• Quick note - the problem with Youtube videos not embedding on the forum appears to have been fixed, thanks to ZiprHead. If you do still see problems let me know.

Fake science is out of control

Puppycow

Penultimate Amazing
Puppycow
Joined
Jan 9, 2003
Messages
32,014
Location
Yokohama, Japan
Incentives matter.

‘The situation has become appalling’: fake scientific papers push research credibility to crisis point

Last year, 10,000 sham papers had to be retracted by academic journals, but experts think this is just the tip of the iceberg
Click to expand...
“The situation has become appalling,” said Professor Dorothy Bishop of Oxford University. “The level of publishing of fraudulent papers is creating serious problems for science. In many fields it is becoming difficult to build up a cumulative approach to a subject, because we lack a solid foundation of trustworthy findings. And it’s getting worse and worse.”

The startling rise in the publication of sham science papers has its roots in China, where young doctors and scientists seeking promotion were required to have published scientific papers. Shadow organisations – known as “paper mills” – began to supply fabricated work for publication in journals there.

The practice has since spread to India, Iran, Russia, former Soviet Union states and eastern Europe, with paper mills supplying fabricated studies to more and more journals as increasing numbers of young scientists try to boost their careers by claiming false research experience. In some cases, journal editors have been bribed to accept articles, while paper mills have managed to establish their own agents as guest editors who then allow reams of falsified work to be published.
Click to expand...
The problem is that in many countries, academics are paid according to the number of papers they have published.

“If you have growing numbers of researchers who are being strongly incentivised to publish just for the sake of publishing, while we have a growing number of journals making money from publishing the resulting articles, you have a perfect storm,” said Professor Marcus Munafo of Bristol University. “That is exactly what we have now.”
Click to expand...

Paid according to the number, not the quality of the papers they publish. Is it any wonder that such an incentive system would result in a tsunami of garbage? Ten garbage papers are worth more than one quality paper, so why bother trying to do high-quality research?

Maybe such journals should just be ignored entirely? A large majority of it seems to be coming from one publishing group, according to the article.
 
Yeah, I have a slightly crackpot view that this problem has been going on for a while but getting worse over time, and is at least one of the causes of the decline in productivity in science (in per researcher terms, if not in absolute terms). This is as opposed to the "low hanging fruit" view.
 
Roboramma said:
Yeah, I have a slightly crackpot view that this problem has been going on for a while but getting worse over time, and is at least one of the causes of the decline in productivity in science (in per researcher terms, if not in absolute terms). This is as opposed to the "low hanging fruit" view.
Click to expand...

The first part of your claim is almost certainly true. The second part raises the question for me of "How does one measure productivity in science?" Maybe there's an obvious answer to that, but it seems like a difficult, or at least complicated problem to me.

Clearly, it cannot be as simple as "number of papers published" or "words published" or anything so crude as that.
 
Last edited:
Puppycow said:
The first part of your claim is almost certainly true. The second part raises the question for me of "How does one measure productivity in science?" Maybe there's an obvious answer to that, but it seems like a difficult, or at least complicated problem to me.

Clearly, it cannot be as simple as "number of papers published" or "words published" or anything so crude as that.
Click to expand...
One place to look is in the productivity statistics. At least one marker of scientific progress is that the more we know the more we can do. The scientific progress that went along with the industrial revolution saw a huge growth in productivity. The second industrial revolution (electrification, etc.) also saw a similar growth. In general scientific progress has led to technological progress, which increases productivity. But over the last 50 years the productivity growth rate has been decline (productivity has increased, but at a lower rate than in the past). Economic models (see Paul Romer, for instance), show that continued growth is pretty much dependent upon technological progress.

So, if we use productivity as a proxy for scientific progress, we see continued but slowing progress.

On the other hand, the total money spent on research, and even the percentage of GDP, and particularly the total number of researchers have increased dramatically during that period.

From what I've seen its plausible that the rate of progress in science has actually stayed steady, rather than declining. But even if that's the case, the rate per researcher has declined significantly.
 
Roboramma said:
One place to look is in the productivity statistics. At least one marker of scientific progress is that the more we know the more we can do. The scientific progress that went along with the industrial revolution saw a huge growth in productivity. The second industrial revolution (electrification, etc.) also saw a similar growth. In general scientific progress has led to technological progress, which increases productivity. But over the last 50 years the productivity growth rate has been decline (productivity has increased, but at a lower rate than in the past). Economic models (see Paul Romer, for instance), show that continued growth is pretty much dependent upon technological progress.

So, if we use productivity as a proxy for scientific progress, we see continued but slowing progress.

On the other hand, the total money spent on research, and even the percentage of GDP, and particularly the total number of researchers have increased dramatically during that period.

From what I've seen its plausible that the rate of progress in science has actually stayed steady, rather than declining. But even if that's the case, the rate per researcher has declined significantly.
Click to expand...

One could suppose that more and more money, resources and people are needed to make progress as the questions and required hardware become increasingly complex. It's no longer exceptional for three dozen authors from ten different institutions to contribute to a single paper.
 
