WHAT I DO
As science impacts our lives profoundly, not just in terms of policy, it is a challenge that deserves a response. So here it is. This is a long one, so grab a glass of your favorite libation and sit a spell. Or flip to another thread. I’ve been a published author for about 20 years. My first paper as a contributing author was in late 1991 and my first as a lead author was 20 years ago this month. I’ve got about 50 papers in archival journals and about 60 in conference proceedings. I’ve edited about a dozen conference proceedings and am an associate editor on a relatively new journal. I’ll put a caveat in that this is strictly a personal recollection. Your mileage may vary.
The first big distinction is between papers in conference proceedings and archival journals. A number of conference proceedings are not refereed, though the editor should be checking submissions. I’ve generally been responsible for editing proceedings. The first time was difficult. I set aside 6 months of my life to organize ICEL-2 (a conference series on organic LEDs that is still going strong). Working on the proceedings after all that work was tough going. Back in the old days, authors would submit physical copies of the papers when they arrive at the conference and the papers would be distributed to referees while they’re at the conference. Effectively, that meant a lot of papers were being refereed over a beer after sessions.
How do I approach a proceedings paper? I’m not going to put my best stuff in a conference paper. If I take the time to write a paper, a process that can run from weeks to months, it’s going into a proper journal. When writing a conference paper, I take one of several approaches. If I am presenting something already published, I’ll supplement the already published work with a few results that fill out the picture a bit more. I just wrote a paper like that for the SPIE conference this august. Many journals have length limitations (usually 3 to 4 pages) and so you have to think carefully about what to put in there. It’s easy to write a conference paper by padding what you’ve already published with a few unpublished results. It is important to note that graphics and text are protected by copyright, but data and ideas aren’t. It is possible to plagiarize yourself. This trap can be avoided by re-plotting an existing figure. Also, edit after copying and pasting. There’s always a different way to say the same thing.
Plan B is to write a paper based on preliminary results. If it’s a minor study that isn’t going anywhere, that’ll be it. In that way, at least the research isn’t wasted. I’m working on such a paper at the moment. That conference papers has a bit of this flavor to it. I did some basic modeling of a particular system. The main paper was published a few years ago, but the modeling hadn’t been done yet. I think it’s interesting, but not interesting enough to merit a full journal publication. So, into a proceeding it goes. I have to be careful if I have bigger plans for the work. I have had authors refuse to submit a proceedings paper on the basis that they didn’t want to jeopardize the main paper that hasn’t been published yet. Scientific double jeopardy. So, I publish a paper with some of the data, but won’t go into full detail. If you will, I show a little skin, but I’ll save the Full Monty for later.
I don’t want to give the impression that I consider conference proceedings a waste of time. Conference papers are a good way for graduate students to cut their teeth. It is also useful to have a record of what was presented. Sometimes, important results creep in. Companies often don’t go to the trouble of writing a full paper, they just issue a press release. They will, however, put something into a proceedings. One of my most hjghly cited papers was published in a proceedings. It was a preliminary report of a particular phenomenon (a phosphorescent polymer light-emitting diode). As it happens, it was also the FIRST report of this particular phenomenon. The paper got noticed and the snowball was off and rolling. It’s been cited close to 200 times. The amusing bit to me is that the paper was for a poster I didn’t present. I was called into substitutthe poster. I’d already written the paper, so I figured what the heck. I tossed it in the hopper and it went into the proceedings.
Even when refereed, conference papers are treated with kid gloves. The abstract has already been accepted for the conference, so it’s difficult to reject a paper based on the work that has already been accepted. There are a number of papers that I would reject as a referee if they went into a regular journal, but are acceptable for proceedings. In the event that a referee is being a stickler, the author is going to refuse to rewrite the paper just to have a conference paper. In most cases, it’s simply because the work lacks sufficient novelty. We try to avoid outright crapola, but I’ve had reservations about some stuff in there.
OK. On to so-called “real” papers. There is a definite hierarchy of journals. There are a few heavy hitters everyone has heard of. Science and Nature are at the top of the heap. There’s the Journal of the American Medical Association. Individual fields have highly regarded journals. In physics, that would be Physical Review Letters. Chemistry has the Journal of the American Chemical Society. Advanced Materials is a heavyweight in materials science. All of these journals are difficult targets to hit.
