Science Is Getting Less Bang for Its Buck | Hacker News Reader

Science Is Getting Less Bang for Its Buck

theatlantic.com

138 pointsdsr12148 comments

This seems like the obvious result of many of the unhealthy forces against science today: replacing professorship with part-timers, rewarding publishing quantity over quality, pressure to promise a quick monetizable result over in-depth basic research.

I don't buy the "low-handing fruit" argument. I think it vastly overstates the fraction of scientific knowledge that we possess. Contemporary scientific research is in a state of crisis: unhealthy and dysfunctional. Until we pay the most talented researchers to spend decades on in-depth study instead of spending pennies on segmented, p-hacked, post-hoc conclusion garbage papers, major discoveries will continue to disappear.

From what I have seen it is quite simple.

1) There is this pseudo-scientific methodology called NHST (null hypothesis significance testing) devised pretty much on accident in the 1940-50s that tricks people into thinking they have discovered something interesting.

2) It is so much easier to generate these "significant" results than to study a phenomenon and come up with a model capable of making a precise prediction, then have multiple people/groups carefully collect data to check that prediction.

3) People willing to do the NHST thing can publish papers orders of magnitude faster than people not willing to.

4) Number of publications is the primary metric of success in academia.

5) People willing to do the NHST thing slowly take over the field as those who want to do real science get pushed out, retire, or leave for greener pastures.

6) Eventually the field is filled with people generating worthless papers that cannot be replicated in principle and often contradict each other.

- a) Because all that matters is you got a significant result published in a peer reviewed journal

- b) The discovery has already been made, the statistics said so, why would anyone need to replicate it?

And to preempt a common confusion. This has nothing to do with bayesian vs frequentist, testing a strawman null hypothesis can be done using bayesian math just as well as frequentist.

Some decent intros to this problem (yes, it was pointed out long ago but there is no stopping the need to publish apparently):

Meehl 1967: https://meehl.dl.umn.edu/sites/g/files/pua1696/f/074theoryte...

Gigerenzer 2004: https://library.mpib-berlin.mpg.de/ft/gg/GG_Mindless_2004.pd...

First time I've heard that NHST could be a source of issues with science. I've heard about p-hacking issues or frequentist statistics getting the blame, but NHST being a root cause seems plausible in retrospect.

Thanks!

Great to hear you are interested in this problem. It is a huge one that goes mostly ignored.

Afaict, the p-hacking/etc people usually focus on are red herrings caused by the incompatibility of NHST and scientific research. Eg, it is good thing for a scientist to analyze the data in a bunch of different ways. It is definitely not something that should be punished.

Calculating p-values falls under hypothesis testing

1) Hypothesis testing, significance testing, and NHST are all different things

2) P-values seem ok as a way to check how well your model fits the data. The problem is testing a strawman model/hypothesis.

The article seems to be mostly about “hard” sciences, where ”discoveries” are not about getting some p<0.05

Unfortunately, I can see its infection of physics is in progress. And testing 0.05 vs .000005 makes no difference if you are testing a model other than the one you believe. It just makes it more expensive and time consuming to draw an inappropriate conclusion from the p-value.

On the other hand, thankfully, in physics areas they usually also test actual predictions of their models. But that is part of the process you can see happened in psych, then medicine. If you sample the literature going back to 1900 or so:

Stage 1: Science - Test the predictions of your models

Stage 2: Hybrid - Test the predictions of your models and also strawman null/background models

Stage 3: Pseudoscience - Test only strawman null/background models

My guess is during stage 2 there are still people untrained in NHST in important positions who hold back the wave of BS by demanding actual science (testing predictions of models someone thinks may work). Once they retire you get stage 3.

One day I want to write this all up, its really quite interesting to read the old psych and medicine papers where people were coming up with mathematical laws, etc then the slow progression to "is there a significant correlation/difference?".

In my experience as a researcher in physics (obviously anecdotal to my field and not generalizable across physics) all we do are variations on Stage 1 - we either have a model, a prediction and an experiment that confirms it (these go to the best journals), or we only have a model + prediction (and couldn't do the experiment) or an experiment (but not a model for it), so we report those in the hopes that someone else will fill in the missing half. Since I mostly do theoretical soft matter work, having a model with an interesting but hard to verify prediction is the common case (which is far from useless, as experimental techniques are advancing incredibly quickly). Either way, if I brought up p-values I'd be unceremoniously laughed out of the room in my department. If I wrote up a prediction that's not just verifiable in principle but verifiable with existing (albeit sometimes new, rare or hard to implement) experimental methods, I wouldn't get published, unless it was some kind of a novel theoretical technique with further applications in that regard.

I'm not saying your assertion is wrong - physics is wildly different across different fields, and I don't pretend to have any meaningful overview of all of them. I can only report that I see no signs of this supposed encroachment of pseudoscientific statistics in my physics work or the literature I read professionally - I hope it stays that way.

I'm speaking from what I saw in the LIGO and CERN (Higg's Boson) papers. They spend way too much time on rejecting some questionable model of the background rather than testing actual predictions.

From media interviews, I'll also say on the theorist side of it they seem to harbor all the usual misconceptions about p-values as seen in psych and medicine. But they do eventually fit an actual prediction to the data, like a model of black holes colliding, or whatever. However, I don't personally know how many degrees of freedom they have in these fits...

LIGO and the Higgs boson example are both extreme outliers in physics in terms of scale, how much statistics they necessarily involve, the scrutiny and publicity they get and almost every other property. I'd also note that scientists trying to oversimplify things for a media interview almost always screw up somewhere. I wouldn't draw any conclusions about physics in general from either their papers or (especially) interviews.

I work in a hard science and, while it may not look as egregious as p-hacking, my field is riddled with misapplication of statistics and problems that generally follow the trend of your parent comment.

From my background in the chemical sciences, p-hacking and significance testing in general simply are not relevant. It's not that people aren't necessarily sloppy or disingenuous, I mean literally that statistical testing is hardly ever done by anyone in any form even in most contemporary practice and is virtually nonexistent historically in my field and those adjacent to it. That's slowly starting to change as individual fields transition towards big data-y approaches, but 'p-hacking' as a phenomenon is something I would associate as an idiosyncrasy primarily of biomedicine and the social sciences; I do not think it has generalized explanatory power.

I think there are consistencies in your second paragraph. You don't buy the low-hanging fruit argument, however you then state that we need to spend decades on in-depth studies. Well, those are exactly the opposite of low-hanging fruit, those are extremely hard-to-reach and abstract fruits that the vast majority of the scientists of the past would have been unable to grasp, simply because they did not have access to all the foundations we do today. You can't attack Poincare's conjecture without basic arithmetic, nor can you understand antibiotics without a knowledge of infection and germ theory.

Look at some of the most sea-changing discoveries in the history of science and you will see it was not the result of decades of laborious and thoughtful calculation and musing. Penicillin was discovered by coincidence or luck; the production of rubber for tires was also lucky. How about chance astronomical observations like an eclipse, whereby a great deal can be learned in a fleeting moment? How about displacement of water, or the pythagorean theorem, or Michael Faraday's cheap and quick experiments into electricity. These are not tours de force, by and large they are little fun divertissements, interesting observations with great consequences.

I think there's no doubt that MANY of the low hanging fruits have already been plucked. It's the same way in many other fields, such as music, art, computer science, or anything else. People naturally discover the simplest things first, and then build upon those, with later discoveries taking far more investment of time/learning to even be capable of understanding, let alone discovering.

I suspect to some extent, performance metrics of research grants steer scientists away from low hanging fruit in unknown areas, and into well known paths that can be justified in grant proposals. Most labs I would suspect have little time to just play around, or even stop and contemplate anomalous behavior like you're describing.

Agreed, BUT:

Sorry, academia is broken, and the professors are a big part of the problem. We don't need more of them, or full timers. Let's stop teaching lectures to undergrads in medieval style in 500 person lecture halls, and instead get them engaged in research as assistants right away.

We need more science and less "publish or perish" politicians posing as professors engaged in grant grabbing.

I have tremendous resentment against the lazy turds I had to sit and listen to as they wasted my money not teaching and not researching either. Tweed jacket welfare recipients if you ask me.

> Let's stop teaching lectures to undergrads in medieval style in 500 person lecture halls

Although I agree with this sentiment, you have to realize that the alternatives are either a) worse, or b) don't scale.

I prefer the ones that don't scale (small number of students in a class who get real continuous feedback and fully participate in the lecture). It works really well, but this model is enormously expensive compared to the 500 person lecture hall.

500 person lecture halls suck, but what's the scale-able, better alternative?

MOOCs aren't the answer -- that's just the old 500 person lecture hall model with a pause button. Woohoo. Plus, it doesn't actually solve the real problem, which is that you need the humans running the course anyway. For a well-oiled machine of a course, maybe 10% of the actual time spent on teaching is spent in the classroom. And for non-well-oiled courses, creating the MOOC and keeping it up to date is a hell of a lot of up-front cost, and with a lot of risk.

> and instead get them engaged in research as assistants right away.

