Thoughts on business, engineering and higher education

Rejecta Mathematica

The online version of the Economist has an interesting article about a new journal called... Rejecta Mathematica (Huddled Maths, July 29, 2009). From the article: "As its name suggests, the new journal publishes only papers that [...] have been previously submitted to, and rejected by, others." The Economist mentions several famous articles that initially suffered a rejection:

• Paul Lauterbur's work on magnetic-resonance imaging, for which he won the Nobel Prize in 2003. See the Wikipedia pages on Lauterbur and MRI, which has become a common technique in radiology, as well as Lauterbur's 2007 obituary in The Economist and in particular the first sentence: "The whole history of modern science, Paul Lauterbur once joked, might be written on the basis of papers turned down by academic journals." The whole obituary is worth reading, but is available only to Economist's subscribers.
• Peter Higgs's work on the particle that now bears his name, the Higgs boson, which has yet to be observed but is supposed to explain the origin of mass. See the Wikipedia pages on Higgs and his boson. In this interview with Times Online, Higgs explains why the paper was rejected back in 1964: "I hadn’t said enough to convince anybody that this was really important physics. So I added on some extra paragraphs." The journalist continues: "The revised article – still less than two sides of A4 paper — was accepted by an American journal, the Physical Review Letters at the end of August [after being rejected by the European Journal of Physics Letters]."
• (This one is not mentioned in the article itself but in the comments.) The seminal option pricing model developed by Fischer Black and Merton Scholes, discussed in this Wikipedia page, was rejected by both the Journal of Political Economy and the Review of Economics and Statistics before finding a home, in a revised version, in the Journal of Political Economy. (Source: Advanced options trading by Robert Daigler pp.128-129, as shown on Google Books.)  

The author of the "Huddled Maths" article explains: "The serious aim [of Rejecta Mathematica] is to highlight papers that, although perhaps flawed, may still be interesting." I was a bit skeptical at first: a lot of papers are rejected during the first round of reviews; sometimes they have severe weaknesses, and sometimes they are simply not ready for publication at that stage but will be accepted once they have been revised. Breakthrough papers in particular might have been written in a rush so that their authors could submit them before a rival came up with the same idea; some reviewers have a low tolerance for that kind of behavior. When first drafts are not rejected outright, the reviewers often recommend a major revision. Is this worth a new journal?

Then I did look up the table of contents of Rejecta Mathematica and was surprised to see a name I knew: Stephen Boyd, a professor in Electrical Engineering at Stanford University with a long and distinguished track record in the field of convex optimization, who gave a plenary talk at the ISMP (International Symposium on Mathematical Programming) conference earlier this week and is the co-author of a popular textbook I use in some of my classes. His article in Rejecta Mathematica, co-authored with his student Siddharth Joshi, is on "subspaces that minimize the condition number of a matrix."

The most enjoyable part of the journal is that authors must submit an open letter describing the previous rejections. Joshi's and Boyd's is quite entertaining. In particular, they write: "The first rejection was based on the reviewers and editor noting that someone had written a paper that seemed to cover similar material. But a cursory reading of that paper, and ours, show that while the other paper shared a few key words with ours, the results were in no way related." And later: "We then submitted the article to another journal. In this case, the editor apparently did not even understand the result, which is stated very clearly, in completely standard, and elementary, mathematical language."

Maybe the real impact of Rejecta Mathematica on the field of mathematics will be to push for a better reviewing process in other journals.

Data Mining Competition

FICO is sponsoring the 2009 Data Mining Competition hosted by the University of California at San Diego. Here is the information from the website: "This year's contest consists of two classification tasks based on e-commerce transaction anomaly data. The first task is to maximize accuracy of binary classification on a test data set, given a fully labeled training data set. The performance metric is the lift at 20% review rate. The second task is similar to task 1, but provides a couple of additional fields that have potential predictive information." There's a total of $8,000 in prize money.

The contest runs until July 15 and is open to "all current undergraduate students, graduate students, and postdoctoral researchers studying full-time, in residence at an accredited university or college." Students rarely have the chance to manipulate real data and the contest offers them a unique opportunity to see what the real world of operations research looks like. Participation is very easy. This contest sounds like a lot of fun!

The Dark Side of Academia

I've just read a fantastic article from the 2007 Best American Science Writing collection, "Manifold Destiny," by Sylvia Nasar and David Gruber, which originally appeared in The New Yorker. It provides a fascinating account of the controversy surrounding the proof of Poincare's conjecture, a result about three-dimensional spheres that had stumped scientists for decades until an extremely talented but eccentric Russian mathematician named Grigory Perelman solved the problem.

