Hacker News Reader

Hacker News

Comment by whatshisface | Hacker News Reader

Parent Full thread

whatshisface

Here is an incomplete list of things that it's impossible to research in industry:

1. Astronomy.

2. Physics.

3. Geophysics concerning the parts of the Earth deeper than the crust.

4. Biology aside from medicine.

5. Chemistry aside from industrial chemistry.

6. Theoretical computer science.

7. Mathematics.

I'm not blaming you for not knowing this, but I am holding my head in my hands - how can people not know about astronomers? They've been a part of our culture and the prestige of civilization for thousands of years.

View on HN

"Bell Laboratories has been the recipient of 11 Nobel Prizes in Physics, with notable laureates including John Bardeen, William Shockley, Walter Brattain, and Arthur Ashkin. Other notable achievements include the invention of the transistor, the discovery of the cosmic microwave background, and the development of optical tweezers."

Microsoft Research has a ton of people working on theoretical CS.

Biology - there is a ton of research in agriculture too - e.g. Monsanto and GMO seeds.

The discovery of the CMB was accidental. Bell Labs didn't fund it intentionally.

Bell Labs was at its peak from 1960s-1970s. Since the 80s, corporate governance has completely changed due to Jack Welch’s short-term shareholder maximization ideology taking over the corporate world.

I don’t think there are current private organizations doing research similar to what Bell Labs did as the current corporate-governance systems wouldn’t allow for it.

Currently, industry research is more for profit-maximization at the expense of greater human prosperity/economic growth: such as you mention Monsanto making patented seeds, increasing profits by disallowing farmers to regrow crops more cheaply which otherwise could’ve been passed onto consumers/wider society.

There is tons of research on online safety being done at platforms.

It can never be released.

Things aren't impossible to research in industry. Managers with an R&D budget can fund whatever research they want, and it's easy to find examples of companies doing research in most of the fields you name. The fields where it's not are because you've defined them in a circular fashion e.g. "chemistry aside from industrial chemistry" is not an argument for "there are things you can't do in industry", and space research is primarily driven by the private sector now but I guess you are defining astronomy to only include study of things too far away to matter to anyone except astronomers.

Apple definitely had an internal group doing mathematics research, which i know first hand. But yes there are, to your point, topics in science probably only done in academia etc, but, to my point, several are seriously funded in industry.

I couldn't say more without knowing the details, but industrial labs don't usually research mathematics, so much as ways to apply mathematics to their industry. These are called "mathematics research departments," because they hire mathematicians.

> how can people not know about astronomers?

Light pollution. Lots of people now live in places where the stars are hard to spot.

And now we also all have infinite distraction cuboids, many who can still see them if they look, won't look.

And if you don't know what's out there — or mistakenly categorise all of space as scifi — then why would you be curious about it? Why even ask who might research it?

> Astronomy.

What benefits do you expect to see from the kinds astronomy that require this sort of funding? Sure, knowing things can be nice but this ignores opportunity costs, eg. would practical knowledge like fusion research be further along if talent weren't focused on impractical knowledge?

> Physics.

Not strictly true, see quantum computing for instance, lasers, semiconductors and so on. There are some types of physics that aren't viable in this sense, but why does that automatically translate into some need to support them? For instance, consider the decades spent on supersymmetry which ultimately produced bupkis. In a world in which we weren't so focused on ideas so divorced from empirical data, what other types of knowledge or engineering would we have done?

> Geophysics concerning the parts of the Earth deeper than the crust.

What benefits do you expect to see?

> Biology aside from medicine.

Such as? What benefits do you expect to see?

> Chemistry aside from industrial chemistry.

Such as? What benefits do you expect to see?

> Theoretical computer science.

Untrue, Google and Facebook have advance distributed computing considerably, for instance.

> Mathematics.

Unclear, there's a lot of math involved in predictions of all sorts, like weather forecasting, stock market prediction. If your argument here is that math will be more application-focused, this strikes me much like the physics objection where it's unclear that we'd really be worse off.