Ryan O'Dine said:
One could suppose that more and more money, resources and people are needed to make progress as the questions and required hardware become increasingly complex. It's no longer exceptional for three dozen authors from ten different institutions to contribute to a single paper.
Click to expand...

Yep, that's the low-hanging fruit theory. Basically, we've already made all the easy discoveries, now what's left to discover requires more [people/time/resources]. Galileo could make a discovery by pointing his telescope at Jupiter, and seeing its moons. Now we need the JWST. Rutherford could do his experiments in a small lab, now we need the LHC and thousands of people and billions of dollars working together to expand the boundaries of human knowledge.

It's an entirely plausible viewpoint. I'm just not convinced its the answer. Note, though, that if the low-hanging-fruit view is correct, the thing it's explaining is still real: the declining productivity per researcher.

Here's the somewhat famous paper that first demonstrated this issue pretty clearly:

https://web.stanford.edu/~chadj/IdeaPF.pdf

Are Ideas Getting Harder to Find?
By Nicholas Bloom, Charles I. Jones, John Van Reenen,
and Michael Webb

Long-run growth in many models is the product of two terms: the effective number of researchers and their research productivity. We present evidence from various industries, products, and firms showing that research effort is rising substantially while research productivity is declining sharply. A good example is Moore’s Law. The number of researchers required today to achieve the famous doubling of computer chip density is more than 18 times larger than the number required in the early 1970s. More generally, everywhere we look we find that ideas, and the exponential growth they imply, are getting harder to find.
Click to expand...

(the whole paper is there at the link for those interested)
 
Last edited:
There are also many areas of science that don't in principle lead to any direct growth in productivity or any sort of direct economic productive value. Just growth in human knowledge. Like astronomy or cosmology or even particle physics. Do the findings of the experiments in the Large Hadron Collider lead to some sort of economically relevant increase in productivity?
 
Puppycow said:
There are also many areas of science that don't in principle lead to any direct growth in productivity or any sort of direct economic productive value. Just growth in human knowledge. Like astronomy or cosmology or even particle physics. Do the findings of the experiments in the Large Hadron Collider lead to some sort of economically relevant increase in productivity?
Click to expand...

Sure, but do you expect progress in those areas to be happening at a different rate from progress in the areas that do impact growth? The paper I linked looks at crop productivity, medical research, Moore's Law, and other areas, and finds basically the same trend in all of them. Is there some reason to think that trend exists only in places where we can measure it and not elsewhere? If the low-hanging fruit has been picked in medicine, but not in astronomy, that seems surprising.

I think the same researchers published something more recently that tried a different methodology less tied to productivity figures, but I've forgotten the details. Something about looking through Nobel prizes and what they were given for. Anyway, the upside is that the results are the same. Sorry, don't have the link, but if you're interested you might be able to find it.

Just on the "vibes" side of things, while the number of researchers has increased by an order of magnitude, it doesn't seem like the rate of progress in science has increased by an order of magnitude in the same time. I'd be pretty surprised if that turned out the be the case, so the results in that paper are pretty much in line with my intuitive sense of things. But again, the rate of progress staying constant means productivity per researcher is down (by an order of magnitude).

Of course, under the low hanging fruits view, this isn't anything about the researchers, it's just that the problems they are working on are getting harder. It's just diminishing returns. And that seems to be the most accepted view. I have some reasons for doubting it, or at least doubting that its the whole picture, but doesn't seem to be worth going into those if we don't even agree that the phenomenon its explaining exists.

ETA I didn't mean to hijack your thread, which is an interesting and important story. My comment was meant only as a comment as its related to some thoughts I've been having but which aren't really fully formed yet. This does seem tangentially related, but this thread should focus on the issues raised in the OP.
 
Last edited:
rjh01 said:
One measure of how good a paper is, is its impact factor. That is how often other researchers reference the paper in question. Ditto to journals.

More info here: https://researchguides.uic.edu/if/impact
Click to expand...

I would imagine that even that could be gamed to a certain extent when "paper mills" publish enough fake papers. Just have fake papers citing other fake papers and hey-presto, the papers now have "impact" too.
 
Puppycow said:
I would imagine that even that could be gamed to a certain extent when "paper mills" publish enough fake papers. Just have fake papers citing other fake papers and hey-presto, the papers now have "impact" too.
Click to expand...

You can also get things like citing your own papers when not really relevant or reciprocal citing: I'll cite you if you cite me, which can inflate citation numbers.
 
Puppycow said:
I would imagine that even that could be gamed to a certain extent when "paper mills" publish enough fake papers. Just have fake papers citing other fake papers and hey-presto, the papers now have "impact" too.
Click to expand...

True. But this can be detected and corrected. It is one of the problems Google has with ranking pages.
- If a limited number of people cite each other's work then they are all ignored.
- A cite from a good source is worth far more than an unknown source. And papers that cites many other papers does not count.
- Certain people have qualifications and experience. Anything they cite is worth heaps.
 
angrysoba said:
There’s a really good book on this called Science Fictions by Stuart Ritchie.

