Hacker News Reader

Comment by oldprogrammer2 | Hacker News Reader

Systems don’t remain constant, though, and every system gets “gamed” once the incentives are well understood. I’m 100% for investment in scientific research, but I’m skeptical that the current system is efficient at allocating the funds. We’ve seen so many reports of celebrity scientists committing fraud at our most elite institutions, and a publish or perish model that encourages that bad behavior as well as junk science that will have minimal impact on their fields. We pay taxes to fund science so that universities or corporations can claim ownership and make us pay for the results.

View on HN

>>> We’ve seen so many reports of celebrity scientists committing fraud at our most elite institutions

Can you define "many"? 100k reports? 10k reports? 1k reports? 150 reports? 15 reports? What's the incidence? What's the rate compared to the public and private sectors? What's the rate for defense contractors? Are we talking social sciences, hard sciences, health sciences? What's the field?

"many" is just intellectually lazy here. The reality is you read a few stories in the media and now have written off the entire model of research funding.

Failures (ethical or otherwise) are an everyday occurrence at scale, and the US research and funding model is at a scale unparalleled in the world.

Even if it's a few. Imagine if honest researchers start chasing the fraudulent results. Now you have several people's time wasted. If the honest researcher is junior (PhD or Postdoc), their career is almost certainly over. Worse, assume the junior researcher is dishonest or marginal. The incentive is to fudge things a little bit to keep a career. The cycle begins anew... inherent in our system there is positive selection (in the 'natural selection' sense) for dishonest researchers.

This should give you pause.

Without claiming that any given administration is taking any action with deliberateness or planning... What is even more counterintuitive is that if the dishonesty hits a certain critical point, defunding all research suddenly is net positive.

I would also suggest you keep your ear to the ground. Almost every scientific discipline is in a crisis of reproducibility right now.

You might think crisis of reproducibility means everyone is faking data. No, that does not mean that. There are many factors to a crisis of reproducibility. One is fake data. A bigger one is a lack of incentive and a lack of complete data gathering details on some metric. Generally even if there is a crisis is subjective.

There's also usually a mismatch between what older scientists and younger scientists think are the right approach to studying something.

But generally, science is pretty good. You're reading small slices and assuming it actually represents all of science. It doesn't. Please give me a better sense of what ground your ear is on. I don't think it's generally representative of most science fields. Science has a cool thing where you could post totally fake data, but there are enough actors that also would question it if it's entirely unreproducible. Most issues are small nudges or selective data (e.g, retesting when data doesn't support your expectations), not blatant lies. The blatant lie stories you hear are not actually common and I'd love to hear where you think they are.

> Most issues are small nudges or selective data (e.g, retesting when data doesn't support your expectations), not blatant lies.

Yeah you missed it. When you do small nudges or selectively report data that's even worse than faking data. Not all villains twirl their mustaches. It's the ones that don't that are the most dangerous, these are the ones that are going to suck time and effort away from the collective endeavour the worst. Everyone knows that leclair can't do synthesis. But how certain are we that Phil Baran's Xenon oxidation really worked?

OP, please grapple with this.

This is precisely why Ted Cruz, etc. go on TV and read out the titles of silly-sounding research about beehives and condoms. Because they know that most Americans have no sense of very low-N statistics. A few examples out of hundreds of thousands proves the point!

Of course it doesn't.

Do you understand that? If so, then why are you casually throwing around those talking points that are contributing to the destruction of scientific infrastructure and human livelihoods? This isn't a game.

> I’m skeptical that the current system is efficient at allocating the funds

I think everyone would be. There's a lot of bad science that gets funded. The point, though, is that you can't pick the good science from the bad without DOING THE SCIENCE.

The easiest thing in the world is to sit back and pretend to be an expert, picking winners and losers and allocating your limited capital "efficiently". The linked article shows why that's wrong, because someone comes along to outspend you and you lose.