I can tell that you haven't worked with lots of undergraduate students on research projects.

Most undergraduate students aren't ready to participate in research. Of those that are intellectually prepared, few have the grit/persistence required. Of those with the intellectual preparation and persistence, fewer still have the imaginative faculties to actually do research (as opposed to e.g. acting as a tech).

This is true even in very applied areas. You'd be surprised how few CS undergraduates would succeed at implementing something as simple as an experimental change to a kernel or a tweak to an ML algorithm.

At some point you simplify the research problem enough that it's indistinguishable from a course project or homework assignment. In fact, that's even where some course projects and homework assignments come from. But at that point, you're not "involving students in research", unless "doing homework or programming assignments" = "research" that is.

> We need more science and less "publish or perish" politicians posing as professors engaged in grant grabbing.

At times I have certainly felt this way about academia! But I could say similar rude things about every profession.

> Tweed jacket welfare recipients if you ask me.

There are ineffective people in every field, and most of them manage to stay more-or-less employed throughout their working years.

The two of you must have had a different experience than I. At least for my classes, fundamental stuff that hasn't changed in decades was taught in big lecture halls- single variable calculus, physics & chemistry 101, etc. Then as you advanced in your discipline, class sizes shrunk, all the way down to 5-10 pupils in some of the most specialized classes, which was where the close interaction with the teacher was most valuable.

Most undergraduate students aren't ready to participate in research. Of those that are intellectually prepared, few have the grit/persistence required.

More indirectly, few have any kind of serious investment in the research. In college I volunteered to help on a plant study. I was assigned to meticulously clean dirt from dried plant roots and weight each plant. For like four hours at a time. Thousands of specimens. It needed to be done, but I had no investment in the project aside from general excitement, so I just didn't care enough to suffer the drudgery week after week.

Your experience matches mine in CS; the large lecture halls were only used by a handful of early required courses that bottlenecked the students, a basic prerequisite for a variety of other courses all done in smaller classrooms.

Yes, CS is a completely different beast for two reasons.

First, our intro-level courses are fairly stable but nowhere near as stable as Calculus or Physics intros. If you don't change your intro for a decade, it starts to show. Whereas I'm pretty sure my children will take more-or-less the same calc course that I took, and my calc book was pretty indistinguishable from my mother's. To really emphasize the point, my mother's CS course was done on punch cards...

And it's not just content. Student preparedness has been changing a lot lately. We reached "peak physics/calc" in high schools a long time ago (hell, my grandpa took calc in high school). But students showing up at university knowing how to program is relatively new for most institutions.

Third, enrollments have been increasing exponentially (yus, really exponentially) year over year (yes, every year) for close to a decade now (yes, a decade).

So. Yes, CS ed is very different from the more stable STEM fields, especially recently.

> Most undergraduate students aren't ready to participate in research.

Part of the reason may be that most undergraduate students didn't enroll to university to do research; they came for the degree they need to have a shot at getting better-paying jobs. Reintroducing vocational schools and making university research-focused again could help increase research-readiness of undergraduates.

I would argue that if you come to a university for science or engineering, you come to do research. whether that is through working in a lab or internships, extracurricular work is the most important factor in getting the high-paying job in your field.

CS may be different, but all the recruiters I have talked to, and all of the people doing actual hiring don't care about the degree, they care about projects, and whether you are going into academia or industry, your undergraduate degree is not nearly enough to land a job.

Your post is a perfect example of the utterly massive entitlement that young people have today. I learned a TON in my undergrad CS classes, and that only began to scratch the surface of what I needed for grad school. That you didn't learn anything is a reflection of you, not your professors.

Life takes work, you are in no position to just jump into research. You don't have the background, the skills, and clearly not the maturity either. You are not experienced, and even when you graduate you will not be experienced. While the economics of development today might mean that you will get a high starting salary, you are still a junior with a lot to learn.

"That you didn't learn anything is a reflection of you, not your professors."

That's a little insulting, isn't it?

I'm glad you learned a lot in your CS program, but going to university was one of the biggest mistakes I ever made. There were a few good courses, and a few good professors, but the vast majority of what I learned was out of school - weekends and summers with textbooks and lecture videos of my own choosing.

> We don't need more of them

> 500 person lecture halls

I love evidence-based pedagogies like active learning, which are demonstrably better for most students than "medieval style" lectures. I would like to be able to teach that way all the time, but if you want professors to actually be able to interact meaningfully with each student, you need small classes. If you want small classes, you need more professors. So yes, we do need more of them, in that case.

You can't run an (experimental) research program without a lot of money and even then most academic labs are run off a shoestring budget (reagents, lab space, instrumentation + maintenance/electricity for such, access to university and regional instrumentation e.g. synchrotrons, high field NMR, supercomputers, paying your full-time research staff i.e. mostly grad students and post-docs a working class wage)- even at elite institutions there's a heavy reliance on used and archaic equipment (hand-me-downs from the 80s and 90s). 'Grant grabbing' is strictly necessary for having the money to run a research program beyond something utterly barely bones (and likely fatally limited in scope in many fields as a result) on both instrumental/facilities access and output- and you can't get tenure and thus remain employed as an independent researcher without grants anywhere big anyway (grants are proof of promise/success and are necessary for engaging in ambitious projects- and the demand for academic tenure track positions is quite saturated so if you can't manage they can probably find someone else who can). Most faculty I know work (primarily writing grants and publication and to varying extents chewing through data and models with students) on weekends (and expect their personnel to do the same), especially ones still vying for tenure track (who often still work in the lab directly as well). The pay and hours are pretty poor relative to industry (abysmal if you consider time spent as a post-doc) and the stress is quite high- it's not something done in this day and age for the sake of wanting an easy road through life.

Essentially everyone planning to go into the basic sciences as their primary specialty needs to go and work in a laboratory anyway to some extent to get into grad school (my entire cohort did one to three years of lab work in undergraduate); and those who don't in my discipline are typically pre-meds (but who are still rewarded in apps for doing research so plenty of 'em did so) who have to do a lot of hospital volunteering and whatnot for their admissions process. However, undergraduates have less time to work, less experience/knowledge due to being partially through coursework, and are often less motivated because research isn't do-or-die for them like grad students and post-docs- taking them on can have opportunity costs for laboratories even though their labor is typically cheap or free.

I think you need to ask the question about the unhealthy forces in science today. Are these forces breaking science? Or Is the investment model (grants) creating these forces? I think it is the latter. That's the whole point of the article. We need to rethink how we allocate dollars.

Any ideas how to fight with grantoeaters?

> Until we pay the most talented researchers

It's not even that. One of the contributing factors is that we are no longer hiring for talent or merit. If you're a male child interested in science and in high school right now you've been through 8 years telling you that you should not go into science because they need more women of colour who follow the right religions and have the right sexual orientation. You're constantly bombarded with "SCIENCE DOES NOT WANT MEN".

And sure, we can say "well if he had the necessary grit, this would not discourage him". No. No amount of grit will get you in when governments are shaping science funding to punish having male staff. No amount of grit will get you in when indoctrination training is mandatory for the selection committees. No amount of grit will get you in when the first filter in a hiring process is to remove all males applicants.

I'm sorry but science has to go through a decline in the West for a bit. We still have some remnants to be worked through the system, but the next couple generations of Western scientists are going to be generally pretty low quality. On the bright side, science is not exclusively a Western tradition at this point. Other cultures that at least pretend to still focus on merit and talent will carry the scientific tradition forward.

> If you're a male child interested in science and in high school right now you've been through 8 years telling you that you should not go into science because they need more women of colour who follow the right religions and have the right sexual orientation. You're constantly bombarded with "SCIENCE DOES NOT WANT MEN".

Do you have any examples of teens / young adults being discouraged from going into science because they are white and male? There are certainly some (I'd say good!) programs to encourage women and minorities, but that's quite a different thing from men being actively discouraged.

Scientific fields are still significantly more male than female, even among recent graduates.

> Scientific fields are still significantly more male than female, even among recent graduates.

Why "still"? Majority of mans are genetically selected to be adventurers, while minority of woman are adventurers. It's impossible to change genetics with advertising, so we will never have 50/50% split, until human genome will change. Why not just accept natural distribution?

I'm not convinced that joining the scientific community is an adventure or that this genetic difference exists. The differences can be explained without using genetics.

For example, there is a popularly held opinion that girls are bad at math. Girls who believe they are bad at math might try less in math courses. Girls that receive lower grades in math from trying less might be intimidated by the math requirements of a science program.

It's not an adventure, but it is a high-risk with a slim chance of a very high payoff endeavor, and men are typically more drawn to that than women, for example there his a huge gender disparity in fishing the Alaskan fisheries or base jumping.

Offtopic: Can you explain to non-native English speaker, please, why "adventure" does not match "a high-risk with a slim chance of a very high payoff endeavor"? I saw "adventure in science" few times, so I assumed that someone who does this "adventure in science" can be named "adventurer". What is wrong in my conclusion?

