A Bad Idea from Congress About How to Fund Science | Hacker News Reader

A Bad Idea from Congress About How to Fund Science

forbes.com

59 pointsxg1545 comments

One flaw comes to mind when the thought of using prizes to fund pure science is discussed. It's actually an issue that is even larger than prizes.

The people who go down the wrong path in science are often performing an important service. It is obvious enough that we cannot know beforehand how to solve all of science's hard problems (they wouldn't be hard problems...) and of course the people who don't succeed still need to be paid.

The other, maybe less immediately obvious thing, is that a person or group who demonstrates the ways a previously plausible line of thought just doesn't work is performing a valuable service. Progress in science depends on what they publish, even if their work ends up being a dead end.

My father researched semiconductors at Cornell his entire career. One time he told me of a way he and his colleagues envisioned producing semiconductors that would have radically shrunk the cost of solar panels. After months of research, they found that the physical realities simply did not permit the hypothesis to function.

Dead end.

In the eyes of Congress, this whole affair was a waste of money, but for every scientist with the same idea, it means one less path to waste time and money on.

If the negative results weren't published, then it was a waste of money. Sadly, this happens all the time.

Incentives in science publishing generally work against the publication of negative results. The publication-based economy lies at the heart of most of what's wrong with science today.

It's not clear to what extent prize-based funding could change this culture. I'm not sure I agree with the size of the initial trial, but I do support experimenting with new models of government funding for science research.

The results were published. Dad was disappointed that it didn't work but he was content to know that he saved someone else from trying the same.

More than that, I'm guessing that it also shone some light on those physical realities and therefore advanced the field in general. Maybe not in a big way, but much of science, technology and manufacturing is build of many small advances.

>It's not clear to what extent prize-based funding could change this culture.

It'll make it worse.

Reproducing the work of others is also important in science.

No one is suggesting that all grants be replaced by prizes. Even research projects which are ultimately unsuccessful are valuable and need to be paid for. But does it really seem like a bad idea to pay the research team a bonus at the end if they manage to actually _solve the problem_ definitively? And if we're already doing that, then why not also offer the bonus to everyone else who can solve the problem, regardless of whether we gave them a grant in the first place? This is a tiny fraction of the budget, and it seems to me like it's worth doing some experimentation to see what works.

If the results of taxpayer (not government, no such thing, it's our money) funded research is hidden behind paywalls and made difficult to use freely through patents the net effect is that we are bound to repeat many mistakes or waste a lot of time and money reinventing wheels invented and patented using our own money.

Here is the offending section of the bill in question:

https://www.congress.gov/bill/114th-congress/house-bill/6/te...

Apparently a panel of 9 persons will be in charge of giving away $80 million in prize money for biomedical advancements every year to any person or company (even foreign).

The 9 people are primarily congressional appointments and the director of NIH (who is already appointed) and the administration of this program can be outsourced to a private company who can take 15% of the money.

Personally, I'd have to agree with the author that this is a pretty stupid idea.

This is not a colossally bad idea, it is a colossally good one (though maybe not as suggested), and perhaps the best way to end the stupid patent circus that is ruining biomedical research.

We need not replace grants this way, of course - but we can supplement, because grants, as we all know, can't even begin to cover the cost of clinical trials, for the most part. This means that only large pharma companies can foot the bill, here. Then they walk off with a patent and collect monopoly rents from the American public to the tune of $300 billion a year, in excess of the market cost of drug production.

If, on the other hand, we can offer a large prize that covers the costs of development after the fact, then we get the goods - development already works this way, the "prize" is simply in the form of a drug company buying you out for your IP.

In the alternative, the US government "buys you out" for several tens of millions. The drug, instead of being patented, goes into the public domain. The developers walk off with tons of money. The public saves hundreds of billions of dollars. Income inequality lessens.

80 million dollars wouldn't even begin to cover the development / trial costs of a single drug, let alone many. Many drugs also fail late in the clinical trial process (phase 2+), meaning that 80 million dollars would just disappear upon failure. I highly doubt the government has the stomach to eat those losses.