Ok... If it's not the most efficient way to allocate funds, it's now your job to design a more efficient way. Good luck and let us know what you come up with!

Sure, but what has that to do with the administration's attack on funding and independence? As someone whose lost a grant award under the current administration's attack on science, I can tell you with assurance that this is more about political power and revenge than it is about improving scientific rigor. If we continue on this path, we will only get worse at science as a nation.

There are reforms that should be pursued: restructuring grants away from endless and arduous begging for money through the tedious grant process of today towards something more like block grants

> As someone whose lost a grant award under the current administration's attack on science, I can tell you with assurance that this is more about political power and revenge than it is about improving scientific rigor.

I'm sorry to hear this, but curious what makes you certain of this? Revenge for what? I ask, because I hear this same template over and over with this administration. eg. DOGE isn't about government efficiency its about revenge.

Literally nothing about their approach resembles an attempt at efficiency. Efficiency is a ratio of input resources to output. No part of the DOGE program I've seen or heard of even considers that relationship. Simply firing people at best results in reduced output, or hiring more expensive contractors. And you've flushed institutional knowledge down the toilet. It's like turning a car off and pretending you've boosted its fuel efficiency because nothing is burning. Except that the car saved you time on other tasks, oops. Firing people and then immediately having to rehire them is hilariously inefficient. Rewriting legacy software like they're attempting at Social Security is a classically inefficient blunder.

I don't know if it's all about revenge, but it's absolutely not about efficiency. It's an edgy teen's idea of tough governance. It's the epitome of penny wise, pound foolish. It's false economy all the way down.

If for no other reason, if you terminate a grant for cause, you have to specify why.

Echoing this. I've had two grants pulled in the last admin, and one in this one, and all of them were very sweeping - and wildly inefficient, killing projects during the phase of ramping up, rather than productively working.

> I’m skeptical that the current system is efficient at allocating the funds

Probably. But the solution almost certainly doesn't involve the federal government policing what is and isn't researched, discussed and taught. We had a system that worked. We're destroying the parts of it that worked, while retaining the parts that are novel. (Turning conservatives into a protected class, for instance--not even the CCP explicitly reserves seats for party members.)

Why would the people paying for the research not control what it can be spent on? Letting the people who spend the money decide is typically not a good system.

typically

Pure science may not be a typical case, though, because the people who control the funds don't really have any idea whether the work they are funding is ultimately going to turn out productive or not. The work involved is far from routine and basically a jump into the unknown.

I get the risk of fraud and nepotism, but in some other situations (Bell Labs etc.), "choose very good people and let them improvise within certain limits of a budget" turned out to be very efficient. The devil is in the "choose very good people" detail.

...Why would the people paying control what it's spent on...?

They do control what it's spent on. There are volumes of compliance about how you can spend the money. For example, can't use the funds on food, alcohol, paying rent, bribing people (yes, seriously, some idiot tried it and then they had to make a rule about it), you have to fly US carriers where possible, etc.

There are also reports you submit showing your progress and how you spent the money, to check that you are spending it on things you said you would.

This thread (not just the person I'm replying to) demonstrates a lot of misconceptions about why we have research funding, how it works, and what the results have been in practice. Please, everyone, don't rely on stereotypes of how you think research funding works.

The system isn’t really designed to be perfectly efficient at funding research. The inefficiency typically corresponds to scientists doing weird un-proposed research that produces new breakthroughs in other areas.

It’s not surprising to me that this post ends with an unsupported “so many reports” coda about research fraud. Research fraud is not zero but it’s extremely rare. It’s unsurprising to me that the “we really care about research integrity” crowd has joined forces with the “let’s defund all research institutions with no replacement” crowd, because it was always obvious that was where this would end.

Whether or not it's efficient isn't as much of a concern as if it's being gamed. Reports of growing university administrations, increase in the cost of an education, and biases in the publish-or-perish model show the old model is no longer effective.

Reply on news.ycombinator.com