Lucent Reexamines Breakthrough Research

s20451 writes "Bell Labs' claims to have manufactured transistors consisting of a single-molecule switch are being met with skepticism in the scientific community, following difficulties in reproducing the experiment. Now a panel has been formed to investigate research misconduct related to not only that claim, but others regarding organic transistors." We've run several stories about the extremely tiny transistors and the innovative ways of assembling them which Lucent has been working on. A reader's summary of a subscriber-only story on Science's website suggests that there is strong evidence that some of the data in the published papers was faked.

Comments

Fake results

[brejc8]

It makes me sad when I see companys trying to hype up their research to pump up their share price.

Now when my group does any research which has positive results we are scared to release anything because everyone assumes its simply another con.

Currently we have an asynchronous processor which releases so little EMI it looks dead in the graphs.

We tried showing this to other people but everyone nowdays refuses to beleve anything unconvesional can be good.

Re:Fake results

[Alsee]

asynchronous processor which releases so little EMI it looks dead in the graphs.

Sounds like an obvious result to me. Asynchronous means the transistors don't fire at the same time, and they don't fire at a consistant frequency.

Listening to EMI from a normal CPU is like listening to someone pour 100 pounds of bricks.
Listening to an asynchronous processor is like listening to someone pour 100 pounds of talcum powder. You just get a low whoosh noise.

-

Before the bashing begins....

[CHUD-Wretch]

Just think where we would be without Lucent (well, Bell Labs in particular)....

They have invented, among MANY other things...
"the transistor, the laser and wireless technologies."

90% of the tech you love and can't live without originated at Bell Labs.
You know...computers...unix...voice communication...redundant/fault tolerant data networks...etc...

Oh, and for the patent lovers in tha house...

"Bell Labs averaged one patent per business day from 1925 to 1995,
and since March 1996, patents assigned to Lucent have been issued at a rate of more than three per business day."

(Disclaimer - I do realize this is off topic a little, but I want people to think about how much great tech comes out of there!)

Re:Before the bashing begins....

[Jeppe Salvesen]

Actually, good friends tell good friends when they think they're fucking up. That is why Europe and America are currently being pretty damned good friends, and we would be good friends of Lucent to tell them to straighten up.

The problem arises when the friends can't figure out which one is fucking up, and get annoyed with each other.

Re:Before the bashing begins....

[Zathrus]

The question is the same as it always has been - "What have you done for me lately?". In the business world that's largely what it comes down to. Of course, you also have to execute on innovation, which they've failed at too.

They were buying companies left and right and increasing thier already huge debt burdon. And then failing to utilize many of the technologies they bought.

Once upon a time Bell Labs was the leader in numerous fields, largely because they invented them. Now they're in second place or worse in virtually every field. They'll occasionally come out with something innovative and then someone else will do the same thing without violating their patent, for less, and with a better business plan.

Disclaimer - I do realize this is off topic a little, but I want people to think about how much great tech comes out of there!)

I'd also put that in the past tense.

Lucent is rapidly becoming the Xerox PARC of the 1990s/2000s - lots of nifty stuff which proceeds to rot or be taken by other companies.

Re:molecule size vs. atomic size

[Conspiracy_Of_Doves]

A bowling ball is not one single molecule. Try a diamond. That is.

Lucent!

[sulli]

You've got some 'splainin' to do!

More info of the fraud

[Anonymous Coward]

This was submitted yesterday to slashdot, but not posted for some reason...

For the past two years, a team at Bell Labs/Lucent, led by a young physicist named Jan Hendrik Schon, has published a dizzying array of groundbreaking work in the field of solid-state physics, which has previously
inspired discussions at Slashdot,
here
and here.
However, as reported tonight in Science (look under
the "ScienceNow" link), and I'm sure soon in Nature, it may all be a fraud. It looks like Schon has used identical data curves for very different experiments in different papers. The scale of the deception is enormous--there are duplicated graphs in at least 5, and as many as 20, papers. The fallout from this will be huge, not just for Lucent, but for the physics community as a whole, as a large number of these papers made it through the review process at the two most prestigious journals in the natural sciences, Science and Nature.

For a comparison of two plots from two seperate papers about two seperate experiments with remarbably similar data, check out here here. Scroll down to thursday may 16...

impacting

They just need to do a demo

[Waffle Iron]

It's too bad these researchers are meeting so much skepticism. I'm sure they could prove their results if they would just pull out their single-molecule transistors and show them in action. Unfortunately, I bet that somebody just dumped the molecules into a desk drawer, and now they're hopelessly lost amongs the crumbs, dust and fuzzies at the bottom of the drawer...

Here's the text

[Anonymous Coward]

Pioneering Physics Studies Under Suspicion

Officials at Bell Laboratories, the research arm of Lucent Technologies in Murray Hill, New Jersey, are forming a committee of outside researchers to investigate questions about a recent series of acclaimed scientific studies. Outside researchers presented evidence to Bell Labs management last week of possible manipulation of data involving five separate papers published in Science, Nature, and Applied Physics Letters over 2 years.

The papers describe a series of different device experiments, but physicists are voicing suspicions about the figures, portions of which seem almost identical even though the labels are different. Particularly puzzling is the fact that one pair of graphs show the same pattern of "noise," which should be random.

The groundbreaking papers include Bell Labs physicist Jan Hendrik Schön as lead author and his colleagues at Murray Hill and elsewhere as co-authors. Schön is the only researcher who co-authored all five papers in question. Everyone involved agrees that the questions need further investigation, but many fear that the impact could be devastating for Bell Labs and for solid state physics. Schön told ScienceNOW that he stands behind his data, and he says it's not surprising that experiments with similar devices produce similar-looking data.