Why not? The government can easily, easily pick up the tab for clinical trials, which amount to pennies compared to the amount it is shelling out in prescription drug costs via Medicare. It already does, to a great extent, and it's so heavily involved in the process it seems natural to make them eat the costs.

> It already does, to a great extent, and it's so heavily involved in the process it seems natural to make them eat the costs

As far as I know, the government has minimal involvement / skin in the private development process as it stands currently, and only steps in to perform regulatory oversight and helps define trial design for regulatory approval. Almost all costs are borne by the private sector and all work is done by the private sector.

"Almost all costs are borne by the private sector"

Doesn't every penny of "old people drugs" development come from selling drugs to old people paid solely by medicare? I'm just saying if you trace back where those supposed private sector $$ come from, don't almost 100% of them come from the Medicare program?

On the good side you'd cut out a lot of middlemen. No point paying them off thru multiple steps for doing basically nothing productive.

On the bad side you'd cut out a lot of middlemen. Their corporate owners finance a lot of re-election campaigns and PACs....

> Doesn't every penny of "old people drugs" development come from selling drugs to old people paid solely by medicare?

I'm not sure what this statement means. Drug development costs are accrued by the company before the drug goes to market. Costs are borne by the company's investors or offset using profits from previous ventures. Some of these profits may be from drugs meant for the elderly or they may not.

> I'm just saying if you trace back where those supposed private sector $$ come from, don't almost 100% of them come from the Medicare program?

No, much of the initial development capital comes from investors or from other profits which are from many sources. If you're talking about who is paying for the drug, then you're completely forgetting about private insurance companies, which spend at least as much each year on health care costs as medicare. You're also forgetting about the entire ex-US market.

> On the good side you'd cut out a lot of middlemen. No point paying them off thru multiple steps for doing basically nothing productive.

Government is not synonymous with efficiency, especially when there is no profit motive involved.

Government is often quite efficient. For example, public health insurance has smaller overhead (like five fold less) than equivalent private products. In terms of R&D there is no contest. The entire NIH budget produces prodigious amounts of research, from the basic to the translational, for less than what the entire pharma industry spends just to refine their drug development. The private sector sucks at research spending.

> Government is often quite efficient. For example, public health insurance has smaller overhead (like five fold less) than equivalent private products.

I don't have the numbers so I can't speak to the overhead, but I can point at the VA health care system as a counter to your statements about government health care being efficient.

> The entire NIH budget produces prodigious amounts of research, from the basic to the translational, for less than what the entire pharma industry spends just to refine their drug development. The private sector sucks at research spending.

There are a number of problems with this argument.

First, the NIH can afford to take a shotgun approach to research funding which pharma can't take because the risks of loss are so much lower. The NIH can give out large numbers of grants per year, have a large number of them fail to produce meaningful results, and call their work a success as long as some good work is being done. That model is untenable at the major clinical trial level because the losses would be astronomical. It's the same reason why the Y combinator investment strategy wouldn't work if you wanted to apply it to big private equity deals: You can't fail massively more than a few times before you're in trouble.

Second, the standards for "success" are completely different for drug development and basic / translational research. Most basic research is looking for (at best) proof of concept in an in vitro or in silico model; most translational research is looking for efficacy in an animal model using small populations. A pharma research project has to consider statistically significant efficacy, formulation, dosing, acute and chronic direct toxicity, delivery, off target effects, drug-drug interactions, manufacturing, storage, population differences etc and are held to much higher standards at every stage than your average research lab. By the way, even with these differences in requirements, it is well known that a lot of published, NIH funded research coming out of academia is not reproducible: http://www.nature.com/nrd/journal/v10/n9/full/nrd3439-c1.htm..., http://www.nature.com/news/announcement-reducing-our-irrepro... If this were to happen in pharma, you would have deaths and lawsuits.