I wouldn't consider a career in science to be high risk. It's often easy to transfer from academic science to private engineering. Science degrees holders have respectable earnings on average.

The argument I like most is that young ladies are just smarter, and don't go as much to a field that promises low pay, long hours, and doubtful prospects; they choose careers more wisely.

How does that one square with the fact that the median income for women is lower than the median income for men while the median income for all groups correlates positively with intelligence?

Seems like that kills the hypothesis that women are making better career decisions across the board optimizing for pay.

Pay is not the only issue when choosing a job. Job Stability, stress and work life balance are also huge factors rational actors should consider.

Science does not simply pay poorly, the job also sucks.

So does nursing, but it is heavily female-dominated.

“Registered nurses (RNs) made a median salary of $68,450 in 2016.” That’s not really bad pay considering you can get an associate degree which only takes 2 years, and you make above median US wages.

Being a scientist requires vastly more education, and long periods of low pay. I actually know a programmer that switched to nursing and he seems reasonably content.

The reason why science sucks goes beyond pay. It's the repeated failure, long-ass hours (I once worked for 150 days straight, pulling >100 hour weeks), only to find out that in the end playing politics was more important than doing a good job at the work.

A life tip: Always 50/50. Talent is great, politics helps you direct it to where it will maximize impact. If nobody thinks what your working on is important you may be an unrecognized genius, but its also perfectly possible you are on the wrong track.

If you can't convince anyone what your doing is important it should make you stop and reconsider your efforts.

I'm in a position where I succeed by not playing politics. Since i'm not playing politics (which others in the company are doing), i am spending my time doing development work. I am now building a team, and my only concern is to deliver results on time. My CEO thinks I'm competent and has expressed interest in helping fund me a startup, when the time comes. I'm not going to say that I didn't strategize the a politics "meta game" (picking a place to work where I would be visible, choosing to work on projects where I can have an impact) but the amount of time or effort I am spending on it is vanishingly small.

In any case my point about science is that playing politics is MORE important than delivering quality results. In my lab as a grad student, there was a grad student who delivered extremely sketchy data and then won the grad student of the year award and now he's a tenured professor at a top 50 US research institute.

> Do you have any examples of teens / young adults being discouraged from going into science because they are white and male?

I know a post-doc who was told to hire at least one female undergrad, even if there's a better man.

The crazy thing is that in some companies/jobs you could be fired for posting this.

In most (US) states, you can be fired for anything not specifically protected. What makes this case anymore interesting?

As a scientist, definitely. It would be evidence that you've lost your grip on reality and are entirely a creature of reactionary twitter rage, and trusting your judgment is probably not a great idea.

>"As a scientist, definitely ... trusting your judgment is probably not a great idea."

You should never trust someones judgement when it comes to science. That is practically the idea that started it: https://en.wikipedia.org/wiki/Nullius_in_verba

I'm in grad school at the moment, and it sounds to me like you're describing another planet from the one I live on. My entire lab has one woman in it, and nobody's come around to punish us or anything. Hell, I'm not just male, I'm white and male. I'm not even gay! My ethnicity (Jews) are over-represented in science.

What sort of data have you collected, or received in summarized form, to come to these conclusions of yours?

Biology as a field ranges from majority female to 1:1 parity in the modern postgraduate cohorts I have seen; the physical and logical sciences at the postgraduate maintain large male imbalances and notably seem to do so across all western countries regardless of social ideology from the figures I have seen across various sources. It is true in my observation that gender-specific encouragements are consistently emphasized at the university level even oddly in scenarios where parity has been reached or exceeded.

One must be delusional to believe something like this. I work in science and it is still extremely white and male dominated. All the students I teach are also mostly white and male and are very much enthusiastic about continuing in science. If anything, the people who are typically less enthusiastic are the non-white/non-male students. And understandably so, because - despite some minimal effort to change in recent times - the environment just isn't all that welcoming to them.

I might quibble with the white part here. I see many people of Asian and Indian descent in STEM fields.

This is profoundly contrary to my experience mentoring young students, the approach of my teacher friends and acquaintances, the experiences of family members. When this topic does come up -- and it does -- it's done specifically with the notion of normalizing the idea that people of all backgrounds can, should and will be scientists and mathematicians. It's entirely unproductive to the project of establishing a more welcoming future by, as you say, declaring "SCIENCE DOES NOT WANT MEN". The loss of an exclusive status does not necessarily imply a discriminatory regime, though you've clearly framed things that way for yourself. Governments aren't "shaping science funding to punish having male staff" they're preferentially awarding funding to labs that are not exclusively or predominately male. I know enough excellent scientists who've been chased out of labs on account of sexual harassment to know that there's _plenty_ of reasons for a lab to be exclusively male besides "merit and talent". In fact, I argue that if you allow systematic inequalities to shape a field you're specifically undercutting people who might otherwise have succeed because of their talents and drive.

> I'm sorry but science has to go through a decline in the West for a bit.

The real issue, as others have pointed out in this thread, is the need for scientists to chase funding at all times, whether by seeking out grants or side-gigs in graduate school to supplement a meager stipend. The issue is emphatically not that your favored subset of humanity isn't esteemed like you think proper.

Fix your heart.

You conflate "losing your exclusive status" with the push to equalize representation in workplaces.

There isn't an infinite amount of jobs for any particular field or workplace at any one time, if you pursue a campaign of equality of roles going to men/women in a male dominated field, your going to have to reject some men who otherwise would have qualified.

> You conflate "losing your exclusive status" with the push to equalize representation in workplaces.

I am not. You have done that for me and incorrectly.

Firstly, It's absolutely appropriate to inform children that people of all backgrounds can, should and will become scientists and mathematicians and to encourage children with these interests to pursue them. Separately, it's entirely appropriate to be suspicious of behaviour as a society that artificially filters some subset of humanity out of a field they would otherwise be able to contribute to owning to individual talent and ambition.

> There isn't an infinite amount of jobs for any particular field or workplace at any one time, if you pursue a campaign of equality of roles going to men/women in a male dominated field, your going to have to reject some men who otherwise would have qualified.

This viciously cuts the other way too. If you leave a systematic inequality in place you prop up otherwise untalented schlubs who edged out people of merit. In my own field, software, I've seen talented software engineers leave companies that needed them because of continued, unwanted and rebuffed sexual advances by some EXTREMELY mediocre software engineers. These dullards then merrily added negative value alone in their pit of reinforced homogeneity. More effective HR would help in this particular example but, hey, so too would having a shop that wasn't almost entirely of homogeneous background in the first place.

So what if we do inform kids of all backgrounds they can do whatever they want and we still end up with 95% of programmers being white males? That would be ok and would require no more effort to correct in your book?

To your second point, then my recommendation to employers is that they should hire only these great women who have been sexually harassed. It seems clear whoever does this will have a large competitive advantage.

> When this topic does come up -- and it does -- it's done specifically with the notion of normalizing the idea that people of all backgrounds can, should and will be scientists and mathematicians.

This line of thought always confuses me a bit. Unfortunately the image of those that question such things is so poor it's rather discouraging to even bring up. I should really try to stress that I'm not trying to say how things should be. I think these issues are more complex than most anybody gives them credit for, and everybody has their own desired outcomes, timelines, and acceptable consequences. All I'm trying to do is explain perceptions, really.

The main line of thought that tends to result in all the focus on women in science and programs to help women succeed in science seems to include the idea that the societal conception that scientists are male pushes females away from science. There are of course other issues like you brought up that impact retention of female scientists, but much of the science-specific efforts aren't focused on retention but rather initial interest. That's fine, and a decent enough issue to go after.

However, what many people seem to have trouble seeing is how these programs can reasonably provoke reactions like the one you responded to using the very same logic that promoted the creation of those programs in the first place. If you can see how media depictions and general expectations so severely shunt women away from science, is it really so hard to understand how the focus and programs promoting women in science may be having a negative impact on men wishing to go into science? The pressures aren't exactly the same, what with one being sort of a general thing while the other is more purposeful and focused. Who knows how they balance relative to each other, and people won't even agree on where that balance should be. However, I do understand how people can take the more purposeful one more personally. You have specifically putting out these influences that you can latch onto, while you can't really do that with a vague general feeling.

So, hell if I know how to even determine who is right here, though I suspect nobody is or ever will be. But I'd really like to see more thought put into how things effect everybody and some more understanding that people may have justifiable negative responses to things you personally consider to be positive. And neither of you has to be wrong. Or fix themselves.

> If you can see how media depictions and general expectations so severely shunt women away from science, is it really so hard to understand how the focus and programs promoting women in science may be having a negative impact on men wishing to go into science?

Hi, yes it is hard to understand. If you are a child and you see an example of people of all backgrounds working together to do science, including people that look and behave like you do, and your take away from this is "Welp, I clearly am not welcome here." then you need mentoring help from an adult to prop up your self-esteem. If you are an adult and you see an example of people of all backgrounds working together to do science, including people that look and behave like you do, and your take away from this is "OOOOH I AM NOW OPPRESSED" then you are either pitiable for how far off the rails you've gone or are now pushing a motive distinct from leading your people out of the inhospitable land where talent and ambition are primary.