Schön, who joined Bell Labs in 1998, has worked most closely with former Bell Labs physicist Bertram Batlogg--now at the Swiss Federal Institute of Technology in Zurich--and Bell Labs chemist Christian Kloc. His work has focused on efforts to make novel types of transistors using organic materials. He was the lead author on at least 17 papers in Science and Nature in the last 2.5 years.

Until this week, many physicists believed the impressive string of results was worthy of consideration for a Nobel Prize, although other groups have reported no success in reproducing Schön's most striking results. Last week, several physicists began to present their doubts to company managers. Bell Labs spokesperson Saswato Das says that company officials take the concerns "very seriously." Within hours of hearing of them on 10 May, Das says that Lucent management decided to form an external review panel chaired by Stanford University physicist Malcolm Beasley. Das says, "The panel will be given full freedom to make an independent review of concerns that have been raised." Physicist Paul McEuen of Cornell University, one of the first to question the data openly, says that Lucent is taking the right step: "Malcolm Beasley has great stature in the community. ... Everybody wants to get to the truth." --ROBERT F. SERVICE

Figure legend: Striking resemblance. Bell Labs is investigating a possible duplication of data in several publications. (* The author has corrected the bottom graph.)

no suprise

[greenrom]

This sort of thing really shouldn't be much of a suprise at a company like Lucent. Since the telecom winter started, they've slashed half of their workforce, and have been closing facilities right and left. You can bet that any research project that wasn't getting results got cut. This puts a lot of pressure on the researchers, and some will inevitably be willing to falsefy data in order to keep their jobs.

I'd be skeptical of any research done by Lucent in the last year, or at any other company with such serious financial problems.

Newsbreak

[MarkusQ]

...but physicists are voicing suspicions about the figures, portions of which seem almost identical even though the labels are different. Particularly puzzling is the fact that one pair of graphs show the same pattern of "noise," which should be random.

Lucent scientists today reported the remarkable discovery that, contrary to conventional wisdom and accepted scientific theory noise isn't random. Said one researcher, "We'd expected self-similarity, due to the fractal nature of noise, but this is amazing!"

Researchers estimate that there are actually less than a dozen examples of true noise, which are repeated endlessly through out nature. Some observers have expressed concerns over the fact that most, if not all of them are already copywritten by the RIAA.

-- MarkusQ

Bell Labs is not just one guy.

[Ungrounded Lightning]

They have invented, among MANY other things... "the transistor, the laser and wireless technologies."

But remember that Bell Labs is an institution, not an individual. It is composed of MANY scientists. It is not impossible that the barrel has acquired a bad apple. The trick is to find the bad apples and pull them out before they spoil the barrel.

Of course it COULD be that the research in question wasn't faked, with the anomolies coming from a clerical error, a jackpot, or a previously-undiscovered bit of physics. That's why they're INVESTIGATING, rather than just recalling the papers and canning those connected with 'em.

Value of research.

[Restil]

From what I can tell, this should mostly be an in-house issue. Let's say that these discoveries are frauds. That means they won't ever be used in applications, and Lucent will lose out. If no other scientists are able to replicate the process, then they won't be able to reap the benefits of it either. There is a REASON for this peer review. And it seems to be working. There are many possible reasons why things are turning out like they are, fraud being one of them.

However, its possible that the procedures involved are not trivial. Its also possible that either the procedures involved to produce or the procedures involved to confirm the findings are in error. Observing a single molecule is NOT trivial. It's certainly possible to think you've got what you were looking for, when in fact it sometimes takes another pair of unbiased eyes to take a different approach and discover that all is not what it appears to be.

I'm not saying this isn't outright fraud. The only problem is, what does it benefit anyone? If the scientist involved was pocketing all the research cash and running with it, I could understand. But if the research is legitimate, and nobody is able to actually profit from any of this without a working prototype in a useful medium, which may take up to 10 more years to produce, fraud would serve little purpose except give a black eye to the researchers AND Lucent.

-Restil

that's how science works nowadays

[dario_moreno]

Look at the space occupied by the Physical Review on the shelves of a library : went from less than 1 meter a year in the 1950's to maybe 20 meters nowadays...and the number of quality scientists has not increased 20fold ! At this time, it was enough to publish a paper once in a while, when a real discovery was made. When looking at the summary of a scientific journal of, say, the 30's, you see Fermi, Einstein, Brillouin, when nowadays articles are just a proof that someone did some work with the money he was given.

The review process has become a joke : either the paper goes to an indirect friend thanks to the editor (submit wisely !), and there is no actual review, or it goes to a concurrent, which makes irrelevant points (in one occurrence I know of, delaying the publication by more than a year making stupid points, and when all the objections were met, asking to change the units, and pointing minor misprints !). The referees usually do not understand the scope or actual point made in a paper, and make the stupidest comments possible (so one of my former bosses recommended to write papers in one afternoon, since the real mistakes would not be spotted anyways). This is also natural because they tend to be flooded by cut-and-paste papers from scientists who are in science only because there is some (ridicully small) money or career to be made, and they could not find a "real" job elsewhere. This is sadly true of the 3rd world, where scientists are underpaid (150$ a month anyone ?) and eagerly look for positions in developed countries, so need published papers, but their lack of money and bad education mean that they often submit utterly uninteresting papers.

This is also true from people under pressure from their supervisors because they are all on short-term contracts, so that they often resort to faking data to get the expected effect. A nice positive result created with the Gimp
(or vi data | gnuplot ) is way nicer than a boring negative result and easier to publish,
even if faked and wrong.

Sometimes the referees even resort to say "
please cite this guy", meaning, "hi, it's me,
hope you do not forget me when I need something or you refer my papers".