Slashdot Mirror
Big trouble In The World Of "Big Physics"
klevin writes "Hey, scientists are human too, who woulda thunk it? Nice bedtime reading for anyone who thinks science is an impartial search for knowledge and understanding. `Six months ago, Jan Hendrik Schön seemed like a slam dunk nominee for a Nobel prize. Then some of his colleagues started to take a closer look at his research.'"
... If you tie funding into results the way that it is in universities at the moment (OK, the guy worked at Bell Labs) then people are forced to chase and publish results. In some more theoretical fields it may be less of a problem, where there is more room for disagreement and differing opinions. In this case though, when you are publishing experimental results, then they either work or they don't. If you have people who spend their careers chasing money then temptation to take shortcuts is going to catch up with them, maybe thats why the guy published so much; perhaps he wanted to get caught out and pushing into the most prostigious journals and flooding the system will get you noticed.
Question for the physists though, the article was a bit scant on details, what did the guy claim he could do? How was he claiming to turn materials into semi-conductors?
Slashdot: where don knuth is an idiot because he cant grasp the awesome power of php
And as we all know, there's no such thing as scruples and honesty when "Big" is involved.
; -- the corruption of government starts with its secrets. a truly free people keep no secrets. --
"think of the untested O-rings on the space shuttle Challenger that froze stiff in the upper atmosphere"
The O-rings weren't frozen stiff in the atmosphere. Ice worked into a joint while the launcher was on the pad. The rupture happened even before the thing left the ground.
I think the Salon author was confused - what the group at Bell labs was working on (that I'm aware of anyway, though I'm not working in the field myself) were organic superconductors. Taking organic molecules, crystallizing them, and measuring their properties basically. What the Bell Labs group has a history of is being very good at crystallizations - Bertram Batlogg, who hired and supposedly supervised this guy, got some extremely impressive results from the crystals they made of the High Temperature superconductors in the early 1990's.
Energy: time to change the picture.
This is actually usually called "small" or "benchtop" physics, as opposed to the real "big physics" that goes on at accelerator labs etc. with hundreds of physicists working together. Making this worse is the coincidence with the bogus element-118 discovery at Berkeley, which was also revealed over the last few months. The APS, where I work, has some rules people are supposed to follow: the 1991 Guidelines for Professional Conduct - but investigation and resolution of problems (which happen more often in lower profile cases such as contested authorship of papers) are left to the institutions where the people involved work; it's starting to seem that perhaps more is needed.
Energy: time to change the picture.
It is a reasonable criticism directed at Science and Nature that they seem to compete with each other to publish attention-getting results (the recent bubble fusion experiment comes to mind), but what it comes down to is that a reviewer of a paper has no way to validate experimental data given to him. You have to take the research group at its word that the data are not fabricated. You can question their data reduction and analysis methods, but if they said they did this measurement and these are the resulting data then you have to take them at their word.
One of the ways science operates is that results like these are presented, and if the results are interesting enough (i.e., unexpected or never seen before) then other labs repeat and verify the experiment. When the results are confirmed, then great. If not, then the results (or at least the conclusions drawn from them) become suspect. This happened with cold fusion and it looks like bubble fusion is heading down the same road. This has happened in the past (N-rays are another example), and it will happen in many other instances that don't draw the big press stories. That is how it should work. The Salon article seems to suggest (among some valid points) that the paper reviewers should have had some all-knowing wisdom and immediately questioned the data.
I also doubt, as the article suggests, that the reputation of physicists has been harmed and that all over the world school children are crying "Say it ain't so Jan Hendrik." The biosciences have many many scandals related to data forging, or at least questionable massaging or analysis of data, because the stakes ($$) are much higher for a new drug to come to market as well as the difficulty in collecting consistent data. The biosciences continue to draw huge numbers of people into the field and it enjoys (deservedly) a positive reputation.
I also thought the article was way over the top with regard about the government funding aspect of this. It made it sound like that all the government money spent on R&D is a waste as it obviouly is going to charlatans and rouges. The author should have looked up the research dollar amounts in relation to the total government budget (such as its percentage of the GNP) as well as in relation to the total non-DoD R&D budget and see how well the NSF or the DOE compare to, say, NIH (I'll give you a hint, they are quite neglected). This isn't "Big Science" by any stretch of the imagination.