Here's why the story warrants an article in The New Yorker: the Chinese mathematician Shing-Tung Yau, who has achieved super-stardom after his breakthroughs that "helped launch the string-theory revolution in physics and earned him, in addition to a Fields Medal - the most coveted award in mathematics - a reputation in both disciplines as a thinker of unrivalled technical power" (he's the Yau in the Calabi-Yau manifolds), tried to gain credit for the proof and push Perelman to the sidelines. He is portrayed in the article as a power-hungry, recognition-obsessed man who, according to a professor at Stony-Brook, "wants to be the king of geometry. He believes that everything should issue from him, that he should have oversight. He doesn't like people encroaching on his territory." (My only source of information on the matter is the New Yorker article. I tend to find the articles in that publication well-written and well-researched, so I have no reason to doubt Nasar and Gruber's work.)

Perelman's proof was dry and compact, which made it hard to follow for the average scientist (and even the not-so-average one), and Perelman was already known for his antics back in graduate school, saying for instance "If they know my work, they don't need my CV; if they need my CV, they don't know my work" when a member of a hiring committee asked him for his resume. He's proved results but hasn't bothered to publish the proof in peer-reviewed journals, and after proving the Poincare conjecture, declined the Fields medal on the ground that "if the proof is correct then no other recognition is needed." In other words, Perelman belongs to the idealist school where the work speaks for itself and doesn't need to be promoted.

Which turns out to have been a very naive position, because of Yau. The journalists explain that the two main ways to reach academic fame is to (1) find a new proof, or (2) identify gaps or errors in an already-existing proof and fixing them. But in most cases, gaps in proofs are gaps in exposition, where the researcher simply didn't bother writing down all the implicit steps, rather than mathematical gaps, where a part is truly missing. The first type of gaps is of course nothing special - graduate students deal with those all the time. The second type is more interesting, because it allows the scientists who fixed the proof to gain recognition along with the original authors, but it raises the question: how do you distinguish between the two?

This is at the center of the controversy surrounding Perelman's proof. The New Yorker states that "on at least one occasion, Yau and his students have seemed to confuse [mathematical gaps with gaps in exposition], making claims of originality that other mathematicians believe are unwarranted." The journalists give the example of a proof by a young geometer at Berkeley named Alexander Givental. Later, "a former student of Yau's who taught at Stanford gave a talk at Harvard on mirror symmetry [...and] proceeded to present a proof strikingly similar to Givental's, describing it as a paper that he had co-authored with Yau and another student of Yau's." In the final paper, Yau and his co-authors "mention Givental's work only in passing", write that his proof, "which has been read by many prominent experts, is incomplete" but do "not identify a specific mathematical gap." An inquiry later established that "Givental's proof was complete."

The fit Yau had when his former student Gang Tian was chosen to give a plenary address at the IMU's congress in Beijing (IMU stands for International Mathematical Union), and the incredible accusations he leveled at Tian to discredit him, are worth reading on p.35 and p.37-38 of the paperback edition. I'd repeat them here, but then you would have less incentive to read the whole piece (also available for free on the New Yorker's website), which would be a pity. Let me just quote this one line: "[Then the leading Chinese mathematician died] and Yau's efforts to insure that he - not Tian - was recognized as his successor turned vicious."

To gain credit for Perelman's proof, Yau apparently used a similar approach as he had for Givental's work, and again "mathematicians familiar with Perelman's proof disputed the idea that [Yau's co-authors] had contributed significant new approaches to the Poincare." A professor stated: "I don't see that they did anything different." Yau himself said: "If you can attach your name in any way, it is a contribution." On p.45: "As Perelman put it, 'if everyone is honest, it is natural to share ideas.' Many mathematicians view Yau's conduct over the Poincare as a violation of this basic ethic."

Sadly, Perelman was so revolted by the lack of integrity in the mathematics community that he decided to quit. He told the journalists: "It is not people who break ethical standards who are regarded as aliens. It is people like me who are isolated."

Yau has obviously not been thrilled by the article, published in August 2006. A more positive profile in the New York Times, published in October 2006, still quotes a professor as saying that Yau has "an outsized ego and great ambition, and has done things that dismay his peers" (the title itself, "The Emperor of Math", does nothing to change that impression), but he claims he's doing all this for "the younger generation" (even in the New Yorker article, he justified part of his behavior regarding Perelman as a way to help one of his young co-authors, who is a professor in China): "I have to fix my reputation in China in order to help younger students." According to the New York Times: "The New Yorker has said it stands by its reporting." And with the story's inclusion in the Best American Science Writing series, it is going to remain alive for a very long time.

