Hacker News Reader

Comment by renewiltord | Hacker News Reader

Well, it's no surprise that people optimize to the metric that's available, right? Private labs aim to make production things. Public labs aim to collect grants.

The only real hope for modern science is a business with a long-term view. That's because that business is only rewarded if their thing works.

If there was ever a time when public academics worked for science, it is no longer that today. Now it is occupied by grant vultures. This makes sense since we never rewarded science. We always rewarded a proxy. And in any sphere where you do that, eventually the proxy becomes the target as more sophisticated strategies evolve and each generation of participants gets better at optimizing against the proxy.

View on HN

The follow on from this is rather more problematic. If you are prioritizing production things, then it means you'll severely limit the topics you explore. Anything that would take too long to get to market, or is too risky, or is too theoretical, will never be prioritized.

This is the reason for the tenure system, to give true freedom to explore novel and future oriented work. Things that won't pay off for decades, if ever. Ironically, the grant system does cut against it, especially with some grants now looking for more near term impact. Also, publish or perish cuts against this, because it optimizes for quantity rather than quality. Similar to industry labs, risky work or things that will take too long to get to publication are now discouraged. Instead we have a plethora of weak, simple, and short publications, often duplicating previous work due to a lack of consistency in terminology.

mRNA vaccines took ten years before they "made it" and they needed a black swan event to really take off. Moderna IPO'd eight years in.

I think this is a false premise that we have. That there is any long-term science going on at all in government labs that wouldn't happen in private labs. It's conventional wisdom but not likely to be true.

The market is very sophisticated now. There are players who make decisions outside of any human's life. There are instruments to insure against hundred year events, and it is possible to price both rare and far off successes.

If you've peeked behind the curtain you know there's a lot of fake science there, including photoshoppery of microscope images that's now considered science. It's out there, I've seen it, I know it's real.

Public labs have decoupled success from knowledge and, in doing so, many long-term fundamental science projects are now fake.

People thinking in the same line of you is why I think China is going to be the global leader in scientific discovery in a decade.

We've stagnated while they've organized their efforts for long term discovery and long term technical brilliance.

I guess time will tell if public or private sector research is the best possible solution.

I see no evidence that China has organized around long-term discovery or technical brilliance. They have many scientists, and relative advantages in certain fields, but their system of science still has major structure issues.

For example, one aspect of Chinese science policy is paying scientists based on the impact factor of their published papers. This, however, has lead to a much larger issue of "salami-slicing"/"least-publishable unit" publications aimed at passing the minimum bar of a "good enough" journal rather than contributing to science. Another issue—these payouts are only given to certain authors of the paper, which can lead to issues and controversies over authorship position. These are issues in the U.S. and Europe, but less severe, and not incentivized with cash payments.

China does not have some secret sauce that makes them better at discovery or technical brilliance. If anything their policies are organized towards maximizing prestige. Whether large-scale competence follows prestige, we will have to wait to see.

I know some folk that studied abroad in China as part of their US-based undergraduate program (US faculty teaching their courses in China), and they had a very stressful experience due to the seemingly widespread amount cheating by the Chinese students throwing off the grading curves. The university seemed to ignore it for political reasons.

I guess we'll see. It will be hard to tease out the effects of what is public funding vs what is due to other factors.

I think that Chinese science will advance far more rapidly because the Western ethical schools have converged on the Do Nothing school of trolley problem solvers and rampant safetyism is ruining our ability to study immediately important dangers, that Western society is organizing towards protecting its power structures (overwhelmingly elderly over the young), and that the infantilization of Western populations leads to the inability to study interventions since no volunteer can be said to understand any risk.

On the other hand, while Chinese R&D spending is hockey sticking (see Red Moon Rising issue of The Economist for lots of info on this), I am 50-50 on whether they'll manage to solve their fakery problem faster than ours. Theirs is far bigger but they know it is a problem and it is smaller than we think. Ours is smaller but we don't think it exists and it is bigger than our governments think.

But pouring more money into a device that exists simply to extract money is clearly insufficient.

The other problem with the tenure system is that professors - even the ones that want to try cutting edge research and have tenure - cannot risk the career of the PhD students in their lab on some very speculative and high risk research that might backfire terribly.

Or worse, they do and 10% of the lab goes onto fame and glory, while the others bash their heads against the wall.

> This is the reason for the tenure system, to give true freedom to explore novel and future oriented work. Things that won't pay off for decades, if ever

A long time ago, I heard the opposite view. Researchers only establish tenure once they've spent a long time on a specific path, and empirically, later-career don't suddenly throw out their decades of commitment to one path to explore ground once they're protected by tenure. This rings pretty true to me, based on everything I know about the scientific establishment (and my model of human psychology). There's a reason "science advances one funeral at a time" is such an old saw.

