Slashdot Mirror
How Did Those STAP Stem Cell Papers Get Accepted In the First Place?
bmahersciwriter writes The news team at the scientific journal Nature turns its investigative power on the journal itself. The goal: to try and understand how two papers that made extraordinary claims about a new way to create stem cells managed to get published despite some obvious errors and a paucity of solid evidence. The saga behind these so-called STAP cells is engaging, but sadly reminiscent of so many other scientific controversies.
Peer reviewed. Yeah, right. And just who is reviewing the peers?
And found out in short order.
As for what causes researchers to do this sort of thing, well it's the same thing that other humans do. The biggest difference is that science accepts it's lumps and corrects them.
You'll have someone feeling pressure to get results, and they fudge, make hopeful assumptions, or even fake results. This woman appears to have done all three.
Then she manages to get some highly respected researchers to sign on. Laziness on their part. The pee reviewers see the names, and get a little lazy themselves.
Viola. But in all this, the fact remains that others will be fact checking, and trying to reproduce the results claimed in any paper, especially one making these extraordinary claims And that's the strange thing. If you fake the results, you're going to be found out to a high degree of certainty. She toasted her career with a similar degree of certainty.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
It's probably not the case that she wrote the paper cackling to herself madly and proclaiming "Those suckers will never find out!"
It's probably the case that through self-delusion and carelessness she managed to partially convince herself that the results were true, and this, coupled with pressure to produce results, caused her to take a few shortcuts to get it published. What she did was wrong, and her career is over. It's not something a rational mind would have done. But scientists are just human and sometimes prone to making irrational decisions. The great thing is that we have the scientific method to weed out the good from the bad.
A fool and his hard drive are soon parted.
And where would the money to reproduce the results come from? Journals typically do not even pay peer reviewers, so you're essentially expecting them to use their own grant money and time to reproduce results knowing that it will get them precisely fuck-all in terms of career advancement. If that were actually the expectation the simple reality is that it would kill off the peer review system.
Seriously though, I would have imagined that the papers should only get published if the results themselves were reproducible. Somehow those are skipped and the whole peer review system is in trouble.
Actually, the whole peer review system is not in trouble. See, the peer review done by (volunteer) reviewers for the magazine is just the first step. The next step comes when the article is published and the entire world gets to see the paper. The fact that the fraud was exposed in pretty short order after publication shows that, indeed, peer review does work.
The biggest difference is that science accepts it's lumps and corrects them.
Yes but...
Science has bigger problems correcting them, and takes much longer to do so, when political and financial pressure tempt people to look the other way. Scientists are people too.
The biggest difference is that science accepts it's lumps and corrects them.
Yes but... Science has bigger problems correcting them, and takes much longer to do so, when political and financial pressure tempt people to look the other way. Scientists are people too.
Beg to differ. go to retraction watch.com.
As far as I can tell, there is no correction mechanism for politically based incorrect assumptions. Just as example - the job creator myth. It's still cited as gospel by many, while the evidence is clear it doesn't create jobs.
Galileo Galilei finally got forgiven by the Catholic church, when Pope Paul admitted the churches "errors" the church made - almost 400 years later. Regardless of political bias, I don't see either of the big two parties having a website of fact checked "We were wrong about this" stuff - at all - much less in a timely fashion.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
retractionwatch.com is great for the most influential journals but it's safe to say it's only catching a small minority of the infractions, and of those really only the ones published in english. One cynical solution: as grant money continues to dry up, scientists will realize that the easiest way to improve their chances to get their grant approved is to thin out the competition: exhaustively check their competitors' publications and old grant applications for plagiarism and fraud, then complain anonymously.
Until grants provide funding for reproducing old results, publishing a paper will remain the first step in getting the work reproduced. Typically by someone who needs the technique/result for their own work.
Perhaps one source of misunderstanding here is that some people assume that peer review is supposed to be some kind of ultimate validation of a work. It's not, it's a basic sanity check and a validation that what the authors *claim* to have done is sufficiently interesting. It's not an endorsement by the publication venue that the work is correct and the authors are honest, because it's impractical to validate something like that without investing a lot more effort than feasible for a peer review. Reproducing a work might take months, even years.
Exactly. This.
The road to scientist's Hell is paved with journal articles that eventually have been shown to be incorrect. Just because it's published doesn't mean it's gospel. It is Science after all, not religion - even if half of Slashdot seems to think they're one in the same.
Faster! Faster! Faster would be better!
Journals aren't arbiters of Truth (capital T), they're just what they say they are: JOURNALS of the ongoing work of science.
Someone records that they have done X in a journal. Because said journal is available to other scientists, other scientists get to try to make use of the same notes/information/processes. If they are able to do so, they journal it as well. Get enough mentions in enough journals that something works, and we can begin to presume that it does.