There are plenty of good journals out there. You need good work to get published there, but the odds are decent. I’m probably on about a dozen papers in Physical Review B (the solid state physics journal published by the American Physical Society). I’d put Journal of Applied Physics in this category. Organic Electronics falls there.
Then, there’s the easy A’s. Thin Solid Films, for example. I am amused when I read a paper that goes along the lines of we started with this really lousy device. Then we applied our special sauce and now it’s 10 times better! Yeah. Apply your special sauce to a decent device, kid, and then we’ll talk. Incidentally, the journal for which I’m an editor falls into this category right now. Part of the problem is that it’s difficult to start a new journals. We started by incorporating conference proceedings. You hope to get a few decent papers, get into a citation index, and eventually move up the quality chain.
Open access has made the situation worse. There are plenty of faux journals out there. Pay a thousand bucks and they’ll be happy to accept whatever BS you send to them. There was a celebrated case a few years ago where some folks used an automatic paper generator, which was accepted for presentation at a faux conference. If you have an axe to grind and a thousand bucks, there’s some Indian publisher who will be happy to accommodate you. I actually got an email recently from some “publisher” that wanted me to write a chapter for a book and pay >$1000 for the privilege. No. Definitively NO.
With all of this to navigate, how can you tell what a good journal? Citations are the currency of scientific publishing. A few years ago, something called the h-index cropped up. It’s the number of papers you have with that number of citations. I have an h-index of about 25, which means that I have a couple dozen papers that have each been cited a couple dozen times. The index is considered a way of weeding out one hit wonders, a paper like the first one on organic LEDs that has been cited thousands of times. If an author on that paper only wrote a dozen papers, they’d have an impressive average when it really just means they were on one big paper. Damon Huard (third stringer behind Brady and Bledsoe) has a two rings. I doubt anyone would mention him as a better QB than Dan Marino. The h-index also weeds out people who tend to write lots of minor papers. Papers with a few cites indicate productivity, but not impact.
It gets increasingly hard to increase your h-index the longer you keep going. I’d say a decent mid-career scientist will have an h-index in the 20 to 50 range. Big deal people go over 100. One tends to top out over time. If I don’t return to academia, I’ll probably top out in the 30s. Some people do well by collaborating all over the place. For example, the group of John Anthony of the University of Kentucky makes materials. He sends them all over the place and picks up co-authorships as a consequence. That’s legit to me. One can also pump your h-index by citing your own papers. The search indexes can suss this out, though.
Journals have their own version of this: the impact factor. In a given year, the impact factor of a journal is the average number of citations received per paper published in that journal during the two preceding years. Nature and Science have impact factors around 30. Now, the system can be manipulated. I was just reading a Wikipedia article on the impact factor (http://en.wikipedia.org/wiki/Impact_factor). One amusing case had the editors cite every paper in that journal, pumping up their impact factor. In another case, a paper practically begged to be cited and picked up 6600 citations, pumping the impact factor of the journal (Acta Crystallographica Section A ) above that of Nature and Science.
HOW I WRITE A PAPER
What do I do when I write a paper? It generally starts from a result that interests or excites me, preferably an unexpected result. I’ll give you an example. We’ve all seen phosphorescence. Takes some material, expose it to light, and it’ll emit for hours. The physical phenomenon is an optical transition that is only weakly allowed, so it’s slow. A strong transition (fluorescence) occurs on a time scale of nanoseconds. Phosphorescence goes from microseconds to seconds (or longer). The first phosphorescent materials used in organic LEDs were a big deal. The use of an organic phosphor meant a 4X increase in efficiency. The first paper with a phosphorescent OLED went into Nature.
We had some results in the lab which indicated the way people thought these devices worked wasn’t the case. That’s the ah hah! moment. Most people thought it was due to energy transfer (a process important for photosynthesis). Our results indicated that there was no evidence for energy transfer, yet the LEDs worked quite well. We could make a good case that the devices worked by charge trapping rather than energy transfer. As a consequence, the phosphorescence was much stronger than you might expect.