There seems to be this automatic assumption among some people that pure research with no direction or constraints is an unmitigated good and that we can't do better. I used to think so too, but I just don't see it anymore.

Basically none of modern optics would exist without astronomy (well at least astronomy is a convenient cover for military/intelligence interests funding better optics). Most of statistics and efficient cameras originate in astronomy/astrophysics (mostly because you have to count all the photons and you are never getting a second relevant measurement point)

There are huge parts of physics which are only publicly funded. Results are often spun out into companies, but there is no institution that can fund experiments that require timelines of multiple decades (even things like fusion power is nearly completely government funds)

And those are the only two parts where I actually have some competence. So yeah.. I wouldn't buy

> well at least astronomy is a convenient cover for military/intelligence interests funding better optics

Right, defense can and has funded research for its own purposes, and sometimes those purposes can find wider commercial application (like the internet). That's all great, national defense is one of the government's primary purposes.

> There are huge parts of physics which are only publicly funded.

Yes, and? Is this an argument that they cannot be funded in other ways, or an argument that the parts of physics that cannot be funded in any other way ought to be publicly funded? There's just this blanket assumption that this is true but it simply doesn't follow.

For instance, the newest super collider project that some people are pushing for completely misses the opportunity cost of not funding other projects that could be far more impactful, like wakefield accelerators, which would reduce the size and cost of particle accelerators by orders of magnitude.

> For instance, the newest super collider project that some people are pushing for completely misses the opportunity cost of not funding other projects that could be far more impactful, like wakefield accelerators, which would reduce the size and cost of particle accelerators by orders of magnitude.

This is not true in many aspects. There are many problems with plasma and laser wakefield acceleration. First, the beam quality (emittance and stability) is orders of magnitude below collider requirements. They have demonstrated GeV-scale acceleration over centimeters, but scaling to multi-TeV and maintaining luminosity is not even close to solved. There are no concept for a full detector-ready experimental program exists using wakefield accelerators. But on the other hand, we have "FCC" being based on mature accelerator technologies, with well-understood cost scaling and detector integration that builds on decades on experience building accelerators. Actually it is much safer option than what you are saying.

But the important point is that you are making it binary choice, we can still investigate and work on wakefield accelerators while working on more mature projects. Remember than it takes decades of work and thousands of scientists to make any of these things work. And it is not the question of accelerator itself but what detector can use it and for what physics exactly. We can produce much more interesting physics colliding muons instead off protons but this is much more challenging task and will cost more efforts and will cost more.

Also I would say that Scientific value isn’t measured by compactness or cost alone. This is a VC mindset not a scientist pushing boundaries of knowledge.

> Right, defense can and has funded research for its own purposes, and sometimes those purposes can find wider commercial application (like the internet). That's all great, national defense is one of the government's primary purposes.

Well since we are here. I know it is a cliche by now and many people HN doesn't like to be reminded about that but guess that is the most beneficial CERN output ?

> But on the other hand, we have "FCC" being based on mature accelerator technologies, with well-understood cost scaling and detector integration that builds on decades on experience building accelerators. Actually it is much safer option than what you are saying.

The question is not whether wakefield accelerators are "ready" for something on the scale of a supercollider, the question is what is the expected return per dollar spend? From what I can see, there's very, very little we can expect from the energy levels achieved by the next radio frequency supercollider. It's basically "explore the Higgs sector a little better", and that's it, and we're not expecting to find much there. $20B is a high price tag for producing basically nothing new.

I'm saying that if you took $18B of $20B for the supercollider they've been tossing about and invest it into wakefield research, it's very plausible that we could solve all of the problems you describe, and with the $2B left over we could build a wakefield accelerator of comparable energy, and that we'd be better off in that world.

> But the important point is that you are making it binary choice, we can still investigate and work on wakefield accelerators while working on more mature projects.

Investment dollars are finite, therefore it often is a binary choice. You could fund tens of thousands of smaller experiments in domains where we have actual uncertainty for the cost of this one piece of equipment that's good for only a few experiments.

> Also I would say that Scientific value isn’t measured by compactness or cost alone. This is a VC mindset not a scientist pushing boundaries of knowledge.