> So, hell if I know how to even determine who is right here, though I suspect nobody is or ever will be

Nobody is... nobody is or ever will be right? Listen, here's my motivating principles. I believe in the equality of opportunity. I believe we live and die and have a responsibility to leave the world more just and free than we found it. I believe we're working off partial information and even with the best information any attempt to leave the world more just and free will, inevitably, do a little bit of the opposite. That's why we should leave wiggle room for future generations. Now, access to opportunity in this particular discussion is limited owing to a systematic preference for a subset of humanity, shutting or pushing out the otherwise talented and ambitious from science and mathematics. It is entirely possible, though I argue not probable, that by pushing for a world in which all people from all backgrounds can do science if they have a talent for it is going to somehow have dystopic results. I know for a fact that our current approach of letting in-group dynamics and selective representation act as a filter of otherwise talented and ambitious people is actively dystopic. So, move along the gradient. Move from "absolutely very bad" to "very, very unlikely to be just as bad but also in a different way".

In passing, I will also note that your comment and grandparent weight a hypothetical bad equally with a demonstrable bad. I reject this weighting.

> And neither of you has to be wrong.

Seems like pretty dichotomous positions to me.

> Or fix themselves.

The grandparent should absolutely fix their heart.

> Hi, yes it is hard to understand. If you are a child and you see an example of people of all backgrounds working together to do science, including people that look and behave like you do, and your take away from this is "Welp, I clearly am not welcome here." then you need mentoring help from an adult to prop up your self-esteem.

You are grossly misrepresenting the problem here. I grew up in the 90's in a more progressive country than the US and there were plenty of science events that I thought would be cool to attend which were girls only (almost all of them actually, not many events for kids here). My spot on a science competition team was given to a girl since the competition required at least one girl on each team. Also it seems like people taking stock photos loves including girls and women in them. Every time I got to the next level I was surprised over how there were so few girls there, nothing I learned stated that women were not common in STEM as I grew up.

When you go through all of that, is it really that strange you start believing that the odds are stacked against you? And is it strange that when every time you hear "equality" they go on to say "boys need to take a step back in favor of girls", that you read it as boys need not apply? So when almost every job posting is asking for "equality", what you read is "boys need not apply" because that was what it always meant before. I don't care what the exact definition of these terms means, in practice this is what boys hear when they grow up in such an environment. And when you get ghosted by such job postings explicitly encouraging equal representation, then it is very easy to assume that they ghosted you due to your gender.

Anyhow, all of this has lead to nobody taking the shortage of women in STEM seriously, I mean this has been going on for around 30 years and they still can't attract girls to these programs...

> If you are an adult and you see an example of people of all backgrounds working together to do science, including people that look and behave like you do, and your take away from this is "OOOOH I AM NOW OPPRESSED"

That... is not at all what I said. At all. I am confused how you got that from what I said. I said that women have the problem of not seeing women in science, while men have the problem of seeing public interest and support for only women in science. Nowhere did I suggest seeing women in science would be detrimental to men. The reason I bring them up together is because one is a result of attempting to fix the other.

So, to try again. Remember I am not talking about end results, or ideals or where the world should be, or any of that. I'm talking about how things make people feel. You accept that not seeing women in science makes women feel they are unwelcome. Meanwhile we have people repeatedly stressing the importance of women in science, and seeing many groups formed explicitly to promote women in science. Men are not seeing any of this directed towards them. The parallel here is pretty trivial. Men can see this and feel unimportant and like they are not welcome.

Is that an acceptable or even necessary consequence to achieve a more egalitarian society? Maybe? I don't know. But either way it's rather callous to brush off these feelings as merely a result of people being "broken" and "wrong."

> Nobody is... nobody is or ever will be right?

Basically? I don't even know how to define "right" in a meaningful way here without the ability to run world simulations or predict the future. Your motivating principles are all great and I agree with them. The differences emerge in the details of how to act on them. For example, some people view the goal as exact equal representation in all fields, presumably with respect to population demographics though I don't hear that often said explicitly. Other people think the goal should be for each person to have equal opportunities without requiring exact proportional representation. Some people think the time limit on reaching the goal, whatever that goal is, is immediate, while others are fine seeing this take a more natural (and lengthy) course. Some people are more okay with the ends justifying the means than others.

These are all significant details that people can argue about at length while all wanting to see the exact same outcome. I want the outcome you do. That does not mean I need to agree with you about how to get there.

> In passing, I will also note that your comment and grandparent weight a hypothetical bad equally with a demonstrable bad. I reject this weighting.

I'm not sure how you come to the conclusion one is hypothetical while the other is demonstrable. We're seeing men react poorly to the current actions, and not always in appropriate ways. Considering I'm arguing largely about how people are justified in feeling a certain way, I'm not sure what else you need here before you consider their feelings to be sufficiently demonstrated.

> Seems like pretty dichotomous positions to me.

Hopefully by now you see how they are not. You are largely saying that certain actions are good things because they solve problem X. Grandparent is saying those actions are also causing another problem Y. Actions need not be purely positive or negative.

> The grandparent should absolutely fix their heart.

You're really getting needlessly personal and as I keep trying to show, are basically telling them their feelings are wrong because there are other problems in the world. It's really rather condescending and makes it harder to take your argument in good faith.

I have an 8th grader in California public schools, and this is just not at all the case here. There is a general STEM push that is sweeping many of the kids up. There is occasional special note of girls who are good at these subjects, but it doesn't seem to be at the expense of boys.

The notion of "following the right religions" is even more out of touch with reality. At least in California. Are you seriously suggesting that, say, Muslim children are being specifically encouraged to go into science, and (say) Catholic kids are discouraged?

> "There is a general STEM push that is sweeping many of the kids up."

This is awful for science though. It is much better for it to be thought of a "nerdy" thing so people who aren't really interested in it stay away.

EDIT:

I'd like to hear from a downvoter who has tried to convince someone who actually doesnt care too much about science that they should test their own hypothesis rather than a strawman hypothesis when the latter gets them more rewards. It is infuriating, and you do not want them informing doctors/etc on how to advise you:

"We are quite in danger of sending highly trained and highly intelligent young men out into the world with tables of erroneous numbers under their arms, and with a dense fog in the place where their brains ought to be. In this century, of course, they will be working on guided missiles and advising the medical profession on the control of disease, and there is no limit to the extent to which they could impede every sort of national effort." Fisher, R N (1958). "The Nature of Probability". Centennial Review. 2: 261–274. http://www.york.ac.uk/depts/maths/histstat/fisher272.pdf

What? Get this sexist crap out of here. Do you seriously think that any educational programs are rejecting men today? Have you walked into any STEM department anywhere across the country? You can count the number of women on your hands. There is no discrimination against men in STEM.

MY university said they aimed to get 50:50 gender representation of professors within a few years from the current 80:20.

Now, you could achieve that in 2 ways - fire a bunch of old men, or hire mostly women for new jobs.

As a young man, I'd much rather see the first and I think it's fairer... however there's tenure. That means the only way for them to achieve their goal is to hold down young men, who are made to pay for the 'crimes' of the old

I would like to see some sources for those claims.

I'm not sure where to begin on this, so I'll resort to a brain dump.

The Nobel prize has become steadily more political. More senior researchers in the field have had more time to accrue political power, so expect the awards to reach further back in time.

Research productivity has dropped, though. Today a professor in biology has a more than full time job just to get money for their lab. So you go through gradschool, go through postdoc, and when you've got all this training, you stop doing science and become a grant writer. If you don't bring in money, you don't have a lab.

For the postdocs and grad students, they live very insecurely. Their pay is poor, they live knowing they will probably relocate somewhere entirely different in three or four years, so there's no point having links into the local community. They work for someone has no training in management and has no time to deal with them if the money is going to flow. And there are still lots of places where the postdocs aren't considered full time employees of the university and don't have, say, health insurance. These are hardly the conditions that you can expect to produce good work.

Fields are mined down to the last details rather than looking at unexplored areas nearby. There are many reasons for this. As a grad student you work on what your advisor works on. As a postdoc, you work on what your lab works on. As a professor, you work on what you worked on as a postdoc and it requires multiple years and lots of luck to reorient a lab. You need a community to provide evidence of your competence in order to advance at each stage and get tenure, which you can't easily get if you wander out of an established field. Nor is there any training on how to look effectively find things to work on.

Is this only a US and European problem? when I was in college in the oughts their was a popular trend in Physics PhDs to move to china following there Grad program as China provided a block grant to them to start a new lab.

The pattern whereby professors have to apply for small grants continuously to keep a lab funded and operated is an inordinate waste of time vs. a block grant with periodic evaluation.

Keep in mind by small grant we mean 100-200k towards a million.

Which is a small grant. You would be hard pressed to run a five year software project on a million dollars, much less a project that requires real laboratory overhead and materials.

> Is this only a US and European problem? when I was in college in the oughts their was a popular trend in Physics PhDs to move to china following there Grad program as China provided a block grant to them to start a new lab.

