E=mc^2

When I heard about a new album by Mariah Carey titled "E=mc^2", I decided to come up with a rule, similar to one of those Bill Maher New Rules.

New Rule: you can use physics concept or equation in a commercial project (music, movie, book etc.) only if you are capable of explaining what it means. Somehow I doubt Mrs. Carey can.

sexism in hiring decisions

I often read blogs like those of FSP or YFS and I am quite shocked at the level of blatant sexism or chauvinism some of their male colleagues display. I assumed that by now sexism exists in a far more subtle form. Most of my colleagues seem like very reasonable, intelligent and polite people, and I just have real trouble imagining them telling female graduate students, for example, that they should stay in kitchen instead of going into science - or many other asinine and offensive things commonly attributed to male science professors.

During a recent faculty meeting we voted on the first batch of offers to new faculty members - both being women. Both votes to make offers to each candidate were unanimous (votes are by secret ballot) - which wasn't the case in a recent vote for a senior male professor where department was split. There was also a lot of enthusiasm and support for both female candidates in discussions prior to the vote.

I am not sure what if anything this single data point tells us, except this: even if there are indeed some blatantly sexist women-hating male professors, they appear to hide their feelings well - not only in closed-door meetings, but even when it comes to secret ballots.

Of course, in addition to sentiments along the lines of "women never get hired because faculty meetings are dominated by sexist males who vote against them", I also hear conspiracy theories from male candidates along the lines of "women get hired disproportionately more, because departments are under tremendous pressure to hire women (or minorities), even at significant cost to quality". I don't think this is true either - the pressures to hire top-notch candidates is still dominant.

Basically, I feel that the hiring process, while flawed and somewhat random in many ways, is actually a lot more fair and less biased than various groups of candidates tend to think. Or, it's more equally unfair and flawed than many people think, based on isolated examples of "Department X hired male candidate A instead of female candidate B, it must be due to gender" etc.

As a side note, recently there was a lot of discussion in the news about the conclusions that conservatives are more happy than liberals, even after correcting for how much money they make, their religion, marital status etc. Apparently it may have something to do with the fact that liberals spend a lot more time thinking or obsessing about fairness, whereas conservatives are more likely to zen-like accept the world as is - fair or not. If that's the case, perhaps I should spend less time obsessing over fairness in life, work, science etc. and this should make me a happier person.

Arkani-Hamed on front page of CNN.com

Randomly visiting cnn.com, I find the first photo on the front page, usually reserved for "breaking news" like cyclones or Hilary Clinton withdrawing from the race, is featuring a giant photo Arkani-Hamed.

here's a link, as I am sure they will replace the front page soon.

Business or personal?

Here's what I was pondering recently. Say someone gives you an award, and it can be either X dollars for you personally (a gift to spend only on yourself or your personal life) or Y dollars that can only be spent on science (basically a research grant). What is the relationship between X and Y that would make them equivalent?

Most people in other professions would opt for X, rather than Y no matter how small X is - or what ratio of Y/X is (show me a hedge fund manager that will turn down X dollars in his own salary to make hedge fund Y dollars richer, or a lawyer that would rather make money for a client than for himself). The end game in most professions is X, and the question is moot.

In science things are different - most PIs are truly ambitious about getting research done, and getting rich is not the primary purpose of doing science (otherwise we would all become hedge fund managers). But the relationship between X and Y is not simply "I will always pick Y no matter what X is", or even X=Y. I would find a lot more use for $1,000 of personal gift money than $1,000 in research grant. This is because I can't do much for $1,000 in terms of research - this is a drop in the bucket. But for $1,000 of personal money I could make myself a lot more productive, for example by hiring a maid so I don't have to clean my house, or eating out more often instead of cooking.

I think the relationship is closer to 5X=Y or even 10X=Y, or maybe even higher?
It's not because I am greedy, it's simply because research is expensive and my life improvement is cheap in comparison.

What is your Y/X? Does it vary for various values of X - say from $100 to $1,000 to $10,000 to $100,000?

Retention offers

Related to "why so few mid-career job searches" - I admittedly don't know much about this, but talking to some people it appears one way to bump your salary (beyond standard ladder increases that probably don't keep up with "real" inflation) is to get an offer from some other place and make your department make you a retention offer (or counter-offer?) to stay. According to some sources, a large part of budget increases that could have gone to new hires, goes towards this retention business.

Just how commonplace is this (any more senior folks want to weigh in)? And is this maybe one of the reasons that mid-career searches are not so commonplace - you go through the search, make an offer, just to see it being used as a leverage to get more salary, less teaching etc.

I have more important things to wonder about (like getting tenure), but I am just curious. Apparently it's typical to go on the job market the year your tenure is being decided and use any offers you may get to leverage for tenure. Of course it could backfire too - some people may be offended and call your bluff by saying - "fine - Go!".

Am I too naive in my thinking that if I simply do greath research and all other stuff, I don't have to play these games of getting offers from other places just to get noticed by my colleagues in my department?

Robert Reich on China

From Freakonomics blog: Robert Reich answers your questions

Q: What steps, if any, can the U.S. take to consider staying on par with China — in terms of economic growth capital, G.D.P., etc?

A: China is still playing catch-up with the United States and other advanced nations. It’s relatively easy to chalk up fast growth when you’re not the technological leader. All you need to do is follow the leader!

But that’s no reason for complacency on our part. If we want to stay in the lead, we’ve got to invest far more than we now do in basic research and development (green technologies, for example), as well as education (starting with early childhood).

Amen to that!

students who cheat

since Doug has mentioned plagiarism, I have to attend a hearing in case of academic misconduct in the course I taught. The student was confronted with clear evidence of blatant cheating (worse than plagiarizing a report, in my opinion - if not caught, it could have also affected another student's grade), but yet the student denies denies denies and requests a hearing.

This means I have to meet with ethics coordinator, meet with dean, meet with my TAs, prepare all the pertinent documents, attend a hearing at which I need to testify. Not a big deal (for me personally), and I feel strongly enough that justice must prevail, but I wonder how many other professors would feel as strong, especially if this is their 50th case of cheating. Would some of them hush-up blatant cases of cheating, reprimanding the student and avoiding all of this administrative measures? If they do, they only encourage students to cheat more.

