden results falsifying the plum-pudding model, and the quantum Hall effect were predicted theoretically before they were discovered empirically? I thought that these were three of the more spectacular cases of unexpected experimental results that bore great theoretical fruit.

Possibly you intended to post your comment in reply to one of Jensson's comments rather than mine?

I accidentally read your comment the wrong way around -- sorry about that!

No worries!

It requires collaboration and mentorship too. These don’t scale super well.

No, but they do scale exponentially. Consider a random mathematician from a century ago, who I selected because he had a student in common with Sierpinski: https://www.genealogy.math.ndsu.nodak.edu/id.php?id=15165

Rajchman had two students, one of whom (Antoni Zygmund, who also studied under Mazurkiewicz, and founded the Chicago school of analysis) had 40 students, five of whom had over 100 students of their own. 18 of those 40 had at least one student of their own. Consequently Rajchman had 1658 descendants in only a century, a mentorship growth rate of 7.7% per year despite Rajchman himself having his career cut short by being murdered by the Nazis in 01940 and apparently ceasing to mentor anyone officially for the previous 15 years of his career.

Yes but the exponential scaling doesn't really work to anyone's benefit. It just means that N people can each mentor an average of N people and so on and so forth. An active community means that the people who are working on the problems can all share results and bring people up to speed. I'm not convinced THIS scales well.

I can buy more people in physics working on more problems. There are a wealth of interesting problems in physics and more people all going in different directions would be great. But ten times as many people working on the LHC? A hundred times as many people working on string theory? I don't buy it.

To me, the ideal model of fully open, accessible research is the speedrunning community. I don't see speedrunning as all that different from experimental work. You probe, you hypothesize, you have breakthroughs, you compete in what's generally a pretty healthy way, and you communicate and document. Look at how this scales, how many people get in and get obsessed, etc. To really master quantum hall, you need to have have a devotion to the field that's comparable to "completing Super Mario 64 with half an A press."

Yeah, although maybe mentorship can scale reasonably well (or at any rate much faster than we are scaling it at present), I agree with you about collaboration: ten times as many people working on the LHC (or HEP in general) probably wouldn't be very effective. Now that we have Sci-Hub, the General Index, Wikipedia, Stack Exchange, and Google Scholar, we can probably collaborate a little more effectively than before, but not enough to cram orders of magnitude of people into a given subfield.

There might be a path forward in the work on making computational work easily reproducible, by people like Konrad Hinsen, Yihui Xie, Jeremiah Orians, Eelco Dolstra, Ludovic Courtès, Shriram Krishnamurthi, Ricardo Wurmus, and Sam Tobin-Hochstadt, but clearly it hasn't been a panacea so far. Speedrunning results are in many cases reproducible by virtue of nailed-down console hardware and bit-identical game images, but that's harder to achieve even for FEM simulations of turbulent MHD systems, much less actual experimental MHD systems like a Farnsworth fusor.

How do people initially get up to speed on speedrunning? Are there tutorials, the equivalent of a textbook with problem sets, some other onramp? Can we gamify learning quantum mechanics? (I've tried QiskitBlocks but so far haven't been impressed.)

Scientific apparatus is by definition “pre-engineering.” It’s low volume and generally designed by a few people with no BOM optimization. OTOH, everything that’s engineered is generally just EE or optics lab stuff that’s already high volume and aggressively cost optimized. Good luck making a cheap MBE or dilution fridge, and good luck making a 10 GHz oscilloscope cheaper.

I've been seeing a lot of 3-D printed FDM parts showing up in labs in recent years. Of course you can't 3-D print an MBE or FIB, but maybe you could automate the manufacturing of some significant apparatus to the point where you really could download a cutfile from Thingiverse, cut it out on a CNC plasma table, and have it MIG-welded together by robots, so that, like custom T-shirts or FDM-printable parts, it can be cheap even without being high-volume. Even some high-vacuum apparatus might be accessible by that kind of route. A friend of mine has been doing a lot of optics fabrication via UV stereolithography.

I really detest the word "just." It's the ultimate signifier that someone hasn't properly engaged with a problem.

Duly deleted. Thank you for the feedback. I'm doing my best to properly engage with the problem.

Reply on news.ycombinator.com