If only one mention in one journal is ever made, then it is just another record in another journal of another thing that one scientist (or group of scientists) claim to have done.
Peer review is just to keep journals from expanding to the point that there is too much for anyone to keep track of or read. It is emphatically NOT the place at which the factuality or truthfulness of notes/information/processes are established once and for all. That happens AFTER publication as other scientists get ahold of things and put them through their paces.
Seriously, this is all exactly as it is supposed to work. I have no idea why there is such hoopla about this. There is nothing to see here. One group journaled something, other groups couldn't replicate it, they no doubt will reference this failure in future articles, and "what happened" is recorded out in the open for all of science, thereby expanding our pool of knowledge, both about what consistently works in many situations and of what someone claims has worked once in one situation but appears either not to work in the general case or requires more understanding and research.
Again, there is nothing to see here. Let's move on.
STOP . AMERICA . NOW
The vaunted peer review - supposed to eliminate problems like this - failed.
Not really. Peer review is designed to catch holes in their logic or spot errors, such as if the incorrect analysis method was applied or if their scientific evidence doesn't fully support their claim. When it comes to outright fraud, a peer reviewer really has very limited means of spotting it. In exceptionally rare cases they will request that a claim be replicated by an outside researcher, but that is exceedingly rare and I don't think I've ever heard of a reviewer actually attempting to replicate research themselves as part of the peer-review process.
What normally happens is that other people in the field will read the paper and say "I don't really buy this" and attempt to replicate it themselves. If a consensus of groups can't replicate their findings, then the question becomes whether there was fraud involved or if it was just another example of "winnners curse" or maybe something unique about their study that was different from all the rest (like if they were looking at a different cell line or global population than everyone else). In no case is it really feasible for the peer-reviewer to catch outright deceptive fraud, but usually it gets spotted sooner or later. And the bigger the scientific claim, the bigger the bulls-eye becomes on your back.
As a former reviewer working for a very renowned research institute in Europe I can say: Peers typically don't get/take the time to do their job right, and often outsource the job to less experienced people. Reproducing results is a very expensive and time consuming job, which means: unless it is it won't happen. You must be lucky if the reviewers have at least read the paper till the end. Quite often the review happens by people who are "no experts" in the field of the paper. For many conferences, papers with a bad rating still pass because there are not sufficient good papers, or if it is easy to guess the institute the authors work for, the paper passes without proper review.
Once our institute had a paper rejected, but my boss -who was in the review team- managed to get the paper accepted anyway. High profile conference in Electronic Engineering.
As a former paper author I can say: If your paper is rejected for one conference, you simply resubmit to another until it is accepted. Publish or perish is the holy grail of research, something many bosses will make very clear to you, and quality is less important. You don't write a paper because you have results, you write a paper because this or that major conference has a deadline in two weeks. I have a few paper on my name I am ashamed of: Omitting the bad results in the measurements, compare with competitors only on the features you know you would win because the comparison doesn't make sense at all, bragging about results which are very bad, but you hide that by not comparing to (avoiding any reference to) competitors which are better.
As you might understand, I quit the job. I left research and never ever want to have anything to do with it anymore.
The issue is not this one journal. Its a general lack of scrutiny in science itself. They are not being audited. The data is not being checked. The experiments are not being replicated.
I believe you are mistaken, or alternatively I have a different interpretation of events.
No system is perfect, however the system works more or less as follows.
Scientists do work.
They are under pressure to publish because not unreasonably the people paying for the scientists like to see their money is going somewhere useful even if the metric is far from perfect. Anyway it generates strong pressure since science is up or out an unlike many other careers, if you don't get promoted, you're eventually fired.
Work is submitted to a journal or conference. Note in HEP (high energy physics) people just shove ut up at xxx.lanl.gov and send it to a journal something as an afterthought to satisfy the previous point.
Editors filter the papers. At low journlas almost every submission gets sent out to review. At very competitive journals, the editors strongly filter papers. It helps to know an editor to get your papers past this filter. Some journals (Nature) have paid editors. Most to not.
It gets sent out to other scientist volunteers to review it. The review is really a check of reasonableness. Nobody expects the reviews to find deliberate fraud. That's not the point. They're not there to check for actual correctness, only reasonableness. In other words no one excpects them to replicate the results. The reviewers have not much time. Some less scrupulous ones farm it out to students to review. Some are mad as a sack of badgers. Some will insist you cire them more. Others may be able to check some of the results for correctness if they look suspicious (I've done this before when I believe I've spotted a glaring error I can test). In some very simple cases (e.g. Magnesium Diboride as a semiconductor) people are able to verify the results before publication.
The previous two steps repeat until the paper is accepted or rejected.
Now the important part starts.
There are two main possibilities here:
1. The paper is boring. In this case it willprobably get a few citiations bulking up someone's lit review and nothing more.