Also, it seems a no-brainer for students to deny - the student in my case is facing 1-year suspension. So he can accept responsibility and serve 1-year suspension, or he can deny and worse case scenario he serves 1-year suspension. So many appeal and ask for a formal hearing. Apparently from statistics engineers and MBAs cheat more, and also more likely to ask for a hearing - thinking they are smart enough to get away with it, even after being caught. Maybe deans should offer plea bargains, just like on Law and Order. Accept your responsibility and get 9 month suspension, and save everyone a lot of time and effort. Or you can fight it, request a hearing, and if found guilty, serve 18 months suspensions - your choice.

There will be blood

We finally got around to watching this movie on DVD. Mrs. Ponderer hated it, even though she said she would like to read the book (Upton Sinclair's "Oil!"). I was bored throughout this 2hr40min movie that was for the most part excruciatingly slow, but in retrospect, the more I think about this movie, the more I begin to like it. Mrs. Ponderer accuses me of "pretending" to like it just so I can appear fancy and sophisticated, but I can assure you this is not the case - I hated plenty of artsy movies that critics loved and I told everyone about it (case in point: "Rushmore" or "American Splendor").

To those who wonder why majority of successful professors are jerks, go see this movie. The main character is so driven by his competitive streak ("I have a competition in me"), to the point of disregarding normal human relationships he has with his son, half-brother (not going to spoil anything here) or people in the Little Boston.

I am not saying EVERY PI must have a mean competitive streak in them, but the ones that do get selected out to go on to become successful more often than those who don't approach things competitively.

Again, I don't want to ruin what some people call a rather bizarre ending (I thought it was quite typical and within what the Daniel Day Lewis character behaved like in the past), but it appears that he was tired of constantly being driven by this competitive nature and he probably had developed a pattern of self-destructive behavior to stop himself from competing.

Anyways, it took me a while, but I thought that the main character could have been written as a composite of professors I knew back when I was in grad schools, including my own advisor. I have a feeling a lot of them were not liked even by their real families, never mind "academic" families (grad students, postdocs, etc.)

"let me know" filter

I have tried something I read recently, where you star (gmail) or flag (Outlook, Thunderbird) an email that has the phase "let me know". So far it works remarkably well to weed out hundreds of emails that are addressed to me, but not REALLY addressed to me specifically.

Mail trends analysis

Google has released something called mail trends, a set of scripts that can analyze your gmail usage and plot all kinds of statistics. I recently move most of my old archived email into gmail account (so I can quickly search for past emails by sender/subject/keywords) - I have about 50,000 messages, dating back to 1995 - even though I am missing a span of about a year during my postdoc years. It may still exist in backup somewhere.

I also used to have email from as far back as 1993 or 1994, but it wasn't archived.

Some conclusions: there was little emailing back in 1995 and 1996. Or even 1997. A big explosion happened in 1998 - I have received about 4,000 emails that year, an increase of a factor of 5 over 1997. Maybe it's because I didn't archive everything back then, who knows. But then the numbers are increasing only very gradually, with about 6,000 emails in 2007 (I was getting an average of 500 emails a month last year). This year the average is about 800 emails a month, with last two months in 900-1000 range. I might be on track to get 8000-9000 emails this year.

All of these numbers are already spam-filtered, but there are still numerous mailing lists that I am sure contribute significantly to the averages. For example, apparently I was encouraged to play frisbee and volleyball on weekly if not nearly daily basis while in grad school - a fact that I have no recollection of nowadays. It's also interesting to look at those old emails, there are some emails from people who went on to become quite famous but I never realized I was in the same department with them during my grad school years.

Plotting daily totals is interesting too. Nowadays I typically get 30-40 emails a day Mon-Fri, with highs going up to 80 on occasions (I got over 90 emails in one day a few days ago). But on weekends the numbers often go down to single digits, or at least 10-20 range. You can clearly identify which days are weekends on the chart.
Averaged by day of the week, I get about a factor of 4 less email on Sat or Sun, with both days being about the same (I hoped to see the effects of Sabbath from my jewish collaborators).

Hourly email usage varies quite a bit too. From 10PM to 5AM it stays constant low, with dramatic uprise around 6AM. The ratio of maximum to minimum is about a factor of 8. The curve looks like a gaussian, with a visible dip around noon when people break for lunch. Email starts to decline around 3PM, but gets a secondary shoulder-like bump around 7PM, while overall declining into 10PM. The curve is quite asymmetric, with a longer tail towards PM hours, but sharp rise on the AM part.

You can do all sorts of other analysis, like top senders and recipients, length of thread, size of messages etc. Pretty cool stuff for geeks like myself.

discrimination/bias study suggestion

Since there was a lot of discussion regarding Fermilab gender bias study, which was even commented on in Nature, some of the criticisms mentioned were the small sample size and the validity of the productivity metric (number of internal reports) in high energy physics collaborations which often hundreds if not thousands people large.

Someone could do a fairly straightforward analysis using the existing rumor mill information (CM-AMO, Astro, etc.). First identify the candidates being interviewed at various schools and compare their publication records (h-index may not work well for recent publications, but a modified metric where each paper gets weighed with its impact factor, with bonus doubling of IF for first-author publication would probably work fairly well). For example, Nature or Science get 30, Nature Materials/Physics gets 15, PRL gets 7, PRB gets 3 etc, double points for first author. Maybe include h-index as well for good measure.

Then you come up with some metric for performance on the job market: for example using department ranking (interview at Kansas State is not the same as interview at MIT). I would focus primarily on interviews (which are more "blind"), but one could add job offers as part of the metric.

Then one would have to demonstrate that the chosen metrics (which are somewhat arbitrary) work, by showing a strong correlation between the productivity/publication metric and the performance on the job market metric. This is a crucial step - otherwise one could prove just about anything.

There will be a lot of statistical noise, as anyone who was ever on academic job market surely knows. But after averaging things out for large enough samples I suspect the trend will be clear - people with better productivity record are more likely to get interviews (and offers) from more and better ranked places.

The next step is to look at the value of correlation for sub-groups, such as women or for example foreign-sounding names vs. american-sounding names. In other words, do women with the same productivity metric as male counterparts have statistically significant loss in terms of job market metric?