I don't know the trend there. In biology there was a trend in the late 2000's of Singapore recruiting professors to come over. Most of them ended up moving back west, saying that they couldn't find students that could work independently.

> The pattern whereby professors have to apply for small grants continuously to keep a lab funded and operated is an inordinate waste of time vs. a block grant with periodic evaluation.

The block grant with periodic evaluation is exactly how Howard Hughes Medical Institute and the NIH intramural researchers work.

Interesting. A lot of this is because of imperfect information and transaction costs. I wonder if firms running R&D choose a different route and if that works. Perhaps even random resource allocation would do a better job since it removes the incentive to falsify though it does add an incentive to participate in bad faith.

Well said.

Any thoughts on researchers like Piotr Wozniak (Supermemo founder) who gave up on publishing and academia?

"I no longer care about publishing in peer review journals. I don't perish if I do not publish. Some of my old publications from the 1990s are listed here. I consider peer review to be a constraint on creativity and dissemination of inspiration (see: Problem with peer review)."[1]

[1] http://supermemo.guru/wiki/Piotr_Wozniak

edit: here is a specific article of his on peer review:

http://supermemo.guru/wiki/Problem_with_peer_review

Its too bad I can't find somewhere he is publishing his raw data, but besides that I appreciate his ability to operate like he does.

Quickly scanning comments I do not think anyone else has brought up: administrative bloat.

More money is being spent on science, but is more money actually making it through the administrative bloat encumbering most institutions to the actual performance of research?

Anecdotally, I have a colleague who has received funding from the NSF and the amount of regulations and paperwork and various travel and meeting-related obligations related to the funding soak up so much of the actual dollar amount supplied. (You have to use your funding dollars to satisfy the various required meetings, travel, and paperwork-filling.) The constraints are so ridiculous that satisfying them consumes nearly all the resources the NSF provided, and the little that remains is actually not sufficient to perform the research with. Worse, he has now wasted months of his time satisfying various oversight requirements administrated by both the NSF and the research institution he works in, leaving him an unreasonably small amount of time to actually achieve any significant progress on his work. Once this round of funding dries up, he will be left with no choice but to repeat the process in order to secure some more funding to continue to barely make progress on his stated research goal.

If I had to make up a number to describe the dollar efficiency of research funding, in some cases I might assert it is negative: Not only is it just being soaked up by self-serving, efficiency-draining administrative requirements, it literally destroys the most valuable resource (time!), leaving the researcher with none to actually engage in their subject matter of expertise.

it's huge. Similar in an european ERC-funded lab: the PI is constantly traveling, there is little oversight of the work let alone actual scientific output. It feels like a large portion of the funding is designed to keep a lot of people people busy doing nothing.

At least the US government treats it like a "jobs program", just like everything else.

Not to mention the fact that the academic institution usually takes a hefty "tax" on any grants awarded. Where I work, they take around 30-40% I believe.

I think overhead rates at most big state schools are around 50%, and I believe they are higher at a lot of elite private universities. This is the link I got when I googled "Yale overhead rate":

https://your.yale.edu/sites/default/files/rate-agreement.pdf

If I am reading it right, it says that on-campus research has an overhead rate of 67.50 or 69.0 percent depending on the funding source. Ouch! I imagine that the federal government knows that it is subsidizing universities as a whole through this system and intends to do so.

My sense is that in a given research project, the proportion of people who do actual technical work vs total people getting paid directly or indirectly has significantly reduced. Even worse, the folks who are not actually on the grounds doing the work gets paid lion share. I'd read about studies on how vast majority of university funds are used on non-faculty staff and tuition increases where directly proportional to increase in highly paid non-faculty.

In my opinion science is getting less bang for bucks because the low hanging fruit has been picked by now. The big fundamental structures describing nature its working are more or less known. It's about details nowadays and it takes more time and effort to get the details right. You can see that it takes multi disciplinary teams nowadays to discover the connections between these large systems and how they (how all in the universe) connect. People are curious which is our most precious gift from nature and science will go on as long as people stay curious. But yes, not all science is about returns of value in economic sense, a lot of it is fundamental and also often results of science is laying in a drawer waiting to either getting monitized or used for further discovery. Science moves by spurts and hiccups, not straight forward with an even pace.

> In my opinion science is getting less bang for bucks because the low hanging fruit has been picked by now.

This is certainly part of of the reason in physics, experimental verification takes more and more resources. It took the LHC to find the Higgs boson and LIGO/VIRGO to detect gravitational waves. The Nebel committees also tend to be conservative as a later u-turn would be unfortunate. For example, no prize has been awarded for Inflation theory.

While I think the analysis is quite flawed with regards to Physics I do think there is some value in looking at the Chemistry and Medicine prizes. It does torpedo the core assumption of the singularity movement, that of exponentially increasing progress.

Articles like this are best considered as a starting point for a conversation rather than firm statement of reality.

But isn't a big part of the conservatism that there is so little that can be tested at a macroscopic level? The existence of the neutron can be demonstrated with some polonium and berylium.

I agree with you, but I would like to qualify that. The low hanging fruit analogy is apt in more ways than one; there are likely to be deeper truths out there that we don't know enough to model; take dark matter and dark energy and the expanding universe for one. In that respect I disagree that "the big fundamental structures ... are more or less known"

However, these fruits of science are so far out of reach of humanity's experience that they will be very difficult to exploit.

Personally, I think that the social sciences have yet to see their renaissance. The social sciences involve complex systems that are difficult to model, so there simply aren't comprehensive quantitative theories in these fields on the order of physics. If we develop efficient, cheap computation, orders of magnitude greater than what we currently have,[1] perhaps the social sciences may see its heyday.

The social sciences are stuck in a vicious cycle of politics, not technicals: There is tons of behavioral science and evolutionary biology that they continue to ignore.

> The big fundamental structures describing nature its working are more or less known. It's about details nowadays and it takes more time and effort to get the details right.

This (bullshit) is what they thought in the 1890s,and is why physics was considered boring, basically a dead field. Oops!

Difference is, this time around we might be standing at the edge of what we can possibly know and/or understand.

And how could we possibly estimate if we are or not?

Perhaps there are many levels to understanding and knowledge? It's not a binary thing.

Models can be correct in a particular domain of application, but incorrect outside of those constraints. Progress is made by determining the constraints that define the domain of applicability for a given model, and then trying to find new models that are valid beyond those constraints [1].

For fundamental physics, we may have reached the edge of the current "level". Our results at this level seem to have a very wide domain of applicability. Who knows how long it will take to make a fundamental breakthrough that will really shake things up and deepen our understanding...

[1] A concrete example: Newton's model of gravity is not "wrong" or "false". Its domain of applicability and validity is simply much more restricted than the models developed by Einstein. Interestingly and perhaps unsurprisingly, many common phenomena that humans care about fall squarely within the area where Newton's models are quite valid and useful.

Here is quote from 1894 by Albert A. Michelson that captures the attitude of that time: ``[...] it seems probable that most of the grand underlying principles have now been firmly established and that further advances are to be sought chiefly in the rigorous application of these principles to all the phenomena which come under our notice.''

That was before quantum physics, GR, transistors, lasers, etc...

Not true in biology, not remotely true in neuroscience. The big questions are hanging very visibly but it is just not easy to answer them.

Aristotle thought the low-hanging fruit had been picked in his time. (I didn't keep the original quote, but it said roughly that most important knowledge that could be figured out probably would be within a few lifetimes.)

This may be accurate in delimited fields like fundamental physics (though even there you have to wonder when you reflect on e.g. the half-century between Turing and Deutsch/Feynman with no one picking up quantum computing). But our understanding of nature is so tiny compared to what's out there. We're only a small percentage of the way to understanding even our own biology in full. Why would the frontier of available discoveries become much poorer, relative to our abilities if deployed well, long before useful discoveries run out? Maybe they really do, and we won't really know until we can look back much later, but it's not how I'd bet.

> This may be accurate in delimited fields like fundamental physics

No, major advances are still possible there.

That observation is decades too late. Big companies set up sizable research labs in the 20th century. After WWII, it was expected that a big industrial company would have a sizable R&D operation.

Then, in the 1980s, those big R&D operations stopped paying off. Gradually, the big corporate labs closed - Bell Labs, RCA's Sarnoff Labs, Xerox PARC, Westinghouse Labs - gone. IBM's labs are far smaller than they once were. The payoff wasn't there. The easy hits from research were gone.

Hum...

My reading is that R&D facilities couldn't compete with the sheer amount of small improvements that came from the uneducated labor force. Japan grew out empowering the employees to do development, and the entire world copied it. What leaded us on a path where people are making the Uber-for-X apps and some actually getting rich that way.

And that is exactly the opposite of a sarcity of low hanging fruit.

Another problem with this data is the rise of team science and huge author lists for papers - something the Nobel prize isn't really able to deal with. Think of the human genome paper or the Higgs boson paper - both good 21st century science but lacking the easily identifiable "super scientists" the Nobel committee is looking for.