This happens all the time with pharma. See SSRIs, drugs that have no theory behind them, tiny effect sizes barely better than placebo, terrible side effects and withdrawal symptoms. They make billions for decades; they are still widely prescribed. What pharma can't do with science, they paper over with marketing.

The NIH already spends about $3 billion a year on clinical trials, mostly in the form of money given to patients. Total spending on clinical trials is estimated to be about $25 billion. Not chump change, but easily fundable, and ridiculously easy to offset with the cost savings in terms of Medicare.

http://report.nih.gov/categorical_spending_project_listing.a...

Also, it's wrong to suggest that the costs are borne by the "private sector" when all of that private sector money is coming from monopoly rents handed to them by the government. That isn't their money, it's ours.

> Total spending on clinical trials is estimated to be about $25 billion. Not chump change, but easily fundable, and ridiculously easy to offset with the cost savings in terms of Medicare.

You're assuming that the government is as good at choosing research directions, choosing which entities to bring to trial, and running the trials as private industry. My bet would be that this is not true.

> Also, it's wrong to suggest that the costs are borne by the "private sector" when all of that private sector money is coming from monopoly rents handed to them by the government.

A lot of new drugs come from start up pharma companies, who are exclusively funded by investors, not the government. They don't see much, if any money from the government until they bring a drug to market. Established companies may use a mixture of funding sources to offset their development costs.

Also, the notion that drug companies are purely profiting from the government is absurd. You're completely forgetting about private insurance company spending and the ex-US market, which account for enormous amounts of money.

The cost to develop a new drug is, on average, $2.6 billion (see, e.g., <http://lifescivc.com/2014/11/a-billion-here-a-billion-there-...), including the amortized costs of failed candidates.

Or, put another way, it looks like a new aircraft supercarrier would be cheaper than a cure for Alzheimer's. Whatever the true cost of drugs are, it's clearly not pennies.

The cost is negative, when you consider how much money would be saved by not having to pay out monopoly rents for decades to drug companies.

It should be noted that gov't funded research at universities doesn't necessarily go into the public domain and can still be patented:

https://en.wikipedia.org/wiki/Bayh–Dole_Act

But we'd make this a condition of the prize, that the drug would go into the public domain upon victory, exactly to avoid the disgusting patent.

Why would anyone apply for the prize in that case? The IP rights on anything reasonably successful will be worth way more than the prize money.

Me and my six scientist friends don't give a shit about the IP rights. We're not going to see those billions anyway, the owners of Roche are. I'd much rather take the same payout from the US government. Plus, after a decade of this shit, without their monopoly rents there won't be any drug companies to buy the IP from you anyway.

So what you're saying is that, if those IP rights are worth billions of dollars, the company that stands to make that won't outbid an $80M prize? I'm a little skeptical. That would be a bizarre decision for a for-profit company to make.

This would be good, but the proposal seems ambiguous at best:

> NO AWARDS TO COMPETITORS LIKELY TO REAP FINANCIAL BENEFIT FROM INNOVATION.—The Director of NIH may not, with respect to an innovation, award a prize under this section to any individual or entity that has a vested financial interest in any product or procedure that is likely to be developed or marketed because of such innovation.

One way to ensure that would be to require that any winning or better submitted innovations come with a royalty-free patent license. But

> (i) Intellectual Property.— “(1) PROHIBITION ON THE GOVERNMENT ACQUIRING INTELLECTUAL PROPERTY RIGHTS.—The Federal Government may not gain an interest in intellectual property developed by a participant in a prize competition under this section without the written consent of the participant. “(2) LICENSES.—The Federal Government may negotiate a license for the use of intellectual property developed by a participant in a prize competition under this section.”.

Clearly no such thing is mandated.

Seems like that would require overturning the act?