One could similarly look into how pedigree affects job market metric (does Harvard PhD perform the same as Kansas State PhD given similar productivity metric?). And which metric is best correlated with job market performance - number of publications, number of citations, publication or citation per year, fraction of first-authorship, publications in top-tier journals, pedigree (grad school vs. postdoc), h-index, etc?

Looking at rumor mills has some advantages over looking only at how many former postdocs got job offers in academia. First of all, a significant portion of students and postdocs (especially in high energy, based on my personal anecdotal evidence) opt out of academia on their own, in pursuit of jobs in finance and consulting. It would be inaccurate to count them as those who tried to get faculty jobs and failed based on their record, because they never tried or aspired to get those jobs. Limiting ourselves to rumor mill or other similar interview data means the sample size includes only those who actively seek these types of jobs.

Looking at interviews as opposed only at eventual positions increases the size of data sampling by at least a factor of 5-7, and eliminates many other issues. For example, top applicant X interviewed at MIT, Berkeley and Columbia, but decided to accept an offer from Syracuse due to closeness to family or 2-body issues. Looking merely at where candidate X ends up may lead us to conclude they were discriminated against, since based on their record they should have ended up in top-10 university, but looking at interview/job offer record would show a different picture.

In these type of study one has to pay special attention to "error bars" (as in everything else). What are the uncertainties and what if different metrics were used?
One of the criticisms of the Towers work was that if one looked at the simplest metric possible - 16 out of 48 male postdocs went on to get jobs, 4 out of 9 female postdocs got jobs, one could draw a very different set of conclusions. Granted, looking at merely job offer outcome percentages may be flawed, for the reasons mentioned above, but at least it's more straightforward than a metric that may or may not have influence on how job offers were made. In other words - if the outcome of the study is not robust, and instead strongly depends on the type of metric used, then any conclusion should be taken with a grain of salt.

graphene, STM and CM-AMO rumor mill observations

Graphene is still going strong as a research topic, by this time I expected the hype backlash reaching the point when even very good papers have problems being accepted at top-tier journals because of graphene-fatigue, but no.

I will make a prediction that Kostantin Novoselov will end up with a big offer from a top school very soon. I am surprised he hasn't been lured away from Manchester yet.

Looking at CM-AMO rumor mill the other day, it appears that top candidate (by far) is Pasupathy, an STM guy from Yazdani group. This is what happens more or less every year - there's someone from Davis (or Yazdani) groups who has a lot of offers. (The same thing happens with AMO if you replace Davis/Yazdani with Ketterle or Colorado/NIST groups). But so far the former disciples of STM has failed to catch up to their former advisors - Davis and Yazdani crank out papers left and right, while their students and postdocs struggle with a lot of various issues.

Some will argue that STM/STS is an extreme example due its highly technical nature, but there are other similarly technical and novel fields where new assistant professors are often as successful as their former mentors.

This brings me to the following question: is it really so reasonable to expect a young scientist to be very successful simply because they worked for a big-name advisor? In other words - in some groups (not implying anything about Yazdani/Davis/ Ketterle groups, just a very GENERAL observation!) students or postdocs are almost glorified technicians, and being a lead author on a few key high-profile publications does not mean that this person will be successful at building the lab from ground up.

Of course all things being equal, why not hire someone from Davis or Ketterle groups (again, just typical examples of top people in the field!), those with intimate knowledge of STM or BEC tricks of the trade? It's hard to come up with argument for why not, except pointing out that not everything can be easily "copied". Otherwise we would have 30 groups in US routinely producing new BEC and STS results, as opposed to just two or three.

One interesting consequence is that despite groups at Harvard, MIT, Colorado, Stanford, Houston etc. having their own STS disciples of Davis group, had their instruments produced as much science as Yazdani or Davis labs, the field could have easily saturated and departments might consider it unwise (in terms of cost-benefit analysis) to hire yet another STM person to compete with these super-groups. In other words, if success is really difficult or impossible to duplicate just by hiring former postdocs in a certain field, this only leads to MORE hires in the specific field. If success is easier to duplicate, the field saturates quickly and leads to less new hires.

Tenure track advice?

Everyone on blogosphere seems to be discussing "how to get tenure" topics
(see here, here and here).

Now, I have no advice for anyone and I don't pretend to know much about the process. A lot of it seems like a black box, or a meatgrinder of sorts - some stuff goes in, then something mysterious happens, then something else comes out.

But here's my thoughts on the process: first off, it really depends on the type of department we are talking about. A lot of blogs seem to mix everything together in tenure=tenure=tenure=tenure. Obviously, requirements for tenure in history or computer sciences may be very different from those in physics. Condensed matter physics experiment may be different from astrophysics or string theory. Community college or liberal arts college may be different from research oriented university, and top 10 department may be different from department ranked #89. For example, in some departments publishing 4 papers in low-key journals over 6 years, would be considered worthy of tenure. In others, publishing 60 papers with a large fraction in Nature, Science and PRL is not enough for some reason.

1. Funding: In most research-oriented departments, funding is the key - having a lot of funding allows someone like me to accomplish other things (large group, solid publication record, etc.). Completely and utterly failing in this category is very difficult to compensate for.

2. Publications: this is next on the list - it's hard to argue against the solid publication record, and it's also difficult to compensate for poor publication record in other categories. There are multiple parts to publication record - roughly separated into "quality" and "quantity", more on this later. A large part of success in terms of publication record can be traced back to success in getting funding.

3. Reputation of being world-class. Since department solicits letters from outside experts, it is important to have good reputation in the field. A lot of it comes from publications, which can in turn be traced to funding. The easiest thing to do is simply do good science.
It may also be helpful to keep in contact with people who are experts in the field (attend conferences, workshops etc.), and it may be unwise to piss them off (a general rule that applies to everyone - including collaborators, department colleagues, editors and referees, grant managers etc.) In other words, it's important to be famous but that "fame" should be good fame, rather than bad fame.

4. Niceness: Making sure that most colleagues are on your side. No matter how good a scientist you are, it doesn't matter if most people in the department don't want you as their colleague.

5. Management skills: This includes being a good advisor and mentor to graduate students. If you are not a good manager, there will be less publications and more grant money spent for less results. Also, good students will not want to join your group. But it's somewhat tangential.