This was a paper fraught with trouble. The problem was that the paper had a bad structure. We first started with a whole bunch of spectroscopy that showed that energy transfer wasn’t effective. Then, we discussed all of the work on devices and came to our conclusion. Problem being that the study made it sound as we knew the devices couldn’t work, but we tried making them anyway. The paper languished on the desk of the big deal professor, until we finally had a meeting and hashed it out. What we figured out was that the paper was backwards. Discuss the device results first, then turn to spectroscopy to understand them. That was a good story.
Ultimately, good writing is telling a story. I might be describing the Förster theory of energy transfer, time-resolved THz spectroscopy or optically detected magnetic resonance. Regardless as to the technique, I’d better tell a good story to the people who care about it. The hardest part of the job is conveying what excites me to someone who isn’t a specialist in that technique. I’ve been working on a particular technique for quite a few years now. I think it gives us deep insight into devices, but I’ve gradually become convinced that it’s scientific marmite. A few people love it, but most don’t care for it.
There are plenty of papers that are, for lack of a better word, evidence of work. I’ve done a few clean-up jobs. In one case, we had a post-doc who left the group and had drafted a paper. It was one of those where he’d made a whole bunch of devices and gotten some mildly interesting results. It was at a time when we were changing the nature of the group (the previous head was forced out) and so had a dearth of publishable results. There was enough there for a decent enough paper. We didn’t come to any shocking results, but it’s still a decent addition to the literature.
PHYSICAL REVIEW LETTERS
This leads me to my latest paper. This is, to quote the current vice-president, a big fucking deal to me. A bit of back story, PRL is the top journal in physics. My first paper as lead author was a PRL. My PhD advisor wrote the paper. I worked my ass off for five months in the lab generating data. My second PRL came in 1996 as a post-doc. That one I wrote, though the basic ideas came from my post-doc supervisor. Mind you, I carried it through and wrote the paper myself. Still, it wasn’t my work beginning to end. I then joined a prominent research group as a lecturer (the British version of assistant professor). I did some great work there, but didn’t have anything that would rise to the level of PRL. I had some work around 2002 – 2003 that I did think was worth a PRL. It wound up being the most frustrating experience of my entire career. We submitted the paper and received two referee reports. Referee #1 said great paper, accept. Referee #2 said great paper, but too specialized. So, we responded to their reports and resubmitted. The paper was then sent to two more referees. Referee #3 said great paper, accept. Referee #4 said great paper, too specialized. Evidently, you lose a tie when submitting to PRL. We ultimately sent the paper to Applied Physics Letters and had a few follow-ups in Advanced Materials and Physical Review B. Generally good work that’s gotten some decent notice. It was always a disappointment to me that I never got into PRL on my own.
Some years later, magic struck. I made some samples for a group up at the National Institute for Standards and Technology (NIST) in 2008. The technique is time-resolved terahertz spectroscopy. Many of you have experienced THz in a different light (those scanners at the airport use THz radiation). As a technique, it is an ultrafast probe of conductivity. THz frequencies permit picosecond time resolution. Two papers on THz spectroscopy of organics were published in PRL in 2003 and 2005. They were wrong. Well, in my opinion. The problem was that specialists in this technique used theory that was developed for conventional semiconductors (silicon, germanium, gallium arsenide, etc.) Some of the conclusions that were logically reached didn’t make sense for someone such as me who had been working in organic semiconductors for 20 years. I butted heads with the lead author on the paper. He knew THz spectroscopy, but was resistant to what I knew about organic semiconductors. We published a paper in 2009 (in the Journal of Physical Chemistry B, a good journal). It’s been cited 4 times in 2 1/2 years (one a self cite), which pretty much means it was ignored.
The magic was a highly talented post-doc. He set up an experiment that allowed us to correlate the THz measurements with ultrafast optical absorption. My own work extended the time domain to milliseconds. I knew that we had the pieces for a great paper, but putting them together would be a challenge. It came together at Lake of the Ozarks of all places. I was driving back to KC to visit family over Memorial Day and stopped off for a couple of days at my brother’s vacation home. Hanging out there was great for thinking. He had a sun room over their neck of the lake and I had my computer with me. I had an idea and set up a crude model using Excel. Sure enough, it looked like the data we had. So, I wrote a somewhat more sophisticated program to model the results and was able to fit all of my data. Like it or not, a purely experimental paper isn’t going to stand the chance that one does with some theory or quantitative modeling.