Undergraduate science students, in particular physicists are taught that to report results of experiments that don't work is just as important as the reporting of positive results. While this may be true for the pursuit of knowledge, it doesn't really fit with human nature. Since when have money, reputation and Nobel prizes been the reward for the scientist who proves nothing. Is there any other way to reward these people or should we just rely on their own sense of scientific truth to continue this work anonymously and relatively poorly rewarded.
It is sad when these things happen. A lot of people have put their careers in jeopardy by trying to keep up with these exciting "results".
Unfortunately, science is only a human activity so it is subject to all of the faults of people. The way the career and funding system works puts significant pressure on the shoulders of aspiring scientists and things like this will continue to happen. Fortunately, the peer review system managed to stop him in the end though it would have been a lot better if it had happened on day one.
The big dilemma is that science both has to be open to new (surprising?) results and extremely critical at the same time. Redoing an experiment can be incredibly hard and is always time consuming and expensive. A negative result is much too often no result at all.
virve
--
Science is an impartial search for knowledge and understanding. Falsifying the results of experiments is most definitely not science.
There's no place I can be, since I found Serenity.
No, I don't mean that researchers who falsify data are doing good science. But you'll notice that the falsification was caught. And it wasn't just the revelation that they included an incorrect figure and some of their plots had identical noise. Collegues have been growing suspecious of the results for over a year now because they've been unable to reproduce them.
This is good science. Scientists individually screw up all the time. I certainly have. Usually, we make honest mistakes. Sometimes, we make dishonest ones. But science is not and has never been about any one person or group. Science is a collective effort. It's not just the group doing the experiment, it's the other groups that try to reproduce it, the reviewers who look at it critically and the opponents who try their hardest to tear it apart. If you want to consider "good science", you need to add all of these into the picture. One of these segments clearly failed in this case (the original researchers) and another didn't catch it (the reviewers), the others did their job.
So, really, while the individual scientist was doing bad work, this illustrates exactly how science should work under real world circumstances.
Anyone who actually has done some science either in the schools or especially in industry knows that science is NOT an impartial search for the truth. Overall it is one form of inquiry that has matured over centuries but it's hardly perfect and possibly never can be. The notion that scientists and science is this impartial knowledge gathering enterprize is pure propaganda.
The history of the activity doesn't bear this out, the work experiences of scientists themselves doesn't bear this out, the public relations apparatus of science by way of popular press books hyping the latest and greatest doesn't bear this out, and actual survey data of scientists attitudes, as opposed to speculations about them, doesn't bear this out. When competing for fame or grant dollars, politics data distortion and hyping/fraud can play a role.
Of course this isn't everybody but it's not as small as some people, mostly those that haven't actually don't any basic science work, like to believe.
Science is a human activity with lots of compitition for prestige/ego and grant money. The problems of self-interest, whether of individuals or whole fields, is a very big part of all this.
Mechanisms that help around this are independent replication of results. However, this doesn't always work as it should either, in fact results can and have been accepted long before proper replications occur...if they ever do.
Some of his work involved molecular transistors. He reported results of successfully making a single molecule transistor. In fact, IBM set up an entire lab based on his findings. (oops). Not to say that these things won't be developed. Most of his "findings" are scientifically sound... at least the theory behind it, but with current technology it just can't be done (according to the 100's of researchers doing the same work). In a rush to publish lots of stuff, he re-used graphs with the scales hastily changed, and other undergraduate techniques for falsifying data.
Interestingly, The articles in question started only after Batlog's name started appearing on the papers (a reputable Bell labs guy). There were some other things he was working on. A lot of what he did at the beginning was quite real, people still believe a lot of his work on Pentacene.
In the end, my point is this guy was working on organic semiconductors used in all fields, as well as exploring the possibility of superconductivity in many organic systems. Hence the confusion.
The only believable explanation for this episode is one buried in the middle of the story: Schön has mental health issues. He's not just a con artist, he's a compulsive con artist, the kind that lies even when he doesn't need to, or is likely to come back and bite him. How else to explain his behavior? He must have known that nobody would be able to reproduce his results. It's a pretty common syndrome.
Of course, if you focus on Schön's mental state instead of on the "failure" of the peer review system you don't have a story!
I miss the good ol' days when scientists gone bad had an evil, ear-peircing cackle.