6. Teaching and service: In my department the chair openly advises new faculty that teaching doesn't really matter, as long as you are not terrible at it. Basically, if you fail one or several of criteria 1-4, it doesn't matter how good of a teacher you are. If you ace criteria 1-4 but suck at teaching, most will consider the situation "fixable" with some teaching training. Same goes for service - even though junior faculty are not put in charge of anything terribly important anyways. However, most people who are ambitious enough to get funded, publish a lot and run good research groups are rarely bad teachers, and are conscientious about teaching and service. Conscientious, that's the right word to describe it.

I am not sure what the right balance between 1-6 above, or if I am missing anything. Perhaps wiser older readers can comment. Right now my ever-evolving tenure evaluation system looks like food pyramid, with funding, publication and reputation forming the foundation (vegetables, fruits and grains), and the rest being the icing on the cake (literally: sweets, ice cream etc.). Being exceptionally good at 4-6 is no good if you are merely in the middle of the pack in 1-3. On the other hand, being exceptional in 1-3 will help to overcome some (but not the most egregious) deficiencies in 4-6. In that sense, 1-3 are primary parameters, while 4-6 are tie-breakers.

Of course, your mileage may vary, and the criteria outlined above are suitable for experimental condensed matter physicists in a primarily research-oriented department. In small liberal arts colleges the criteria may be reversed, with teaching, service and overall niceness taking precedence over research component.

Iron Fever

Doug has already blogged about new family of non-conventional Fe-As based superconductors.
See here,
here,
and here.

I am not working on High-Tc stuff, but here's my (tongue firmly planted behind my cheek) observations so far:

* Originally discovered in Japan, most of the recent stuff on these compounds is coming from China - or at least samples made in China. Why is that?

* The materials are supposedly easier to dope (is that even true?), but more difficult to make (due to presence of As). Is this one of the reasons why chinese scientists are more successful at making the compounds?

* Experimentalists are divided into two groups: those who are trying to make it in their lab and openly admit it, and those who are trying to make it in their lab but do not let anyone know they are working on it. However, if you hear explosions coming from their lab and they come out smelling like garlic, you know what they are working on.

* Garlic-smelling experimentalists (of both groups as per description above) are also seen walking around the halls randomly asking each other if they know where to get some FeAs samples, preferably from China.

* Theorists (who are not limited by having to actually make samples) immediately publish 100 papers claiming their favorite theory has predicted it all along. This proves that: a) it's often better to be first than correct b) theoretical graduate students can produce papers on arxiv within a few weeks of actual work, if they really wanted to. They have to be careful, because it now means they should be able to produce close to 100 papers in 5 years of their thesis. Granted, those papers are likely to be primarily "we could have predicted this" Monday morning quarterback type papers - but then again, this is what most theory papers are nowadays, no?

Why so few mid-career job searches?

I just learned that a colleague of mine moved from a so-so, maybe 50th or 60th ranked small department into a top 10-15 department or so. He has had tenure and was a big fish in a small pond for quite some time, and in fact I am surprised why he didn't jump ship earlier.

I was wondering about a broader question for a while. The ratio of junior faculty searches to the senior/mid-career searches (at least according to rumor mills) is typically something like 10:1. Why is it so high?

I am mostly not talking about super-senior faculty members, but mid-career, or faculty who have recently got tenured someplace else. Unlike junior searches, these searches are easier to conduct - no need to sift through 400 applications - there may be only a dozen or two. And even those are easier to evaluate - in case of junior faculty there's always some uncertainty - will they be successful in getting grants? Advising students? Teaching? Running labs?
In case of more senior (just past tenure) scientists these questions are easy to answer with the existing record.

So why doesn't faculty shuffle happen way more often? I have no idea, as I am still learning about these kind of things. One possible reason is that it may be more difficult for prospective ship-jumpers to openly apply for these positions. If you are a little unhappy with your department, or think you deserve more (internal funding, prestige, labspace, less teaching) - you could approach some people in other departments, but what if they won't make you an offer, and your current department finds out? This could make things uncomfortable.

A lot of time there are many-body considerations - by the time you solve a many-body problem, it is difficult to make yourself go through the brand new solution yet again.

Finally - there may be some backlash against hiring mid-career faculty on administrative level. It's relatively easy to justify hiring someone fresh out of postdoc - they are a possible new young star (whether or not this is the case remains to be seen, but it's hard to object to this statement). And if they are a mistake, they are a mistake that lasts 6 years. For very senior hire, the salary/startup package may very well be much higher, and we are talking about full tenure, so the stakes are higher too. If we are talking about Nobel Prize winner caliber or even National Academy of Sciences member, there are no questions. But other than that, there may be some resistance to the idea - both on administrative level, but also from the rank-and-file faculty (this person gets $1 million in startup - then why not me, we are at the same stage in our careers?!).

However, I still think there ought to be at least a factor of 2-3 more faculty shuffles than currently are, and just like in professional sports and other arenas, open market means more equal pay and opportunities (so state schools like Urbana or Berkeley won't be able to get away with paying a lot less than ivy leagues like Harvard or Princeton).

on discrimination

"This is the worst kind of discrimination. The kind against me!" --Bender (Futurama)

How does one tell discrimination from just "shit happens"? For example, are people who don't want to vote for Hilary Clinton (like myself) secretly sexist, or is it that they simply don't like the particular politician Hilary Clinton who happens to be a woman? What about people who don't like Obama (racism?) or McCain (age-ism?).

I think that too often we immediately look for discrimination as some sort of knee-jerk reaction when the explanation is far more prosaic.

Okham does a full in-depth analysis of the paper on arxiv that looks into discrimination in academia hiring. A very thorough post, but I will add the following: a sample of 9 people is way, way, waaay too small to claim anything substantive. We might as well use "I know a friend of a friend who claims..." type of anecdotes. Second: I have no idea how high energy people attribute credit, with hundreds of people in a typical collaboration - so they picked the worst possible sub-discipline of physics.

Someone could have done a far better job analyzing the applicants by looking at rumor mills - with larger sample size than 9 people, that's for sure. At some point I tried to run some numbers by ranking candidtes with "publication scores" which is the sum of the weighted the publications by approximate impact factors (h-index or total citations are not very useful) - the results were quite well correlated with the outcome of hiring (candidates with high scores got multiple offers from top-ranked schools). We could all argue as to what constitutes a good metric of scientific performance (or promise of thereof), but I suspect it doesn't matter much in the end.