Additionally it is hard to fully evaluate the impact of scientific work fully until many years have passed since publication.

It's no wonder they prefer to find their winners in the 80s. ..

A lot of 'money spent in science' has not really been spent. Just offered with lots of requisites that nobody can reach and then recycled for other projects. If granted, the money can be blocked for months (instead to pay an entire year to scientists, you pay 7 months... for one year of work), and can be partially siphoned off again using bureaucracy. Some politicians are very fond of this 'miracle of bread and fishes' trick.

"In the last ten years we spent '10 millions' in science projects". Looks great.

"In the last ten years we flashed the same million 10 times before to put it again in our pocket". Not so great, often the real thing.

I actually don't see this from his data. It seems more like he started with a hypothesis, wrote the article as if he was correct, and then gathered data to try to support it. Then after seeing the data wasn't as commensurate with his idea, he wrote around the inconveniences.

Yep. I don't know what article the other people in this thread read, but it wasn't the same one I did. At best, the graphs can conclude that scientist's opinion of nobel prize winners stays about the same, regardless of time.

This is why science should be led by the government, not for profit companies. Science is for the greater good of humanity, it should not be solely about making money.

I love science (at least in the ideal form). That being said, if knowledge provides a real, tangible benefit to mankind, people should be willing to pay for it. If it doesn't, people shouldn't be forced to pay for it.

I agree that some valuable research would go undone if performed by companies because the time window to see a return on investment would be too long. Unfortunately, academia is a huge mess, and we really need a new approach to basic science.

What's your definition of basic science?

The requirement for Science to produce real, tangible benefits to mankind for it to get funded makes a lot of it unfeasible, because a good bit of it is useless on it's own, but is good to figure out because of potential uses in the future.

The real problem is we won't know if something is useful until after we get results.

Gravity is a perfect example - physics has a lot of open questions on gravity. What if someone closed one of them? It could mean nothing, just a slightly better understanding of why things fall - or it could be a major change that allows us to create "anti-gravity paint" making a self-supporting colony on Jupiter simple (image all the "land" you could own there - whatever "owning land" on gas planet means)

Come on, such a thing would allow for reasonable space travel even better than for just making colonies. It is essentially Mass Effect Element Zero - you could create gravity wells for things to "fall into", creating enormous accelerations of space ships while at the same time protecting the crew from the effects.

Potentially even a kind of faster than light travel.

As a side boon new kinds of materials can be created in extreme artificial gravity conditions some this maps to extreme and impossible to achieve otherwise pressures. Including just creating stars in a controlled environment making for immense energy availability. Instead of magnetic fusion containment you'd have gravitational fusion containment.

This is how powerful practically breaking gravity can be.

I just hand wave assumed a slightly different set of rules. With your set of rules (which is at least as likely as mine) you are correct. Since both of us are speculating we are both equally right. Of course it might be that better understanding of gravity proves that artificial gravity is impossible.

Gravity is not a hard thing to understand. If energy is released when two objects are gravitating towards each other, then something must release it. It narrows number of possible options. My current best guess is that we are inside gigantic object (about hundred of order of magnitudes larger than our Visible Universe, made of particles about few dozens order of magnitude smaller than Planck scale) with conditions similar to conditions in neutron star. So gravitation is just Casimir effect at scale.

The type of basic science I'm talking about here is stuff that is foundational for other science that can be applied. A good example would be esoteric math. Looking back, not all of it had immediate application when it was being worked on, but we wouldn't have things like transistors if people hadn't worked on it.

There definitely needs to be a cost sharing scheme in place in science to encourage basic research that benefits everyone. I'm doubtful that the current model is optimal though, and I think we should explore other options.

On one hand, I love that people can research things without worrying about generating a profit. However, when I read about someone who has been funded to spend years studying the sports history of central-asian mongols and I look how much I'm struggling, working in an underfunded public-sector school, I feel... well, jealous and hard-done-by.

Yeap, a lot of para-scientific departments and fields get founding under the umbrella of research and science but what they really do is either politics or almost completely useless studies. In countries where funding is a problem, it is a bit raging to see how much money got wasted this way. These studies have their place too, but their makers should at least have the decency not call what they do science.

I think you need both, but an alternative albeit for-profit way of going big without government requirements attached is the patronage model, which funded Leonardo Di Vinci for example.

Several individuals are pumping a lot of money into moonshot projects, and of course Google X type incubators exists. But I'm excited for projects like Elon Musk sending those artists to the moon, because those are the type of unpredictable upsides that could never achieve government funding.

The core problem persists in either model. If people don't think they should pay for science, they'll vote either with their wallet or for the politicians who promise to return (or spend elsewhere) their tax dollars. Add in the increasing level of politics in science and the idea of just giving some people a decent life regardless if they produce result or not becomes a very hard sell.

The root of the problem with either funding method goes back to people not valuing science that may bring no benefit for to them. What's the solution for fixing that? Especially when talking to someone who care barely make ends meet.

We've seen a big shift over the last 75 years or so from most science being funded by corporate research to most science being funded by government agencies.

Politicians come and go and can't be trusted to understand or even appreciate science. Something that will not probably pay benefits before four years is a negative sum game for they.

I’d doubt you’re suggesting that for profit companies should not be allowed to perform science?

If science is worth doing, and companies have money, then they’ll probably do a lot of science.

If you’re arguing that science should not only be left to private companies then I fully agree and think most would.

I'd personally bet on some sort of major breakthrough in space travel or energy sources or both (perhaps due to new physics) to have comparable results to the decades of the atom.

Being that, there is just not enough gain or even potential gain of technologies being surveyed right now, even considering full on genetic engineering and universal nanoconstructors.

Well, unless immortality or nigh immortality of some form happens. Or we found a way to expand our intelligence and performance in some huge way. And then it becomes available.

Compared to atomic and subatomic physics AI is a joke and genetics is at witch doctoring levels. Self driving cars are like the steam experiments in 1800s at best, with even less of an impact.

At least genetics has potential to change everything, more than automated transportation ever can.

Nanotechnology is making some progress but not nearly fast, good or cheap enough.

Another breakthrough could come from side physics like photonics.

This could also be titled: “the foundational truths in a field are more consequential than subsequent findings”.

What if the survey respondents value earlier discoveries more because they understand them better? I remember my mathematics professors saying, once they chose a specialty within a field, there wee only a dozen or so people in the world who could discuss it on the same level. So concurrent appreciation is rare within a single field, let alone across fields.

But, as a high schooler, I was taught the basics of Einstein's general theory of relativity. I couldn't do the math, but I appreciated its value. Maybe the respondents undervalue recent discoveries in different fields.

Effects like that could only be measured if they repeated the same survey many times over decades before publishing. Even then it would be confounded by a lot of other factors.

Instead they’ve marched forward using junk science to discredit science. I found it a really weird and off key article.

They might be right but I wouldn’t use this article to demonstrate it. (And I otherwise have a lot of respect for the authors.)

A huge part of this is the general perversion of science. It starts at the bottom with the abuse of p-hacking by lowly grad students. It goes all the way to the top with fraudulent for-profit science by Big Pharma to get drugs and medical devices approved that don't work. All of it is part of the academic-medical industrial complex that views science not as process for discovering truth, but as a tool that can be used to generate profit.

There are enormous breakthroughs that I know are very close, but simply inconceivable due to the academic-medical establishment that has stranglehold on science.

Ex-professor here. Reason for this is trivially obvious to any outsider observing the day of an academic: A dazzling torrent of time-wasting grant applications, awards and other such nonsense.

The system is basically designed to take productive scientists, and waste them as quickly as possible. After all, how better to protect your mediocrity?

Never been happier than after leaving...

Does anyone else find it fascinating the Patrick Collison (the co-founder and current CEO of Stripe) co-wrote this article?

The dude makes me feel inadequate like no other.

Probably impossible to find out, but seeing a comparison - however loose - to other countries would be ideal.

Annecdotally, it seems there's at least one significant leap per week listed on HN that can be attributed to China. Reaching 100 million degrees from a fusion reaction (from earlier this week) comes to mind.

But of course the general public isn't enlightened by the mainstream media about such things.

The Chinese fusion reactor was not a leap. Already achieved in JET (Joint European Torus).

That said, if China decides to pour a lot more into R&D, it'll be doing leaps in no time, simply due to brute force.

Link to JET achievement? Please??

Chinese? European? Any comparison would be better than none.

Brute force? Does it matter? Results are results, yes?

The arc of my point is, the USA so often as a very self-serving, often myth-based view of itself. It's as if no one else in the world might have a better approach. That's a mistake.

JET achieved 0.7 Q, 24MW in 16MW out.

It operates somewhere around 100-200M Kelvin. https://www.scienceinschool.org/2013/issue26/fusion

> Brute force? Does it matter?

No, I doesn't mean to say it as a negative thing. Fusion research funding is very low compared to what is needed to build big enough devices. Because when it comes to fusion, size matters, as efficiency goes up with size.