If you want to expand knowledge faster then you should consider adopting the VC mindset: reduce costs per novel datum gathered. You can run more experiments in more diverse fields and uncover more surprises. Sounds like something a scientist should value frankly.

> I know it is a cliche by now and many people HN doesn't like to be reminded about that but guess that is the most beneficial CERN output ?

Direct military projects arguably haven't been a focus of CERN for 30+ years. They might benefit indirectly, but ask yourself whether the military would have still achieved the outcomes they needed by funding that research directly rather than indirectly in a way that accidentally produced things they needed. The direct funding is what I'm suggesting is well justified, the indirect maybe not so much.

Each of these has a long answer, so I'll pick this one:

>[Chemistry]? Such as? What benefits do you expect to see?

Everything around us is made up of "molecules," assemblages of parts called atoms. Since it's not possible to manipulate the molecules directly in sufficient numbers (one pound of plastic is made of 2000000000000000000000 individual molecules), we have to assemble molecules en masse by subjecting them to processes that cause each step to happen to all of them at once. How does that work?

Let's say you have a molecule. Its structure will have exposed parts, and some bonds will be weaker than others. If you want to replace a part with another part (one step in the assembly of the final product), you might go about it by letting another molecule come along that has a greater affinity to bond with the location of the part you want to replace, and also has a tendency to be in turn itself replaced with the part you want to add. How can you know which molecule to use for this? You could run a computer simulation, apply a rule of thumb, or look it up in a book. In order to write the simulations, deduce the correct rules of thumb and write the books, scientists need to try a lot of combinations of molecules to see what parts swap with what other parts when they're mixed, and then think very hard about what's happening and why it is happening. This practice is known as, "chemistry."

Once a lot of the rules for a certain molecule are mapped out, engineers with an application in mind can go to the library and ask, "what sequence of steps will take me from available molecules to a molecule I can sell in a way that succeeds very often?" This is called, "industrial chemistry." If there was no library and no knowledge in it, industrial chemistry would be impossible. That is the relationship between science and engineering.

All well and good, but not an argument that the science here has to be publicly funded, and that a commercial research enterprise providing this library is not viable. The question was about what benefits come from the public funding, in particular, benefits that simply cannot be provided by a commercial enterprise or alternative approaches. You merely stated this was the case and I'm asking for the reasons.

Even supposing the costs associated with basic research can't be recouped commercially using existing technology, that does not suggest that alternatives are not possible. For instance, in a world in which this library was not created, perhaps the set of talented industrial chemists in our timeline would have studied physics or computer science instead and advanced quantum chemistry simulations that capture some, most or even all of the utility of this empirical approach.

I can look at what you describe and acknowledge that it's useful without automatically accepting that it's a) not commercializable, and b) impossible to work around, which is what you were implying.

[flagged]

> Why are you parading your ignorance of my position as an informed rebuttal?

Nicely put, but I say ignorance because your post was a flurry of questions asking someone else to tell you information about multiple subjects rather than adding substantive information or viewpoint to the conversation.

> Not an argument that these advances would not have been made otherwise, such as by research directly in medical imaging, nor an argument that this was the cheapest way we could have made these advances.

You've asked a question that is impossible to answer, but the reality is that the benefit happened, and it's not the only one. It seems that the system has some merit, although yes, there's no way to prove that there wasn't a "better" straight-line-to-the-answer way to do it. How can you know the straight-line path ahead of time? You can't map the territory without going out there and looking. Basic research in multiple areas, allowing for cross-pollination has done a really good job at that over the years.

> Ditto for mathematics, which for centuries has progressed without direct public funding.

This one really doesn't make sense. Who paid Riemann? Who paid Newton? Universities are not a new thing, and funding them with state money has been there from the start. Even figures perhaps not as strongly associated with universities like John Herschel or Tycho Brahe got their money from the state one way or the other (aristocrats, or given money to advance the knowledge and/or image of the state).