Random thoughts and compilations of ponderings

Today is April 15, and in addition to being a deadline for filing tax returns, this is also when senior undergraduate students face with difficult decisions of which graduate school offer to accept.
Sean over at Cosmic Variance has an excellent post on this topic. I distinctly remember visiting campus as a perspective graduate student and a professor who ended up my PhD advisor telling me that "choosing a research group is like dating - you need to feel the sparkle".

Meanwhile, as I was attending our department's open house, I was mistaken for a perspective student - despite wearing a jacket. Not sure what I can do, hopefully grey hair kicks in soon enough. Then again, I got carded today buying beer, so maybe I just look like I could be 21. My students were asking my TAs if I was a "real professor".

Angry has an interesting post - especially the part about mood fluctuations. Most people can't stay very happy for too long, and research clearly shows it - it's not how much stuff or money we have in absolute terms, it's how much well off we are relative to our neighbors, friends etc. that really matters. People in "super-rich" countries are not much happier than people in well-off but not super-rich countries. Of course people in third-world countries are not as happy, but they have all kinds of other problems - but basically beyond certain standard of living money doesn't buy you happiness, it just makes you work a lot harder to keep up with the joneses.

I had some good news in my professional and personal life, and the lifetime of feeling good about my life is a few days at most. At the same time, bad news can make me feel bad for a lot longer than that. But then again, I probably wouldn't have been where I am now if it was the other way around.

The part of Angry's post about logos in presentations: I have a hunch - it's all about "branding". Managers and administrators go to seminars where they are taught about value of branding - you have to "sell" your name, and logos is one way of doing it. Rankings, such as US News & World Report, clearly show the benefits of branding - schools that have recognizable names will be ranked high no matter what they do. This leads me to believe that experience of going to those schools for grad school or even being a faculty is not as pleasant as it may seem based on the perceived reputation.


Slate has an interesting article on lack of available bachelors based on the game theory. Basic premise is that not-so-attractive women out-bid attractive women by being more aggressive in their bidding process. Because women hold all the cards in dating/marriage issues ("proposals" are mere formality, and even though men propose, the entire proposal process is still controlled by women), this leads to lack of available men and surplus of available hot women. Not sure if I am buying this theory, but sounds somewhat plausible.


Washington Post has an intriguing article on a social experiment by Joshua Bell, a virtuoso violinist who can easily sell a full Boston Symphony Hall with the cheapest tickets going for $100, playing in the metro station in Bethesda - most people walked by without recognizing the "true art", and Joshua Bell earned just $32, with only 37 out of more than 1,000 people giving him money for his performance.

Someone ought to do the same experiment in science. I guess Alan Sokal already did something similar. Would complete rubbish of a paper get accepted in respectable journal, if it was submitted by a big name scientist? And, reversely, would a good paper by a great scientist go unnoticed if it was supposedly authored by someone anonymous from Wichita Community College?

random ideas for research

here's some random ideas for non-physics, almost freakonomics-style studies:

1. People running into other people on the street while texting, checking emails on their crackberries/iPhone, playing with ipods etc. How many collisions are there? How many happen on college campuses where demographics seems to favor a considerably higher concentrations of ipod zombies? Is the number of collisions proportional to square of density of zombies?

2. With fancy glasses that track your eyeball movement we are learning how different people access various information. For example, men seem to check out others crotch area, apparently - both for pictures of animals and people, like baseball players.

What is interesting to me is how people read manuscripts. In what order do their eyes move? My guess is: title of the paper, first author, last author, affiliations, abstract, figures, abstract again, conclusions, captions to the figures, introduction, the rest of paper.

3. What criteria do people use in establishing collaborations? Why do some people collaborate better than others? Why do some students or postdocs work out in some groups but not others.

4. What kind of "non-malicious" lies do scientists tell each other? For example, I know one guy who claims NOT to work on a project, where I know for a fact he DOES actively work on it. Another one claims to have been offered a position that he also claims to have rejected, whereas his claims are directly contradicted by people within the department. I suspect this stuff happens routinely at internet rumor mills. What other misleading information are we feeding each other? And what is the primary motivation for this and other behavior of questionable ethics?

spending money

One of the grand challenges that assistant professors face is spending money. This may come as a surprise to some - what's so difficult about spending money? A lot, actually.

No matter how small or large startup package is, everyone wants to use it wisely.
Should I spend most of it right away on several big pieces of equipment, or should I wait? If I wait, I can think it through carefully and will be less likely to make major mistakes in terms of optimizing the lab equipment. Maybe I don't need some things, maybe we can build some in-house, and maybe there will be better technology on the market, and better deals. Maybe dollar will be stronger a year or two from now and I can actually afford to buy from european vendors again. But every day that I wait means a day that we don't get to use the equipment, and meanwhile tenure clock is ticking...

What is the proper balance between equipment and salaries? How much can my startup last before completely phased out by support from big grants? There is no overhead for expenses paid from startup, but there's a substantial overhead for grants. But there is no overhead for capital equipment either. But it's also difficult to spend a lot of money on equipment from grants, while startup money have far less restrictions on how its spent.

To those who think spending money is easy, I can guarantee it is not. Making these type of decisions and constantly second-guessing my own decisions results in all kinds of anxieties: should I have bought this used laser or a motorized stage from eBay for just 40% of the price of the new one? It would save me a lot of money, but what if it's just a piece of junk? Then I am just throwing money away. Or what if it works for two months and then breaks? Should I pay technicians from company do the installation and repair, or ask graduate students to learn how to do these tasks? Time is money, but how much is time actually worth? A talented student can do this task in one month, but a not-so-talented and unmotivated student can drag it out for a year - but what kind of graduate student do I have? Nobody, including the graduate student in question, knows yet.

I literally lost a lot of sleep over these type of issues - and still do.