INFORMS Practice Conference

Last month, I attended the INFORMS Practice Conference in Phoenix, Arizona. That was my first practice conference and the format was very different from the Fall meeting, which focuses on academic research - at the practice conference, talks last about 50 minutes rather than 20, focus on high-level ideas and results (avoiding equations for the most part), and are supported by high-quality visual aids. Note to self: practitioners really know how to use PowerPoint. The much lower attendance (in the hundreds of people rather than thousands) and the higher registration fee allows for networking breakfasts and lunches that are just as valuable as the presentations themselves, although the talks offer fantastic opportunities to see how operations research can be applied to real-life problems, and to get a sense of the issues that truly matter to practitioners.

This is the first conference in a long time where I did not skip any talk to catch up with old friends and colleagues; the Fall meeting is notorious for its fifty tracks spread out all over the conference hotels, with a few courageous people hopping from a session to the next and many others just hanging out in the lobby. My two favorite talks were "Incorporating Adverse Selection in Consumer Credit Decisions" by Robert Phillips from Columbia University and Nomis Solutions, which dealt with the fact that if you charge your customers a high interest rate, you're mostly going to get the high-risk people who can't get a better rate anywhere else, and the naive ones who don't realize rates vary a lot from a financial institution to the next, as well as "Range Planning: Supply and Demand Uncertainty in Financial Planning" by Blake Johnson from Stanford University, which introduced the concept of range contracts, a concept Diego Klabjan, who also attended the session, and I found particularly interesting.

At one of the lunches, a young MIT alumna who now works as a consultant in DC mentioned she found the talks presented by academics much more valuable than those by practitioners, because professors' teaching background allows them to make better presentations and convey the information in simpler terms. She was apparently getting a lot more out of these talks. I see her point - while I enjoyed some of the presentations made by industry people, the most interesting ones always had a connection with academia, often because the project was an industry-university partnership. That is not only because academics are more seasoned teachers, but because they practice creativity and out-of-the-box thinking of the highest order, while industry input helps them focus on valuable endeavors rather than toy problems. The worst two presentations were industry-only projects.

In one, the "founder and CEO" of his own consulting company attempted to explain what went wrong with the demand forecasting of a product that turned out to be a lot more popular than expected (leading to massive production delays); people in the audience started bombarding him with questions regarding his analysis and it became painfully apparent he had no clue what he was talking about. While I didn't look into that person's background and I'm not referring to that person's career path in any way, that made me think that sometimes people start their own business because they're unable to find anything better to do, rather than out of true entrepreneurship, and that the quality of independent consultants out there varies a lot. That also made me think that earlier peer input could have saved that man from a really embarrassing moment.

The other talk I didn't like was a presentation where someone decided to run a Monte-Carlo simulation by adding uncertainty to parameters treated as constants in the traditional Net Present Value framework, and look at the NPV distribution and the probability of its being positive to decide whether a project is worth launching. And for this, which is standard fare in simulation software such as Palisade's DecisionTools Industrial Suite, the man claimed that he had invented a new method and put his name and the name of a co-worker on it. How preposterous. His other contribution was to change the internal rate of return depending on the volatility of the main random variable in that time period. People in the audience pointed out that the IRR was set in all companies and their organization would never make it vary like that. The presenter also said that his method ended up giving the same valuation as the Black-Scholes model, which obviously couldn't be true because of the different model and assumptions involved - a point he agreed upon when confronted by audience members. The demand planning talk I mentioned above was a real train wreck, but that one came a close second. Fortunately, those are only two among the many captivating presentations made during the conference.

I enjoyed the presentation made by the team who won the Edelman Award - Hewlett-Packard, for its "successful product management initiatives" (from this press release). The idea was that HP needed an efficient algorithm to identify revenue-generating products and products they might discontinue; in this context, a product is every little thing that makes a computer, not the whole computer itself. The HP research team tried various algorithms to determine the "portfolio coverage" (a novel concept that considers the products in relationship to each other instead of in isolation). Because of the number of products HP manufactures, algorithms had to be able to handle hundreds of thousands of variables and output an answer in a short amount of time. This is not a research problem you solve in a day, and the HP team did an excellent job explaining its progression as team members decreased the running time little by little by changing their approach and attacking the problem from different angles - a great research lesson for anyone and in particular PhD students (few of whom were in attendance, but hopefully they will be there for the repeat presentation at the Fall meeting in San Diego).