He goes through a few different types of problem in publishing such as Fraud, Bias and Hype.

https://www.sciencefictions.org/p/book

Another one that I read recently is Calling BS by Carl Bergstrom and West:

https://calling********.org/
Click to expand...

Good stuff! (although your second link seems to have fallen afoul of the autocensor).

Found this via the first link:
https://www.sciencefictions.org/p/science-fictions-links-january-2024

Which led me to this:
A prestigious cancer institute is correcting dozens of papers and retracting others after a blogger cried foul
And this:
Stop Hocusing Your Western Blots, Maybe (Derek Lowe)
And this (where the whistle was first blown)

And other interesting stuff.
So it's not just a problem in "those other countries".
 
Puppycow said:
Good stuff! (although your second link seems to have fallen afoul of the autocensor).

Found this via the first link:
https://www.sciencefictions.org/p/science-fictions-links-january-2024

Which led me to this:
A prestigious cancer institute is correcting dozens of papers and retracting others after a blogger cried foul
And this:
Stop Hocusing Your Western Blots, Maybe (Derek Lowe)
And this (where the whistle was first blown)

And other interesting stuff.
So it's not just a problem in "those other countries".
Click to expand...

Ah, interesting. Ritchie talks a lot about Western blots in his book.

I see Ritchie also mentions the Dan Ariely / Francesca Gino issue.

That was the subject of an episode of Freakonomics recently about academic fraud.

Link

Another group who get some exposure, in a good way (hopefully!) on that podcast episode is Data Colada.

The people involved in Data Colada have written papers showing just how easy it is to p-hack findings and were some of the people involved with demonstrating how big a replication crisis there was.

As a result, they are not popular with some academics who have managed to fudge their figures enough to make their findings sound sexier and get round the publication bias favouring novel findings as opposed to, say, boring findings that either show a null result or merely support previous research. They are big advocates of open science.

The Open Science Framework is trying to create better accountability for research scientists by getting them to pre-register and to share data to allow other researchers to easily do their own analyses.

One of the issues that the Data Colada people found was something called researcher degrees of freedom which comes down to how the researchers choose to display their findings which may end up purposefully or perhaps unconsciously skewed to give more impressive results. If researchers have to pre-register how they will conduct their research, they will have less opportunity to fudge the results. Presumably. Of course, there is always the possibility that determined bad actors will find ways around that too, but at least it would be clearer that fraud was committed rather than incentives lining up with unconscious bias.

Here is a paper they wrote on this.

There is a podcast episode here in which two academics discussing the paper.
 
Uri Simonsohn in the most recent blogpost on Data Colada (November 2023, but I guess he's been busy preparing for the Gino lawsuit) gives some suggestions for how to improve science:

Suggestions For a Better Future

1. Preregister or justify. It's time for journals to require authors to justify when an experiment is submitted without a pre-registration. Papers without compelling justifications could be desk rejected.

2. Journal checks. Journals should dedicate the 10 minutes per-paper that it takes to ensure that links to pre-registrations actually lead to (the public version of) a pre-registration.

3. The 20 second rule. Authors (and journals) should ask: "How long would it take a reader visiting the paper's repository to find the pre-registration for Study 2 and the data for Study 3"? If the answer is "more than 20 seconds", the paper is not ready for publication.

4. Single links. Toward that aim, I recommend that authors include in their papers a single link to a well organized repository with all files (code, data, materials, and pre-registrations). This is done by default at ResearchBox (an open research platform that is fully integrated with AsPredicted). See screenshots below. Authors posting to the OSF should spend the 20 minutes that it may take to properly organize and label their files.

Since not everyone is familiar with ResearchBox yet, below I include three screenshots of ResearchBox pages for three different published papers. They all look the same, which is why it takes very few seconds to know what's available and where.
Click to expand...

Link

Another pre-registration site, that I had not heard of, is called AsPredicted.

The suggested use of ResearchBox also looks interesting.

Maybe as the culture of pre-registration spreads, scientific research might get better.
 
angrysoba said:
It’s been said that Publish or perish was replaced with publish rubbish and flourish.
Click to expand...
Another way to boost publication numbers is not so much to publish rubbish but instead, publish one set of research as multiple papers. Make what was previously one paper into a quarterly serial. So each paper is actually "good research", but only a fraction of the envelope-push in each one.
 
Puppycow said:
There are also many areas of science that don't in principle lead to any direct growth in productivity or any sort of direct economic productive value. Just growth in human knowledge. Like astronomy or cosmology or even particle physics. Do the findings of the experiments in the Large Hadron Collider lead to some sort of economically relevant increase in productivity?
Click to expand...

If there's something economically useful that requires the LHC to discover it, that might also require building an LHC to produce it. When scientific experiments become prohibitively expensive, that means it's probably less useful to any industry which doesn't have unlimited funds to invest (which is all of them). Don't get me wrong. I'm sure there could be exceptions to this rule. But when there are exceptions, we probably could have made the discoveries more economically, too.

So arguably, this mental model suggests that science is often outspending industry by far too much to be practically applicable, at least in the short term.
 
Last edited:
You must log in or register to reply here.

Back
Top Bottom