Once I had a draft of the paper, I sent it to my co-authors. The discussions were, shall we say, spirited. Both of the THz specialists had problems with my basic conclusions. They knew the technique far better than I. Hell, I’m faking it when I give a talk on THz spectroscopy. I do know the photophysics of these materials cold. I knew that the primary papers applying THEIR technique to MY materials were wrong. Don’t get me wrong. It was important work. The problem was that people who knew the technique, not those who knew the materials, were writing the papers. In this aspect, my naiveté was an advantage.
We hashed everything out and submitted the paper in late August. At the time, I thought I had a really good paper. The referees’ reports came in late September. PRL requires three referee reports. All three referees thought the paper was of the level of PRL, but one in particular struck me. To paraphrase him or her, the referee commented that the authors didn’t realize what they have here. They have observed the intrinsic process of photoconductivity in organics for the first time in real time. The report blew my mind. I knew that previous THz work was wrong and we had a better explanation. The referee’s observation catapulted this paper from important to fundamental. I wish I knew who the referee was as I’d love to buy him or her a few drinks. Not for the positive review, but for seeing what I hadn’t. He or she is a better scientist than me.
Such a fundamental observation sparked a pretty thorough rewrite. I did a lot of reading and came across some interesting work by a Polish physicist who had done fundamental work on this topic. The reports from the referees to the revised manuscript were positive and so I have my first PRL in 15 years. Much more important is that this is the most significant work of my career. But, yeah, I bought a bottle of champagne when I got the official notice.
I’m trying to convey the passion that scientists bring to their work. At least, the good ones do. We’re human. We fight. We argue. Sure, I can write papers that check the box and make sure my productivity meets the standards. I could have written decent several papers over the last 9 months. I chose to write one that I thought could be great. Again, that’s my opinion of the work. If you don’t think you are doing great work or are capable of it, you shouldn’t be in the field.
WHAT GOES IN, WHAT STAYS OUT
I’ve deliberately decided to avoid some of those questions about handling data. There’s a great joke I saw years ago that was a translation key for papers. I loved the acknowledgements. We thank Dr. Richards for assistance with experimental work [Dr Richards did the experiments]. We thank Professor X for useful discussions [Professor X told us what they meant]. Typical result means the very best result. Good agreement means within an order of magnitude.
Nobody ever publishes all of their data. It’d be like hanging out with someone who suffers from verbal diarrhea. Many journals have strict length limitations. It forces hard choices. That PRL of mine will have three figures. I have two years worth of data. Mind you, insets are a way to cheat a bit. My new paper has three figures with four insets. Two are schematics, two are data. It’s tough to do everything in 4 pages. The authors, bylines, abstracts and references will take up a full page.
I’m reviewing a paper right now that suffers from a problem that the authors have tried to stuff it too full of information. It has four part figures with insets. There’s a great deal of supplemental information. They have taken what should have been a good 8 page paper and produced a four page muddle. I had one paper that I submitted as a rapid communication (4 page length limit). The referees didn’t think it merited special treatment and so the editors offered to publish it as a brief report. I refused, pulled the paper and resubmitted as a full paper (6 – 8 pages). I was pissed off at the time, but the referees probably did me a favor. The longer paper was much better than the letter (and is my most cited work).
Selecting what to put in a paper is the hard part. I will spend months or even years taking data. From all of this work, I reach your conclusion. I come to ideas and try to test them hard. I’m surprised from time to time (those are the best times). Here’s my opinion. Good work is when you perform a variety of experiments and reach a conclusion. Great work is when you then use those conclusions to make a prediction and perform the experiments to test that conclusion.
The process of selecting data to present is at the heart of one’s credibility as a scientist. Once you’re there, it’s time to decide what goes in. What effectively makes the case. Come to think of it, it sounds a lot like a presentation by an attorney.
ASK A SCIENTIST
Okay. I’ve put a lot out and could keep going on. I’ve already moved from a letter to a full article and should stop before it becomes a review. I’ll open this one up to the floor. I’ll answer questions about my own work and am happy to offer general opinions. I didn’t get much into specific controversies, but am happy to opine. Have at me!
Filed under: Uncategorized |