Would you please stop posting flamewar comments to HN? You've done a great deal of it in this thread. It's not what this site is for, and destroys what it is for.

https://news.ycombinator.com/newsguidelines.html

Please don't cross into personal attack, regardless of how wrong someone is or you feel they are. It only makes things worse.

https://news.ycombinator.com/newsguidelines.html

> What benefits do you expect to see?

This is exactly the problem: you're describing a for-profit mindset, and it's exactly why research in the private sector floundering.

If you pretense everything with "what value will this bring" then you've already lost. Research is about finding things out for the sake of it. You don't know if it brings value because you haven't researched it yet. That's what breaking new ground is all about.

> If you pretense everything with "what value will this bring" then you've already lost. Research is about finding things out for the sake of it.

No, it's not. If research never brought any value then we would almost never do it, aside from some weird hobbyists, and particularly not with public funds. Everything has a cost-benefit tradeoff, even publicly funded research, and pretending this isn't so is naive at best.

Research whose costs can be recouped on short time horizons arguably should not have public funding because the economic incentives are sufficient. Exactly where to draw this line is not clear, not only because research returns are unclear but also because publicly funded research diverts talented people from endeavours that would have provided direct economic benefits. This second order effect is not widely appreciated. How can you truly evaluate the opportunity cost of this counterfactual world? Seems virtually impossible in fact, so arguments that research X returns Y with no consideration that you can't evaluate the counterfactual should be viewed with extreme skepticism.

And this doesn't even get into the problem of scope creep. Academics charged with pure research develop the exact mindset you illustrate, where no matter how outlandish the idea, well maybe it will be good for something someday, so why not fund it just in case? This ends up producing a whole lot of nothing, as we've seen in particle physics over the past 30+ years.

wild to come across people on HN who are skeptical about the value of knowing things

"holding my head in my hands" to quote an earlier poster

Wild to come across people on HN who don't seem to understand that "knowing things" has a price tag, that there is no such thing as "no price is too high", that science that can economically justify itself shouldn't be publicly funded and that science that must be publicly funded should have to justify itself to tax payers.

wild to come across people on HN who don't understand that putting economic scrutiny on scientific studies and questioning the public funding of science will both slow the development of yet unknown useful knowledge, as well as slow down economic growth

the perniciousness of the "we need to economically justify the kind of scientific research we are doing" is that plenty of the research we've had that has been economically beneficial was NOT obvious when it was being conducted

by restricting research to programs that may have economic benefit, you restrict yourself to funding things that we pretty much already know, which is a bit more like R&D and less research

to give two examples

1) Gila monster venom - research in the 1990s on Gila Monster Venom formed the building blocks that would become GLP1 medications, which are likely to be some of the best performing medications of all time, as well as have the huge societal benefit of reducing the obesity load on the health system, when this research was being conducted its implications were not known and could have very well been on the chopping block if we were trying to "justify it to taxpayers"

2) CERN - the study of high energy particle physics at CERN is a classic case of "how useful is this knowledge?" It's pretty easy to look at this and wonder how it economically justifies itself. What difference does it make to the tax payer if we discover the Higgs Boson or not? Well, the entire digital economy is down stream of CERN. The internet was partially developed to facilitate the transfer of large quantities of data from colliders like CERN to be analyzed elsewhere in the world. For fuck suck, the world wide web was invented at CERN by Tim Berners Lee. If we didn't invest that money into CERN, or other research institutions, who knows what the web would look like today, and how large the digital economy would be.

Yes, these are just two examples of how research without clear ROI has had economic benefit and justified itself to tax payers. The crux of the issue is we don't know how valuable what we don't know is, and we don't know what branch of science will have the next society altering discovery, so a random walk through scientific research for the sake of knowing things is valuable, because there are undoubtedly things we don't know that will benefit us greatly.

So my argument is in a way like yours, the science does have to justify itself to taxpayers. But the evidence is that the process of science, and knowledge seeking at a high level have justified the funding of science, going study by study to figure out what will have ROI and what won't is a great way to ensure that we discover less and less, leaving more and more stones unturned.

Reply on news.ycombinator.com