My most expensive single purchase prior to starting my faculty position was purchasing a car (which was under $20K). Then all of a sudden someone gives you freedom to spend more than 10 times that on capital equipment, with total startup budgets typically in 500K-1 million ballpark (at least in my field of experimental condensed matter physics). My friends in chemistry departments get well over a million, often approaching two. And Mrs. Ponderer, who is a biologist, tells me stories how they used to spend even more than the chemists - especially during NIH vast expansion years. They had multiple vendors treat their entire group to free lunches or dinners every couple of weeks - all I typically get from my vendors is a mousepad, a mug or a pen. But basically startup is worth several times more than an average house - except you know precisely what you expect from the house - whereas even with the fanciest piece of capital equipment nobody can guarantee you will be doing great science.

What makes these purchasing decisions even more difficult is that the transition to grants is very difficult, considering the last two years of continued resolutions and budget cuts. Future funding is a big question mark, and it's difficult to go into "stretching out" mode after spending majority of startup on capital equipment. So I am constantly hedging by bets on multiple fronts, except hedging is not the way to do cutting edge science - at some point one has to make the leap of faith and take some big risks.

When I was still a postdoc, a friend of mine who was an assistant professor some 4 or 5 years more senior than me was telling me how the startup is basically "spent" as soon as you start - most money is "spoken for" or actually spent within the first year. But some of my current (since long-tenured) colleagues tell me they still have some leftover money from startups that were given to them over a decade ago! In current climate when even well-established veteran faculty members complain about difficulty in getting funding it probably makes good sense to not spend all the money - so there's some left to pay students and postdocs and keep the lights on in the lab. But it also means not doing as much science as we all could have, if we could spend the startup right away (as was originally intended), without trying to stretch it out for 5 years or so.

Lab computers: personal or work?

FSP's student has been using his personal computer for research, and now it's broken - who should pay for repair?

I am surprised at the range of responses: from "how dare he ask?" to "of course he should not pay for it himself".

There are several issues here: let's forget about backup issues for a second. Say no data is lost - would it be fair to ask for repair? My answer is Yes.
I use my computer (paid for through grants) for personal use, such as typing up this blog. I suspect FSP does too. But she doesn't want to purchase her student a computer because she is afraid they will also use it for personal use?

If FSP bought a laptop for the student, she would also have to pay for repairs. So FSP saved 2 grand or so on original purchase and is trying to save more on repairs - all at expense of a student. One of the commenters is correct - it's equivalent to asking students to pay for purchasing research equipment out of their own pocket, and then requesting they pay for repair of said equipment as well.

Some people argue that if FSP was concerned that the student would use the laptop for personal reasons, she should have purchased a desktop instead. First of all, use of desktop does not preclude use of computer for personal reasons (I did a lot of that when I was in grad school). But most importantly - getting someone a laptop instead of a desktop is good for everyone - you get improved productivity, because the student or a postdoc can now work anywhere - home, office, plane, cafe, etc. It's like coffee - the best money you can spend to improve efficiency.

Now we get to the backup - I am always amazed about how little some PI's care about data backup. As experimentalists, data is the most precious asset we have. This is why we do experiments - so we can have *data*!

Blaming the student for not backing the data up is just not right - the students don't know any better. As a PI, it is MY responsibility to make sure data is backed up - pointing fingers after the fact won't help. Saying once or twice during group meetings "Make sure your data is backed up" is not sufficient. Make SURE the data is backed up. And then check again. Ask to see the backed up files. I have multiple backup options (hard drives, network backup, data on remote servers) - you can never be too paranoid about accidental loss of data.

As a final verdict - I have to disagree with FSP on both counts. Employees should be provided with equipment to do their jobs - and students are no different. IF they decide to use their own money (we all know how much students make) to buy computers, the least PI should do is offer support in terms of fixing/maintenance - it's only fair as we just saved a lot of $$$.

As to backup issues - the buck stops with PI.

Why I remain anonymous...

the reasons are pretty well summed up in this piece on Chronicle Careers.

US News releases 2009 rankings

With NRC rankings of PhD programs delayed - yet again - till Sept. 2008 (they should just have a permanent sign - "To be released in 6 months"), I just noticed that new 2009 US News rankings are out.

In Physics Department rankings there's absolutely no movement in positions among the top 10 - I even thought for a second I was looking at old rankings. But comparing scores to 2006 (most recent) listings, it appears about half schools got a bump of 0.1, while about half didn't. Which is not terribly significant.

But there were a few more significant movers: U Texas Austin got downgraded by 0.2, moving it from 11th to 16th on the list. Minnesota and Northwestern got upgraded by 0.2, as did UC Davis, North Carolina, Arizona State, Iowa State and UMass Amherst. Indiana and Florida State got downgraded by 0.1. The rest more or less held their scores/relative positions.

Looking at Condensed Matter Rankings, the situation is similar - the top 10 is basically the same, except UC San Diego now ties Caltech for 10th. But the order is another story.
Urbana is still #1, MIT moves from #4 to #2, Cornell falls from #2 to #5, UCSB moves up from #6 to #4, and Stanford goes from #5 to #6. I suspect all of these moves are within the noise.

Unfortunately I have only access to top 10 listings for sub-specialties.

The paper issue of US News with rankings will come out on April 1. No kidding.

Making Mistakes

I am not sure if this quote, attributed to Steve Chu, is correct or not: "When you start something new, make as many mistakes as fast as possible" - but this has become a kind of a motto for me lately.

Maybe not taken too literally (thinking and planning DOES help eliminate a lot of mistakes!), but it seems to me that a lot of times the worst thing you can do is do nothing. Doing the wrong thing often leads to doing the right thing (eventually), and no amount of over-thinking can lead you to this path: as experimentalist you just gotta try a lot of things and make a lot of mistake before you can figure things out. The key is of course to learn from mistakes, and move out of the "everything I do is a mistake" phase as quickly as possible.

In Exponentially Annoying News

I am irritated by recent news that less than 1% of flights have air marshals. Not because I think this is too few - I have no idea what the right number is. But the news outlets continue to exploit public hysteria about risks of flying. Maybe the most important thing prior to these "news" was that public thought the number was much higher than 1%. And I would have liked to keep it that way. We already have to rid our bags of simple and complex fluids and gels, spanning essentially an entire class of soft condensed matter. We have to take our shoes off, remove our laptops, etc. - and I strongly suspect none of these measured will be taken back, even though they are completely unpractical and unnecessary. Who are these people who think that taking off your jacket or shoes (or throwing away a bottle of drinking water) realistically deters terrorists getting on the plane?