Now they just blend in like regular scientists.
Sigh
Table-ized A.I.
[inspecting scientist] found many duplicate graphs in different papers on different subjects. Schön was apparently using the same sets of pictures to tell lots of different stories.
He would have got away with it if he simply had a Mac.
Seriously, though, why leave behind such obvious clues? It is not that hard to generate new phoney diagrams. I am glad evil people are often just as stupid as the non-evil people.
How many are *not* caught because they don't leave behind such silly clues?
Table-ized A.I.
http://physicsweb.org/article/world/15/6/1
Another reason for the breakdown is the hypnotizing effect of reputation. When the names of eminent people and places appear on the top of submitted papers, says Florida physicist Hebard, "reviewers react almost unconsciously" to their prestige. "People discount reports from groups that aren't well known,"
One thing that could be done to cut down on prejudice within at least this part of the review process would be to remove names and other identifying information (like institution) from papers before they get sent out for review. This would allow work to be judged a bit more on it's merits and a bit less on it's authors.
Such a change migh both slow down frauds like Schon's and also allow new ideas more easily into the public realm. Papers don't just get erroneously passed based on name. Sometimes otherwise good papers can be rejected based on the fact that the author is either unkown or out of favour. Anonymous peer review would help mute both of those effects.
OS Software is like love: The best way to make it grow is to give it away.
I hate to burst your bubble but those days never really existed. What I miss are the good old days when scientists raped and pillaged their graduate students ... and got away with it ;-). We were more like the recent catholic priests but we didn't get caught. Oh those wonderful distant memories, lawyers have destroyed all the hedonism that was left in science. It's so sad :-(
That seems to me to be overstating things quite a lot. In addition to the new branches and disciplines you mention, small labs without monster budgets will also benefit from their own improvements in technology. Just because budgets remain relatively small doesn't mean that technological capability needs to remain stagnant; as an obvious example, processor speed seems to be a lot cheaper now than it was fifteen years ago.
But even more than the new scientific avenues that cheaper technology opens up for small labs, how can anyone say that all that's going to be discovered (without an enormous budget) has been?? It seeems to me to be very unlikely that more discoveries, even significant discoveries, are not just around the corner for small labs. It may not be as drastic as faster-than-light with household products, but surely there are implications even of what we already know that have yet to be fleshed out. We don't know what is going to be discovered, but that doesn't mean nothing can be--and besides, people have been saying that everything's been discovered already since ancient Greece, and probably even before then.
"Someone somewhere had to wear pants for the first time. The meek and indecisive do not change our world." -Montville
In a few weeks the results of the investigation will be published.
MURRAY HILL, N.J. -- Lucent Technologies Inc.'s Bell Labs said an independent committee investigating possible false data in published research found one scientist to be responsible for the fabrications.
Bell Labs said Wednesday in a statement that it has fired the scientist, J. Hendrick Schon, and cleared all others involved in the experiments of scientific misconduct. The experiments took place between 1998 and 2001.
The five-person committee of outside scientists and engineers found Mr. Schon committed misconduct on at least 16 occasions, mainly involving substitution of data, unrealistic precision and results that contradict known physics.
Papers by Mr. Schon and his team were called into question in May when other scientists found the experiments difficult to replicate and noted that paragraphs in several of the papers seemed similar. The experiments, involving molecular transistors, had been published in journals such as Science and Nature.
"The evidence that manipulation and misrepresentation of data occurred is compelling," the committee said. It added Mr. Schon "did this intentionally or recklessly and without the knowledge of any of his co-authors." A total of 20 researchers from Bell Labs and other institutions authored the approximately two dozen papers.
Bell Labs President Bill O'Shea said, "We are deeply disappointed that a case of scientific misconduct has occurred at Bell Labs -- the first in our 77-year history. Since Bell Labs' founding in 1925, tens of thousands of Bell Labs scientists and engineers have faithfully abided by the scientific honor code. That's an enviable track record, but we take this one exception very seriously."
In its executive summary, the committee said Mr. Schon "acknowledges that the data are incorrect in many ... instances. He states that these substitutions could have occured by honest mistake. The recurrent nature of such mistakes suggest a deeper problem. At a minimum, Hendrik Schon showed reckless disregard for the sanctity of data in the value system of science."
http://www.lucent.com/press/0902/020925.bla.html