2. The paper is interesting in which case people will try to build on the results.
And orthogonally:
A. The paper is right.
B. The paper is wrong (or bad or fraudlent or etc).
(in practice there is a continuum on both)
Basically in the case of 1 it doesn't matter if the paper is right or wrong. It languishes in obscurity and the rightness or wrongness never ends up having any bearing on the state of the art and "scientific knowledge".
In the case of 2, people will try to build on it and take it further and advance the state of the art because that's where progres is happening. And that's where the scrutiny starts. The more interesting the result, the more intense the scrutiny.
For results of the magnitude of cold fusion and high temperature scrutiny, every scietist and his dog tries to take a look. The results are very quickly found out to be bad (cold fusion) or good (High Tc).
This was a big result, and what you're seeing now is the scrutiny.
The journal peer review process is merely there to provide a good place to find new results and stop the scientific world getting flooded in a mound of utter dross. It's the very first part of the filtering process.
Once a result is public is can undergo scrutiny and generally this happens in the public eye with people posting comments, other papers, commenting on the internet and so on.
Scientific results in practice undergo more scrutiny than almost anything else, and you're seeing a little bit of it right here even on slashdot.
SJW n. One who posts facts.
Peer review works if the people doing the review are honest and competent. Both aspects have been in sharp decline, not only in the biomedical field. These days, positions for Professors and PhD students are more often than not filled with people that can simulate competence and that have no or little personal ethics whatsoever. They will form groups that accept any and all papers from each other and reject anything from others. Anything original also generally has a high chance of getting rejected, unless the reviewers know and like the authors. The peer-review system is so broken and corrupt that it has just stopped working as the quality of the "researchers" forming it is way too often abysmally bad. (And forget about "anonymous reviews". The in-group has all the Tech-Reports from their friends and can recognize all papers written by them.)
This is not a new phenomenon, it seems to just be getting worse again. But remember that Shannon had trouble publishing his "Theory of Information", because no reviewer understood it or was willing to invest time for something new.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
There are some islands of honest and competent conferences and reviewers, but they are usually in not very well known events. All that is mainstream, "Tier-I" conferences and Journals are fundamentally corrupt. I mostly left research for the same reason, but I occasionally still publish something these days. The difference is that I publish if I have something good and interesting, not when some stupid research administrator thinks I should have more papers. And I publish in a venue where I respect the people running the conference even if that gives a lot less scientific "reputation".
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
One very common scenario for knowingly faked "results" is this: PhD Student has his/her funding running out and gets set an ultimatum (explicitly or implicitly). PhD student fakes something, sometimes looking pretty good at first glance. Advisor is too stupid, lazy or full of him/herself to notice. Paper gets published because advisor is "respected" in the community.
I have seen this happening quite a few times, including one case where all authors, except the first one (the PhD student), wrote what was basically a retraction a year later. But did anything happen to these people? No. The PhD student still has his PhD, despite his results being essentially worthless. The other authors still have their reputation. The faked publications were not retracted. I did recognize the fake in 10 Minutes by numerous inconsistent things and numbers that did not add up and did not make sense at all. None of the reviewers apparently did. Just when people tried to reproduce the results and failed were some question asked. But as I said, no consequences for blatant scientific misconduct by several people. For me, this nearly cost me my PhD as my advisor was not even capable of understanding the fake after I explained it to him in detail and somehow though they were doing something vastly superior to my work. While the low-point of my scientific work, it made me understand that most so-called "scientists" do not qualify for that distinction.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Well, then whether something makes it through peer review or not should not be used as evidence that it is correct or not... merely interesting and reasonable.
I've decided to stop wasting my time responding to AC trolls/sockpuppets... so if you want a response from me... login.
Peer reviewed. Yeah, right. And just who is reviewing the peers?
Ha! I knew the denialists would come swarming out of the woodwork on this one.
Consider the stem cell paper that we're talking about here. It was published in January and immediately started going down in flames. Here we are six months later, watching scientists gleefully kick the cold corpse of the authors' reputations. And you're still wondering who keeps the reviewers and editors of a scientific journal honest?
Peer review isn't some kind of certification of a paper's truth. It can't reliably weed out misconduct, experimental error, or statistical bad luck. It's just supposed to reduce the frequency of fiascos like this one by examining the reasoning and methods as described in the paper. It doesn't have to be perfect; in fact it's preferable for it to let the occasional clunker through onto the slaughterhouse floor than to squelch dissenting views or innovation.
That's why climate change denialists still get published today, even the ones who disbelieve climate change because it contravenes their view of the Bible. Peer review allows them to keep tugging at the loose threads of the AGW consensus while preventing them from publishing papers making embarrassingly broad claims for which they don't have evidence that has any chance of convincing someone familiar with the past fifty years of furious scientific debate.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.