And that's what I meant by brute force. China can simply build a bigger one and reap the benefits of size.

> Results are results, yes?

Yes, and more data is always better in plasma science.

Why do they assume something went wrong? At some point, we ill reach a point where there is far less left to discover or invent. (other than AI -> as long as anyone has a job, there's always room for automation, yikes!)

We assume that technological progress can continue forever and ever. I think this is an incorrect presumption.

Excellent book on this topic.

Big Science: Ernest Lawrence and the Invention that Launched the Military-Industrial Complex

https://www.amazon.com/Big-Science-Lawrence-Invention-Milita...

Considering how large the corporate sector is today and how corporate inspired the government sector is today... Science is remarkably unaffected.

We have tenure and sociology departments, a teacher-researcher combo tradition... All stuff that does not lend to ROI calculations or the kinds of resource allocation done elsewhere.

I'll agree with Yaa101 that a big part of the story is the picking of low-hanging fruit, or to put it less metaphorically, that in the first 100 years of institutionalized science (let's say 1850-1950 without getting caught up in the exact dates), there were a lot of fundamental questions that could be addressed through the application of relatively systematic, rigorous observation and experimentation, and modeling with the kind of math you can do on a chalkboard.

Within this time period, though, a lot of these questions were addressed and the new questions that arose required more data, better instrumentation, and more advanced mathematical modeling techniques to address.

In my own field (geology, in particular tectonics and earthquake studies), this was laid out in a very explicit manner: the fundamental mode of observation is geologic mapping, and the terrestrial surface of the earth slowly got mapped. The mapping of the past may be refined or re-interpreted but rarely does it need to be redone from scratch. It is done to a reasonable level of resolution. There is still a lot of unknown under the ocean basins, but we have strong theoretical and empirical arguments for why those areas are not as complex as continents and therefore less interesting.

The late 1960s through the mid 1980s saw the development of plate tectonic theory which completely revolutionized the science. Now, 50 years in, we have some second- or third-order questions but most of the first-order questions have been addressed.

Today, the major developments of the field come from better instrumentation, for the most part. In the sub-field of tectonics, progress comes from the development and application of new methods for dating rocks or other geologic features (including things like exposure dating or 'how long as this rock been at the surface of the earth'), and from using satellite-based measurements of earth deformation (GPS and radar interferometry) to actually measure the motion of tectonic plates and sub-plates. Additional, continuous refinements in seismic imaging of the subsurface (driven by the oil industry primarily) has also been very helpful.

This stuff is really expensive! It's hard to go camp out, hike a bit, make some observations, and write a good paper. The instrumentation to get the age of rocks might cost $100,000 and then when consumables and salary are factored in it might cost $500-$2000 per sample. You might need 10-20 samples to really find out anything new in your 10-km by 10-km area of interest. And I believe that geology is quite cheap relative to high-energy physics or whatever. Major geophysical experiments can cost millions. We piggyback on physics and other tech to a large extent---launching GPS satellites for example, or using obsolete particle accelerators for geochemical measurements. The oil industry spends (tens of?) billions a year acquiring data as well but very little of it becomes public or available to researchers, though the cumulative data release from industry is significant. [NB, I may be 1 order of magnitude short on any of these numbers.]

In general there have been very few major theoretical advancements in the past 20-30 years. We have gotten better at recognizing coupling between tectonic process and earth surface processes, and as instrumental datasets slowly increase (as we observe more earthquakes, etc.) some smaller boxes get checked. However, a geologist from the mid-1980s would be able to navigate today's scientific landscape pretty well. The fads are different but like any fashion, many are cyclical and were fads in the 80s too.

I personally see advancement coming from better statistical and numerical modeling, and the availability of high-quality global datasets (primarily created through large international collaborations, which is a post-cold war thing). I also see a lot of room for improvement in our understanding of the coupling of mechanisms spanning vastly different timescales--for example earthquakes occur in seconds, post-earthquake phenomena last weeks to decades, the earthquake cycle lasts hundreds to thousands of years, and the cumulative deformation from earthquakes and related processes is what we call 'tectonics' over millions of year timescales. It's really hard to make a single numerical (i.e., finite element) model that works over all of these timescales and is driven by basic physics (i.e. an earthquake results from forces applied rather than being imposed). Nonetheless there are almost certainly a lot of really important coupling processes that occur on these different timescales but they are really hard to analyze (and a lot of interesting stuff happens at 20 km depth and at mm/yr rates, which is pretty damn hard to observe).

So I guess I see a lot of 21st century science as bridge building rather than outright discovery. This is fine. An analogy would be moving to a new country. The first bit is the discovery of the place, then you learn what language they speak. The learning doesn't stop there. As you learn the language, a lot of daily stuff makes sense. As you gain an understanding of the culture and the history, and can interact in a meaningful way, the value of that learning continues to increase, but it doesn't feel like 'discovery' like it did in the first year.

Yet we directly observed gravitational waves, and compute power continues to grow.

Human have succeeded most of the applicable rules of our physical world. What left to be discovered requires much more time and effort.

An unpopular possibly "devil's advocate" opinion: Considering that our scientific advances have led to us quickly using an enormous developed capacity to make the earth much less hospitable to human life, maybe a slowdown in scientific advances is not a bad thing.

On the other hand, yes, we are going to need some science to deal with what we have wrought without maximum misery. But I don't really trust us with it.

What a joke. There are hundreds of thousands of papers published each year, and some very small faction of those (but certainly hundreds to thousands) make a bit impact on their field. But we summarize science based on three yearly prizes? Clearly science is not having enough impact on science policy writers.

I find this article preposterous.

>While understandable, the evidence is that science has slowed enormously per dollar or hour spent.

The only evidence for this is Nobel Prizes won, split up by decade, and polled!? Biology, arguably one of the most exciting fields right now, doesn't even get considered in the Nobel Prizes! You can get a Nobel Prize in medicine/physiology for biology-related efforts (see GFP), but there's no Nobel Prize for plant-related biology (that's what the Kyoto medal is for, which isn't mentioned here?!?)

I can't wait for science-illiterate politicians to take this ('Look at what YCombinator is saying!') and say we should defund science.

It has never been as hard to get funding for your science as now - see e.g. https://theconversation.com/with-federal-funding-for-science... - with huge issues (scientists living in precarious circumstances, ridiculously low salaries, attacks and defunding by conservative governments (US, Australia, Harper government in Canada) - writing an article now saying that science is stagnant is foolish at best, dangerous at worst.

>Over the past century we’ve vastly increased the time and money invested in science, but in scientists’ own judgement we’re producing the most important breakthroughs at a near-constant rate. On a per-dollar or per-person basis, this suggests that science is becoming far less efficient.

A much simpler explanation is that we've picked the low-hanging fruits (think of how diverse the work of Darwin was! He picked up novel fossils literally on beach walks), now (especially in physics!) we need bigger and stronger efforts to go for the harder fruits.

>>Over the past century we’ve vastly increased the time and money invested in science, but in scientists’ own judgement we’re producing the most important breakthroughs at a near-constant rate. On a per-dollar or per-person basis, this suggests that science is becoming far less efficient.

>A much simpler explanation is that we've picked the low-hanging fruits (think of how diverse the work of Darwin was! He picked up novel fossils literally on beach walks), now (especially in physics!) we need bigger and stronger efforts to go for the harder fruits.

Your explanation that we have picked the low-hanging fruits is not mutually exclusive with the increased time and money invested in science. They would even be two sides of the same coin- as the easy pickings disappeared, we chose to invest more to get the harder fruits. As the trend continues, the search space for novel, useful discoveries becomes larger and larger, and the cost increases.

How do you prove to the politicians that a particular scientific investment is worth it?

>How do you prove to the politicians that a particular scientific investment is worth it?

I don't think that's possible. The history of science is full of science that shouldn't have worked but did - think of Barry Marshall drinking Helicobacter pylori, which no-one thought to be a causative agent, or think of Norman Borlaug, who came up with shuttle breeding (grow a plant twice a year by driving seeds around, at a time when people thought that you have to let the seeds rest for a while) - but the history of science is also full of things that should have worked but didn't, we just don't hear about those (classic survivor bias).

In some cases, you can tell whether an experiment is going to work and how it's important, especially if it's incremental work - in some cases, you simply cannot predict.

Just think of how delighted G. H. Hardy was that his mathematics was kind of useless to the general public, and how useful his work is nowadays (most importantly the Hardy-Weinberg equilibrium I guess?), he could not have predicted how that works out! What would he have written into his application for funding?

>"How do you prove to the politicians that a particular scientific investment is worth it?"

You* need to make surprising predictions and perform astounding feats that pretty much anyone can verify for themselves. This is an old problem: http://www.hellenicaworld.com/Greece/Technology/en/Syracusia...

*By "you" I mean at least people associated with you or your ideas.

> How do you prove to the politicians that a particular scientific investment is worth it?

You vote someone who is already convinced of it.