To reduce the focus of grant-funded labs on obtaining their next grant (i.e., ensuring their continued existence), funding organizations should deliberately force a soft cap on the number of research PIs. In the context of biomedical research, US universities have been on a hiring spree for the past 25 years. The number of BME departments increased 150% from 1995 to 2005. This expansion was partially funded by Whitaker Foundation money, which ended in 2006. Consequently, the NIH award rate for R01 grants, which are they main lab support mechanism, fell from 27% to ~ 15%. Naturally, funding a lab in a 15% award rate environment takes much more time (probably 3× more; it's nonlinear due to competition), to the detriment of time spent on research and training. This has a corrosive effect on every aspect of publicly-funded science.

The NIH has strong leverage to correct this. Grant awards include both "direct" and "indirect" costs. Direct costs go to the applicant (PI). Indirect costs go to their /institution/ (employer). Indirect costs are calculated relative to direct costs, generally 70% (so the total awarded sum is 170% of direct costs), but the rate is negotiated and institution-specific. The NIH could penalize institutions with too many soft money (grant-supported) positions by reducing those institutions' indirect cost rates, or even requiring institutional matching, until the number of grant-supported positions falls to a level at which they can place their main focus on doing the work they are nominally funded to do. This would reduce the number of publicly-funded scientists, but would let them act more like scientists and less like marketers.

> The only real hope for modern science is a business with a long-term view. That's because that business is only rewarded if their thing works.

There are limits to that logic though. For example, the business can spend billions on researching more addicting cigarettes or can direct money to drugs which merely manage the disease instead of curing it (as disease management makes you a customer for life).

With respect to the products developed, this is an issue where the incentives of business and industry need to be aligned. Resolving the problem can be as simple as paying more for drugs That cure opposed to treat.

Is there any evidence that public labs are producing less science than they have in the past?

I know fraudulent science is performed in many university labs. If you read the papers you will see only what you think is science. But the results are falsified. The slides are fake. So maybe output is high but determining what is true output is hard.

And nearly no one replicates. I'm sorry that I can't give you info. It's not useful to me or my family to prove this. But consider this a warning. If you move subtly enough you might be able to ask the right people without triggering the defensive walls.

I know I sound like a crank, so I understand if your weight on this warning is not high. But it feels wrong to not warn. So I've done my duty to the extent I find comfortable.

FWIW I believe in the scientific process, and I very much like the idea of long-term fundamental science. I'm not a Trumper, or an anti-science loon, or someone who rejects modern epistemology.

I've spent many years in the academic spaces, both in public genetics labs as well as in computer science departments. I've never encountered anything like what you claim, and no one I know in academia has either. Extraordinary claims like yours need proof, no one will or should take you at your word, especially when there is the weight of many others experiences that cut against your claims.

Lior Pachter is pretty good at calling balls and strikes on this sort of thing

Hey, sorry, I googled but he has a lot of work and I couldn't find where he talks about fraud (or lack thereof) in science. If you have a link handy would you mind sharing?

A starting point: https://liorpachter.wordpress.com/2014/02/10/the-network-non...

Much appreciated

That's okay. I understand why you need to apply a near zero multiplier on what I told you but I'm unwilling to get any family member blackballed over outing anyone.

Your epistemological strategy is correct. And for all I know you are more correct. After all, given any random arrangement of points there will be clusters. And assuming you draw circles randomly, some circles will coincidentally appear full of points. But if the circle is the space of visibility of a sapient being, they will mistakenly determine that all of space is full of points simply based on their own circle.

So just because my immediate circle has encountered this enough may not mean that this is common.

Unfortunately, there is no true way for us to safely exchange information on this front for the positive case (there is fraud), only for the negative case (there is no fraud). So I guess we each act as if this circle is the entire plane and maybe the world will apply sufficient pressures to tease out who is right.

FWIW, I believe your anecdote. It doesn’t mean it happens everywhere all the time. You’re posting with your real name and I can’t think of why you’d come to HN to lie about it. I value reading about personal experiences on HN.

> Extraordinary claims like yours need proof

Scientific scam (or scientific mistake, not necessarily deliberate) is well documented in fact. It seems that >1800 published articles were retracted in 2020 by scientific journals, including 72 related with covid [1] so, maybe is not common, but yes, sometimes it happens. There is nothing extraordinary in that idea. The ideology that requires scientists to be a replacement for the figure of the religious saint is stupid.

[1] An example: 5G Technology and induction of coronavirus in skin cells. Biological Regulators & Homeostatic Agents. July 16, 2020. "showed evidence of substantial manipulation of the peer review"

I never claimed it doesn't happen, it's trivial to see that it occasionally does. The OP seemed to be suggesting that it was rampant and widespread, which I have a very hard time believing. Especially to the level of outright fabrication, rather than sloppy work or presenting subsets of real data in a dishonest way.

Reply on news.ycombinator.com