The "colossally bad", "stupid" rhetoric would be justified if the proposal were for the NIH to give away a significant proportion of its budget in prizes: it's reasonably obvious that prizes can't be a mainstay of public research funding. Instead, the proposal is to spend $80 million: about 0.25% of the NSF's present budget, and less than a quarter of the proposed increase in annual NSF funding in the bill. In other words, throw this $80 million into the NSF's general expenditure instead and in future it would be about to fund, oh, maybe about 20.05% of business-as-usual funding requests to pursue tendentious environmental-health correlation findings and non-reproducible treatments for cancer in inbred mice. (Until the supply of these inevitably rose to meet its former "equilibrium" with demand.) If that's the opportunity cost of seeing if prizes can stimulate some successful research which will not otherwise be funded, then it seems like an arguable proposition, especially since prizes like the X-Prize are often claimed to stimulate research spending which is a multiple of the actual prize money awarded. (No doubt that effect would go away if the NSF spent $8 billion in funding through prizes, but nothing like that is being proposed.)

No doubt the decision process of awarding the prize would be dysfunctional; but it would be dysfunctional in a different way to the NSF's usual funding mechanisms, increasing the chance that at least one of the two would produce the right outcome in any given case.

The first thing that came to mind was Beezid. Basically we'll end up with countless teams of researchers attempting to solve problems and only a tiny fraction of them being paid for their time.

I agree with the article and propose that we should be doing the exact opposite - finding a way to fund more garage tinkerers. Just add up the risk of failure in tech (at least 90%), the fact that so much of our daily workday is broken up into frivolous minutia, and that researchers spend on the order of half their time fundraising. It should be obvious that the problem is underemployment, and that the way to remedy that is to go from top-down to bottom-up organizational structures. In other words, recognize the dignity of work and not just the financial outcome.

It's a staggeringly bad idea but for none of the reasons put forth in the article (jingoism? Really?). The big reason is fraud. The culture of fraud is overly endemic in the grant-based system as it is and it's not clear that gaming a system of contests is not going to be the dominant mode of prizewinning.

It seems to me that the prize model really could work in areas where there's a lot of value snagged up in existing knowledge we already have and the prize-givers want to shake out that value in the form of good applications. I think this works for DARPA because DARPA these days seems to want deliverables that are in this vein. The prize model can also work well in a situation where the prize-giver wants the results to enter the public domain or be cheaply available to the public.

I don't think this could work to fund blue sky research, though. It's asking for too much risk to be assumed by a researcher who probably doesn't have the resources to justify these kinds of attempts.

This sounds like a lot of FUD. $80 million isn't that much compared to the NIH budget. This is about one third of one percent of all NIH funding we're talking about.

I don't think we should change the funding system from grants to prizes. The existence of my job relies almost entirely on NIH funded grants, and I am not worried about this in the least. When they start talking about replacing the other $23.9+ billion in NIH grants with prizes, then we should worry.

Why not consider this an experiment in funding efficiency? This should be supported by scientists across the board, especially as it's such a tiny fraction of the NIH budget. It's a bit much to claim to fully know the outcome of such an experiment before doing it.

This being said, my priors are that it's a better funding model for engineering than for science. Engineering is more goal-oriented, whereas science is more exploratory. In science, it's not always a good idea to have a strongly desired outcome, because of problems associated with confirmation bias, or from spending too much time in the wrong rabbit hole. Apart from concerns about gaming the system (which are valid), the wrong phrasing of the goal could make it impossible to reach, or otherwise direct people in the wrong way. And it's not always clear that the understanding of some process or phenomenon will ever be done to some satisfaction; sometimes you get there and it's much more convoluted than anyone had previously assumed.

The economist Tim Harford discusses the merits of this funding model quite a bit in Adapt, and is quite supportive of it, as are other economists such as Alex Tabarrok. But it's really not clear that they understand the differences between engineering and science, either. Most of the DARPA/X-prize research, which are the typical examples of this funding model working effectively, are engineering research rather than science qua science, or the exploration of natural phenomena.