>"(scientists living in precarious circumstances, ridiculously low salaries)"

The problem is the funding is constrained by the average productivity of the researchers. Too many crappy researchers have been allowed into the club (see my NHST post above) who are pulling down the average. Adding more money isn't really going to help. Actually less money may drive some of those people away for a generation and increase the total productivity. More productivity from fewer people -> higher salaries.

One reasonable way to look at the "bang for the buck" invested in biomedicine is to look at r&d productivity of the pharma industry. Pharma spends more on r&d than publicly funded research, and innovations in academic research typically trickle down to pharma

By most estimations, productivity of biomedical research is decreasing. Some estimate it is decreasing exponentially

One explanation is the low hanging fruits model, but I prefer an explanation that combines the low hanging fruit model (that we rapidly exploit the most obvious applications of a new tech) with a "punctuated equilibrium" perspective of innovation -- that every so often, we get a new breakthrough that opens up a whole new space to play in. Maybe gene editing, gene therapy, cell therapy, the microbiome will have that impact today

> The only evidence for this is Nobel Prizes won, split up by decade, and polled!? Biology, arguably one of the most exciting fields right now, doesn't even get considered in the Nobel Prizes! You can get a Nobel Prize in medicine/physiology for biology-related efforts (see GFP), but there's no Nobel Prize for plant-related biology (that's what the Kyoto medal is for, which isn't mentioned here?!?)

It's also a metric that lags by decades, the authors say that "This is too few to get a good quality estimate for the 1990s, and so we didn’t survey those prizes." So even if this metric is useful, it is measuring what was going on in academia through the 1980s.

> but there's no Nobel Prize for plant-related biology (that's what the Kyoto medal is for, which isn't mentioned here?!?)

Well, plant-related biology is something specific, but I guess that if it is a big breakthrough it is a strong candidate for the Chemistry Nobel

In the same way there is no Math Nobel, but there's the Fields medal. Nobel is not all there is to science.

Yes, the point of the parent comment is exactly this: Nobel prize A/B comparisons by extant professors is a rather ridiculous metric for overall science progress.

>I can't wait for science-illiterate politicians to take this ('Look at what YCombinator is saying!') and say we should defund science.

This is an extremely hyperbolic thing to say. While I don't think that in this economy spending levels for science need to be decreased, reducing spending won't be "dangerous". I don't even know what "threat" you're imagining is so important that reducing science spending is necessary.

This article is actually spot on, from my and my partner's experiences in PhD programs. The field is saturated with people and everyone is trying to stake their claim. Pressure is high not to get scooped, frontiers are small and hard to find, and most progress is very small increments which take lots of time and funding to accomplish.

I think the money is fine, but the number of PhDs awarded needs to decrease. Important discoveries nowadays take lots of time and very dedicated people, but the system is set up to incentivize quickly nabbing low hanging fruit. It's rare that I attend a PhD defense where the contribution is anything but a baby step forward.

I don't understand their methodology at all. Considering that the frequency of awarding nobel prizes remains the same , there is nothing that can be infered from their survey.

That said i agree with the premise of their article that science is not living up to expectations, and i believe this started somewhere in the 90s. For example consider the recent "Burden of disease" survey which was linked here a few days ago:

> GBD 2017 is disturbing. Not only do the amalgamated global figures show a worrying slowdown in progress but the more granular data unearths exactly how patchy progress has been. GBD 2017 is a reminder that, without vigilance and constant effort, progress can easily be reversed.

https://www.thelancet.com/journals/lancet/article/PIIS0140-6...

This effect was probably many years in the making , and is only now becoming apparent. One possible explanation for the "sluggishness" in scientific output is that nowadays it is lacking new grand ambitious projects, in other words centralization and Big Science. The proliferation of PhDs has changed the way funding is allocated in recent decades for purely political reasons. It favors thousands of small grants that go to individual independent researchers who are all studying minute effects, are looking for results that barely pass the significance threshold and are publishing in order to build publication records to advance their career. The total number of researchers has doubled since the 80s , therefore this model may be actually detrimental to the process of scientific knowledge discovery at this time. For example, in my field of computational neuroscience, there is currently no equivalent of the "large hadron collider" project for the brain. Some years ago one such project, the 'Human brain project' was proposed and funded, which had a specific and ambitious goal: to simulate the entire brain. Academic politics however fundamentally altered the project and it is now a funding source for various kinds of ordinary research. To be clear, the project was probably ill-conceived from the start (imho), given that we don't have enough info to simulate the brain correctly, but regardless it was one of the few such efforts towards a singular ambitious goal. Subsequent funding schemes such as the US Brain initiative do not have such a focus. The ones who do undertake big science are private institutes like the Allen institute in their attempt to accurately map the entire brain. In any case i think public policy has to focus less on academia and more on science, in order to make progress otherwise we are bound to see detrimental effects a few years down the road.

The graph should be adjusted by inflation, and probably by other factors, in order to accurately back the article's claim. $1M in 1940 would be equivalent to $17M in 2018.

The graph states that it is adjusted for inflation.

Ok thanks. I must have missed it.

science having its own logistic curve ?

A lot of the comments are listing their personal suspicion of causes, without attempting to illustrate or provide evidence. Some of those suspicions are probably true, and some false. Obviously it is also an outlet for all our personal gripes with the system. Irrespective of the veracity of a proposed cause, I would like to see more discussion of actual examples of current progress, and trying to identify what caused or enabled the progress to be made by comparing with the average paper in the back of our mind: what was different, why did they achieve the progress today and did nobody achieve the insight say five years ago?

Ideally, since we are discussing on HN, it should be an example that would be understood by most participants here.

It is in this spirit that I will give exactly such an example of recent progress (with which I am entirely unaffiliated).

First some minimal background which I assume you are not familiar with: quantum chemistry and solid state physics software.

Just take a quick look at this list on wikipedia, you may recognize the names of some pieces of software like ABINIT... make sure you pay attention to the DFT column, and that virtually all packages support DFT calculations, which has been pretty much state of the art for the last decades.

https://en.wikipedia.org/wiki/List_of_quantum_chemistry_and_...

In physics gradients often arise naturally for example forces as gradients of potential energy... But in general even outside physics gradients are useful in nearly all field for optimization...

Now the part you probably already understand: AD, Automatic Differentiation or Algorithmic Differentiation, and in the often occuring case of a single scalar function in N variables, reverse-mode AD... And that it takes on the order of 5 times a single function evaluation to calculate a gradient.

Now specific subfields of physics have been using AD and adjoint sensitivities for a long time (nuclear engineering, oceanography) but it is not a standard part of physics curricula.

Outside of these specific subfields, Automatic Differentiation has been gaining momenta over the last decades (books, comprehensive reviews, ...)

Physics students of course learn differentiation symbolically on paper for short formulas, or in software packages like MACSYMA, Maple, ... but even then you keep the number of variables low for tractability. These students will also understand you can emulate differentiation numerically by using a finite delta:

d f(x1, x2, ..., xN) / dx2 = ~ [f(x1, x2 +delta, ..., xN) - f(x1, x2, ..., xN)]/delta

fully understanding that for a complete gradient you need (N+1) evaluations of f, that delta too large will be inaccurate due to functional nonlinearity, and delta too small will be inaccurate due to numerical rounding of floats... Nobody thinks of showing reverse mode automatic differentiation to physics students as part of their curricula! If you crash into a physics course and ask the students how to calculate the gradient of a big function of 1000 variables, they won't be able to help you, but then you can explain that such a thing is in fact possible!

You see where this is going...

What if some of the numerical computations, say molecular modeling, could benefit from this insight?

That's exactly what happened recently:

https://arxiv.org/abs/1010.5560?context=cond-mat

Now look at the authors' institutions of this paper:

SISSA, International School for Advanced Studies, 2 DEMOCRITOS National Simulation Center, Quantitative Strategies, Investment Banking Division, Credit Suisse Group

That's people with an interdisciplinary background.

Could it be that we are over-specialized? Or perhaps mis-specialized?

Imagine an alternate world where there is so much math that mathematics has decided to specialize thus: after a common course of Fundamental Math, you can specialize in Definitions, or perhaps in Theorems, or perhaps in Proofs, or perhaps in Conjectures... clearly their progress in math is going to suffer!!

Could it be that experts and professionals have become "too polite" to the point of circle j* ? A kind of "politeness omerta"? Imagine a mathematician and a physicist talking: as long as we're discussing math, the physicist nods with interest, and doesn't make suggestions how he would do it differently and vice versa. The "don't criticize a professional in his domain" attitude? Wouldn't it have been better if someone who understood AD say 20 years ago, who ends up in a conversation of molecular chemistry software physicist to say something like "I don't have the domain knowledge, but first thing I would do is find out if the calculation could be viewed as the computation of a gradient, or else as an optimization of an explicit goal" "how dare you make a suggestion outside of math?! and thats thousands of variables, its intractable" ?

We might have had this 20 years ago! Mansplain it if necessary.

it's this upside down world where we are polite and stoic in our papers, but elbow working back stabbers undermining the workgroup next door, all the while feigning this "politeness professional omerta"

Reply on news.ycombinator.com