There was an awkward moment in the Q&A period (the audience wasn't allowed to ask questions during the presentations the day before where the judges determined who should win, so that was the first time attendees could make comments) when someone said that he could see the benefit from bringing running time from days down to 20 minutes, but since the algorithm is run once a month for one application and once a quarter for the other, there didn't seem to be much of a purpose in decreasing the running time to a matter of seconds using a completely different approach, which appears to have played a key role in HP getting the award. (Maybe the person who asked the question was a disgruntled finalist from one of the other teams.)

The HP team hummed and hawed and just couldn't come with an answer for three painful minutes, until another HP scientist in the audience got up and said it was important for validation runs (to check the algorithm worked correctly). That was hardly an exciting answer, although I suspect the real reason they did all this, which might have been obvious to them but which they couldn't articulate because it was early in the morning after the awards banquet the previous night, is that they wanted the ability to answer any what-if question from their bosses in real time - something you can do if the algorithm runs in twenty seconds but not in twenty minutes.       

But there is no denying HP has done a superb job, applying cutting-edge operations research to a real problem they faced in managing their supply chain. Its Revenue Coverage Optimization tool has saved the company more than $500 million between 2005 and 2008, according to the Edelman press release. HP used the algorithm "to rank its Personal Systems Group offerings based on the interrelationship between products and orders. It then identified the “core offering,” which is composed of the most critical products in each region. This core offering represented about 30 percent of the ranked product portfolio. All other products were classified as HP’s “extended offering.” Based on these findings, HP adjusted its service level for each class of products. [... This has resulted] in lower costs and higher margins for the company," ultimately improving customer service by building inventory and decreasing lead times of core items.

Overall, the Spring practice meeting was a very enjoyable and valuable experience, which I will certainly reiterate.

A Convex Optimization Framework For Multi-Agent Motion Planning

Two weeks ago, I attended Jason Derenick's dissertation defense, entitled - you guessed it - "A Convex Optimization Framework For Multi-Agent Motion Planning". Jason was a graduate student in the Computer and Engineering Department at Lehigh, working under the supervision of John Spletzer; he has now moved to the University of Pennsylvania, where he is a postdoctoral fellow in the GRASP (General Robotics, Automation, Sensing and Perception) laboratory.

I served as a member of his dissertation committee, which is how I got to learn about his research. It turned out I had heard about him before - he was one of the team members behind the "Little Ben" car in the 2007 DARPA Urban Challenge. That car "was one of just six driverless cars to complete the 2007 DARPA Urban Challenge and it was the only car of the six finishers whose team had not received $1 million funding from DARPA to prepare for the race," according to Lehigh's press release.

In his dissertation, Jason has exploited recent advances in convex optimization to develop real-time algorithms for motion-planning problems, with an emphasis on optimal team formation changes and optimal collaborative target tracking. These problems lead to second-order cone programming and semidefinite programming formulations, respectively.

I was very impressed by Jason's work, and the wonderful presentations he makes of it. Of course, I only got to see the product of his research in its final stages, not the six years of hard work leading to it, but it is refreshing to bypass the difficulties and hurdles of graduate work for once (I've got my own students for that) and be shown the finished product without all the setbacks. While this is true of any committee I serve on, including for students in my own department, I particularly enjoyed the application to robotics - a departure of my usual line of work and a welcome reminder of my days as an undergraduate.

One reason why people - myself included - can get so enthusiastic about the work is that Jason excels at disseminating his results and explaining them at a high level anyone can understand (it is fair to say he is the National Science Foundation's dream research assistant). In addition to designing an excellent website, he has posted on his dissertation webpage an outline of his research and several videos and simulations of his algorithms in action. My favorite is the one with the aibos taking a delta formation, because everyone likes dogs, even robot ones.

Science and Online Journalism

A colleague forwarded me the link to this editorial in a recent issue of Nature, "Filling the void", which urges scientists to "rise up and reach out" as newspapers cut jobs, and the number of science journalists in particular decreases. It summarizes the issues facing science outreach quite nicely - more and more younger scientists blog, but science blogs can't introduce the general public to scientific issues, since laypeople don't look for them in the first place.