But given the NIH's aims (which are more practical than NSF's) and the recognition that blue-sky research needs to be funded differently, it might just be a great complement to the existing funding model. I would love to have a bit more of it for those of my own projects that are geared more towards tool/methods development than immediate phenomena understanding; I work on them in my spare time or in conjunction with scientific studies, and it'd be great if I could occasionally pay myself back for that time upon successful completion.

> These 9 people will be responsible for giving away $80 million per year in prizes. In the scientific world, that’s a huge program.

It's about 0.25% of the NIH's investment budget. Doesn't sound like that big a deal to me.

Knowledge Ecology International are the experts on innovation prizes, lots of background on http://keionline.org/prizes

I haven't noticed any commentary at https://twitter.com/jamie_love (their director) or http://www.keionline.org/tracker on this bill yet, am I missing something?

Here's someone knowledgeable https://medium.com/@bethnoveck/funding-science-with-prizes-i...

So are the prized determined in retrospect? That seems like a very odd way to encourage direction. It would make a lot more sense if bounties were published ahead of time.

It doesn't seem so: apparently ( https://news.ycombinator.com/item?id=9955286 ) this is the text https://www.congress.gov/bill/114th-congress/house-bill/6/te... , and it seems to be setting up something more like the X-Prizes, with at least the objectives (maybe not all the awards) spelled out in advance, than the Nobel Prize which is awarded retrospectively (and to the individuals who did the work, rather than to the project itself). I assume that TFA's comparison to the Nobels is just inaccurate.

Counterpoint:

http://www.darpa.mil/work-with-us/public/prizes

I think it could work in certain circumstances for NIH. It's been successful for DARPA, so it's worth a try.

Why do people think that this has been successful for DARPA? DARPA has spent billions on all kinds of ideas that don't work, including prize competitions. Once in a while they have a "hit" and they publicize that like crazy. DARPA is also known (notoriously, one could say) for operating a good-old-boys system, where you have to butter up the DARPA program managers to get any funding. So no, I don't agree that prizes have been successful for DARPA.

> Why do people think that this has been successful for DARPA? DARPA has spent billions on all kinds of ideas that don't work, including prize competitions.

Some people regard the DARPA prizes as successful because, in some cases, the technology to win the prize was developed.

> DARPA is also known (notoriously, one could say) for operating a good-old-boys system, where you have to butter up the DARPA program managers to get any funding.

Can you explain this a bit more in the context of the prize program? It looks like a criticism of their grant program, but you follow it with:

> So no, I don't agree that prizes have been successful for DARPA.

> This is a staggeringly bad idea. Why? Where do I begin? Well, first of all, biomedical research costs money–lots of money. Everyone I know in biomedical research, in which I’ve been working for 25 years, needs money before they make their discoveries. A prize at the end is nice, but you can’t do anything if you can’t pay for equipment, supplies, and (perhaps Congress will be surprised to learn this) people’s salaries.

This point would be more powerful if he explained it in the context of the DARPA and XPRIZE competitions. Spaceships are surely expensive, but multiple teams were funded. Is there a reason that won't happen for the NIH prizes?

> NIH will give these prizes to foreign citizens–and companies! So apparently they will take some portion of NIH’s budget, and instead of awarding it in grants to U.S. scientists, they will give prizes to companies in, say, Russia. Don’t get me wrong: I think science should be supported in every country around the world. But each country has its own system, and the NIH is by far the most important source of funding for biomedical science in the U.S. We just can’t afford to have NIH give out prizes to the entire world.

I don't think it follows that because each country has its own science system, then the US can't afford to give prizes to the entire world. The prizes won't be higher if they're given to international teams; they just might be awarded earlier. Surely it would be more expensive (at least in time) to restrict it by country, because someone else might be able to make the advance faster than the US.

It appears that the author isn't motivated by advancing science, but by funding American scientists. If American scientists need or deserve more money, that should be addressed separately from how to advance scientific progress most quickly. There are probably ways to accomplish giving Americans more money that don't slow down scientific progress.

Reply on news.ycombinator.com