But what caught my attention is TSA press release declaring that there are "exponentially more than 1%" flights with air marshals on them. What does that even mean?

The general use of word "exponential" is getting out of control. I am trying to guess whether it is used figuratively or literally. For example "The number of foreclosures has been increasing exponentially over the past year" - what does that mean? Has it really been increasing exponentially, or is it just a figure of speech?

Charlie Gibson's "$200K a year" comment, revisited

Chronicle Careers has an article about Charlie Gibson's comment about the faculty couple making $200K a year.

This mirrors some discussion we had here and here.

Numbers are higher for research university faculty, but still low, especially compared to what other professionals with advanced degrees make in law, medicine, business and engineering - often with less total education/training/work experience.

In bizarre science news

So apparently Doritos are staging a contest for best commercial, with the winning ad being sent into space. Very bizarre (and possibly brilliant marketing strategy), but what attracted my attention the scientific group involved in beaming up the ads in space is EISCAT, which stands for European Incoherent Scatter Scientific Association, practically the name of this blog.

Tipping

I think academia should institute tipping. Students should tip 15%-20% of the tuition that they already pay to the professors, based on quality of service they provide.
I am planning to keep a tip jar in my office hours (I am joking, as I am actually trying to lure students to my office hours with candy and other sweets).

Seriously though, tipping in general makes no sense. In case of restaurants, tipping is often justified as paying for quality of the service, which is on top of the bill - where you pay for the quality of food. Food bill is not negotiable, but service bill is up to you - even though the minimum is supposedly 10%, while 25% is a really generous tip.
However, the quality of food, or at least my perception of it, varies more dramatically than the service. And I am much more picky about the food than the service. So how come service (tips) is negotiable, and food bill is fixed? Maybe we should be allowed decide if the food was up to my standards and decide to deviate from the price chef charges by 10-20%?

But at least there's some justification in case of restaurant and coffee shops - but what do I tip for in case of taxi driver or hair cutters, where the original bill is for the service (getting me from the airport to convention center, or reducing the length of my hair by a quarter of an inch)? I pay for labor, and I have to automatically tip 15% - for what? I carry my own bags, and don't particularly care if the driver is nice - in fact I would prefer to avoid banter and be quiet. I expect a certain standard of service - not driving like a maniac and talking on the cell (but even those expect 15% tip! - and I often give it to them) - or not being stabbed with scissors in the neck. But do are we expected to pay extra for this? Do we also expect to pay extra for our steaks to be NOT poisoned or my bank account not being emptied by a bank employee?

Another example - there's a popular pub on campus which serves beer - if I order two $4 beers, I often leave $10, which is a 25% tip. The service component here is handing me a glass of beer, and I don't see what constitutes nice vs. not-nice service. We tip hotel housekeeping staff even though they clean the room when we are not there. If they didn't clean your room, you would complain to the management - as it's part of the service you pay for in your hotel bill. We do NOT tip receptionists at the front desk, even though those are the people who are nice to us (or not). If anything, housekeepers often wake me up offering cleaning services at strangely early hours, and I still feel obligated to tip them.

If the standard for tipping is the human element being involved, there are a lot of professions where we do not tip - cashiers, for example, bus drivers, doctors, lawyers, car mechanics, etc. Then again, maybe the difference between mean cashier and nice cashier is not all that important to customers. But when I deal IT, administrative assistants, financial managers - I value human interactions and service they provide, yet I never thought of tipping them. I certainly don't want to be treated by mean or moody doctor - as to moody driver - I don't care too much as long as his driving skills and style are adequate.

Students dealing with mean professor who mumbles through the lecture and avoids students during office ours would probably prefer a nice friendly professor who takes time to explain things, so does it mean they should tip good professors? This makes a difference - as opposed to handing me a glass of cold beer with a smile vs. handing me the same glass of similarly cold beer with a smirk. But we tip for beer, not for knowledge.

What are the situations when someone actually DOES something nice to you? Helping you with directions when you are lost, pushing your car out of a ditch, help you put the bag in overhead compartment - yet we never tip them.

This whole tipping etiquette is applied completely arbitrarily.

Upgrade your life, Winners Don't Punish, Gender Gap

Lifehacker has compiled "Update you Life" feature with the best of the lifehacker posts. Some posts are more useful than others, but I am sure everyone will find something to improve their productivity.

As I was reading this paper titled "Winners Don't Punish" in Nature, I thought to myself about how it applies to scientific collaborations. Angry Physicist has immediately picked up on this in an excellent post.

In Science however, the system provides little benefit for thorough purging of free-loaders (do you really care if your paper has 8 authors or 9?), but such purging has a substantial price associated with it (hurt feelings and broken relationships).
The direct costs of keeping a freeloader in a collaboration is very low, but has a potential payoff in the distant future, if the freeloader proves to be useful somehow.

Indirectly though, the dilution of collaborations reduces morale of participants to the point of killing off the entire project when nobody wants to contribute unless freeloaders start contributing first.

Finally, Freakonomics Blog has a post on Gender Gap in Education. What I find fascinating, in addition to the topic itself, is that when people read statements along the lines of "Studies show that Group A is better/worse than Group B at activity X", the reaction ranges from sarcastic "well, duh!" to hysterical "Outrageous!". Someone could do a study of media/public reactions to how very similar results are worded, or issues involved.

Finally, along the same lines, Stuff White People Like
is a hilarious satire, not so much of white people, but of yuppie upper middle class folks. I find that majority of things on that blog applies to me and people around me, but it's still quite funny. A lot of people get outraged at this website, by the way. There used to be a similar website on Lincoln Park Trixie Society - latte-drinking, Jetta-driving, Kate Spade-bag, North face jacket, chocolate lab owning shallow yuppie girls who all want to conform to the same taste in everything.

Supersolidity in "Big Bang Theory"

A few minutes ago, as I was watching CBS' "Big Bang Theory" where one of the science dorks presents a paper on supersolidity at a Bose-Einstein condensate conference. There was only one slide visible, and I immediately recognized as Fig. 2 from the Moses Chan's Science paper E. Kim, M. H. W. Chan, Science 305, 1941 (2004). Using Google Desktop Search I immediately pulled up the paper to show Mrs. Ponderer.

She was not impressed.

(I also told her that the joke about spherical chickens is supposed to be about spherical cows - and she justrolled her eyes at me!)

Does anyone else watch that show?
Update: here's some screenshots from yesterday's episode:

Tenure at MIT article

via Dr. Shelly, a Boston Globe article on tenure process at MIT, titled Tenure at MIT still largely a male domain.

Dr. Shelly picked mostly on the fact that tenure rate at MIT appears to be less than 50%, and I agree - I think this is in fact the real scandal that Globe could have focused on - people spending 10 years on post-graduate education (PhD and postdoc) and another 6 years cranking out research, still fail over 50% of the time - even though these are some of the brightest minds (and MIT can afford to be among the most selective on already competitive faculty job market), and they have access to plenty of resources. I think it says a lot about arbitrary nature of tenure process at places like MIT. Imagine successfully negotiating 2-body problem for 16 years of graduate school, postdoc and junior faculty years, just to find you have to relocate yet again in your late mid-to-late 30ies.

But I have several problems with the article. The article, starting with the title, creates an impression that the tenure process at MIT is sexist, granting tenure to males and denying it to women. Let's examine this premise a little closer:

The first paragraph states that

"Just one out of 25 faculty members granted tenure this year at MIT is female"

- but does not specify what fraction of tenure applicants were female.

The small sample size of looking just at this year is another problem in using these numbers - and as we will see, the average numbers are closer to 1 in 4, rather than 1 in 25.

Second paragraph says:

"Women have been achieving tenure at a lower rater than men at the Massachusetts Institute of Technology during the past 10 years, according to an MIT analysis of junior faculty. Of the tenured faculty, 16 percent are women, up from 10.5 percent a decade ago, but still too big a gap, several professors said."

The first sentence is supported by statistics that will not be listed until much later, and as it turns out the differences are not statistically significant. Instead the authors decide to follow it up with statistics on percentages of ALL tenured faculty, the numbers that are more reflective of the past 4 or 5 decades of hiring practices - these numbers are cumulative and therefore rather insensitive to the current state of affairs at MIT.

Then the article goes into several interviews which do not say anything substantial, except address the issue of gender parity in academia in generalities. Paragraph 8 brings statistics back into the story:

"Between 1997 and now, the number of junior faculty women granted tenure has ranged from zero to eight a year, according to data provided by MIT at the Globe's request. The number of junior faculty men granted tenure ranged from 10 to 24 a year over the same period."

Again, these numbers are completely useless, they would be better off giving total numbers of tenure cases over the past decade, classified by gender and their outcomes. Giving ranges instead of mean, medians or totals is bizarre.

The next paragraph finally gets to the only meaningful statistics in the whole story:

An MIT analysis of junior faculty who could have vied for tenure during the last decade found that 41 percent of 104 women were granted tenure, compared with 48 percent of the 372 men hired.

These differences are actually not as large as I would have guessed - I would have assumed a lot more female faculty decide to bow-out of tenure-track prior to tenure decisions due to 2-body problems and biological clock issues.

But how significant is the difference between 41 percent and 48 percent?

43 women out of 104 got tenure, and the error in counting statistics for N=43 is sqrt(N)~6.6. This means the 41 percent number has 5 percentage point uncertainty.
Uncertainty of 48 percent value will be smaller, but overall it's hard to argue that the difference between 41 and 48 percent is statistically significant - the difference of 7 percentage points is about the same as one-sigma due to small number statistics.

I think a more significant number that should have been mentioned is that 22% of faculty that could have vied for tenure were women - a number higher than PhD enrollment numbers in many disciplines (such as physics). An obvious question that could (should?) be asked - what are the differences between the makeup of faculty that are hired on tenure-track (qualified to vie for tenure) and those who actually DO vie for tenure? In other words, who drops out and why?

There's more statistics in the article, which I would expect to totally confuse an average reader by conflating number of total tenured faculty with the numbers for recent hires, and the statistics for doctoral programs at science and engineering-heavy places like MIT/Caltech vs. those that have less science&engineering emphasis.

Overall, I am not sure the article addressed the right issues, and often misrepresented or misused the statistics. The impression that a layperson would draw from this article is the numbers like 1 woman for 25 male faculty, 10%, 16% tenured female faculty at MIT etc. - and the problem is primarily a sexist tenure process at MIT, which denies opportunities to females.

In reality, differences between tenure outcomes for male and female faculty over the past decade are not that different.

The real reasons for why recent faculty hires at MIT are only a quarter to a third women (actually a fairly high number), instead of a 50-50 split, are more complex: Comparison across disciplines - physical& engineering vs. biological sciences, vs. liberal arts are likely to reveal some of the underlying differences.
Looking at how many women enroll in PhDs or major in sciences and engineering at undergraduate level is another useful bit of information.

Another meaningful statistics is to look at makeup of entering PhD students vs. those who get PhDs, male/female ratio of job applicants for faculty positions vs. the male/female ratio of those who are offered jobs, and then follow it up the pipeline to see how many drop out before tenure and how many do get tenure.

So while the real questions are - why so few women pursue careers in science and engineering - with a long pipeline that includes high school, undergraduate, graduate school levels, these type of news article use "fuzzy math" to paint a picture where it almost seems like gender bias in tenure decision is what results in 25 male faculty and just one female faculty at MIT (using this years numbers that do not seem to be representative of slightly more long-term averaged trends), and the article seems to send the overall message that eliminating this gender bias will solve all problems and immediately restore tenured faculty to 50-50 ratio overnight.

If anything (and somewhat ironically) these misleading articles and the unattractive picture they paint are one of the reasons why female students may be discouraged from pursuing academic career paths.

As a disclaimer, I personally have plenty of reservations about academic career paths for entirely different set of reasons, but that is neither here nor there, as far as these articles are concerned. I also think the real scandal is why MIT or places like MIT that can afford to hire just about anyone, fail 50% of their faculty. Obviously those MIT-rejected faculty members can relatively easily go to second-tier (or even first-tier) schools and get instantly tenured there, but it still means a lot of stress of relocation and uncertainty. The unpredictability and arbitrary nature of tenure process at places like MIT is the real problem that very few seem to address in mass media.