On a related topic, the March/April issue of Technology Review has an article ("But Who's Counting?") about the challenge of determining numbers of website visitors accurately. This matters because web traffic plays a key role in the rates websites can charge for advertisements, and online journalism is moving toward an advertisement-only business model. The article enumerates the following issues with traditional methods to measure web traffic: "in ascending order of impact, overcounting individuals with multiple computers or web browsers; counting 'mechanical visits' by Web 'bots' and 'spiders' (for example, when Google crawls the Web to estimate the popularity of sites) as visits by real people; and overcounting individuals who periodicially flush out the 'cookies' of code that sites stash on browsers so that returning visitors can be recognized." (I often do the last one.)

The journalist uses a couple of annoying words, like "digerati" and "misprisions" - I consider myself cultivatived, and yet I had to look them up - but overall the article is an insightful read, with valuable discussions on how to use "a methodology inherited from television audience research: the panel", where "panelists agree to have their Web browsing monitored through interviews and through 'meters', or spyware, installed on their personal computers." The issue? "[Panel-based audience research] tends to undercount people who look at sites at work."

A company named Quantcast is currently trying to develop new technologies to measure audience; its website offers an interesting overview of its methodologies; for instance, see here for the "Cookie to People Translation Overview" and here for the complete white paper. Quantcast competes for that business with Google, which plans to leverage its ownership of DoubleClick to combine audience data analysis with the ad-serving system, "so that media planners know which sites are best suited for which ads". This of course raises issues of dominance on Google's part.

Companies that had become used to television's Nielsen ratings might never know their web audience as precisely. But because it's possible to track customers' online behavior - especially when the customers click on display ads - online advertisement might also generate higher-quality data than a more traditional medium like television. After all, companies have never been able to tell how many customers they gained by running a TV commercial during this or that time slot. Managers might be forgetting a bit too quickly the uncertainty of the times they feel so nostalgic about.  

Robust Linear Optimization With Recourse and Cutting-Plane Methods

Prof. Marina Epelman from the University of Michigan, her doctoral student Tara Terry and I have finally put the finishing touches to our paper, "Robust Linear Optimization With Recourse," available for download at optimization-online.org. We propose a robust optimization approach to two-stage linear programming with recourse, and solve the resulting model using a cutting-plane algorithm based on Kelley's method. The advantage of the robust optimization approach is that it allows us to model random variables as uncertain parameters belonging to a polyhedral uncertainty set, and does not require the knowledge of the underlying probability distributions. The decision-maker's conservatism is taken into account through a budget of uncertainty, which determines the size of the uncertainty set around the nominal values of the parameters. We show that the robust problem is a linear programming problem with a potentially very large number of constraints, which is why we use a cutting plane algorithm as solution technique.

In addition, we formulate the robust problem with simple recourse as a series of m linear programming problems similar in size to the model without uncertainty, where m is the number of uncertain random
variables, and provide techniques for finding worst-case realizations of the uncertain parameters. We also demonstrate performance via computational experiments, which yield insights into the structure and performance of the robust solutions.

This is a paper we started working on in 2005, when Marina and I ran into each other at the INFORMS annual meeting in San Francisco, where I had given a talk on preliminary ideas I had on the topic, and while an earlier version of the work has been around since 2006, the revised algorithm and the new computational experiments have significantly improved the paper. Feel free to email us your comments and questions!

Grants and Academia

Stephen Quake, a professor of bioengineering at Stanford, wrote a guest column about the academic life in a New York Times blog (Letting Scientists Off the Leash, February 10, 2009), and more specifically about the importance of getting grants. As scientists progress in their career, their main task becomes to secure funding for their research programs; they typically do little hands-on research by themselves, training students to perform these tasks instead.

Quake wonders how much of a researcher's salary should be provided by the university in order to help scientists think outside the box, as opposed to running after the latest buzzword to get money. In many experiments-heavy fields, one cannot perform research at all without grant money, which is necessary to buy equipment. Quake was lucky to get some very innovative ideas of his funded through atypical programs, but he worries that the conservatism of review panels impedes cutting-edge science.

This is of course a complicated issue - I was surprised to read that 75% of some faculty members' salaries come from research funding, because I always thought this only affects summer salaries, which represents at most 25% of yearly pay. (Note that nine-month salaries, guaranteed by the university, are paid over twelve months unless the faculty requests otherwise, so professors do get paid over the summer even if they don't have research grants.) I also felt that Quake's view that, "without a grant there might not be food on the family dinner table" was overly dramatic, and a very distorted take on reality.

Quake also drops hints of faculty members being worried about their mortgages, but what about the people who lose their job (all of it, not 25%)? Does he think they aren't worried about their mortgages too? In this economy, guest columnists might want to choose their words a little more carefully. (The deli employee at the Giant supermarket, who works there as a second job to make ends meet and keep his son at a top liberal-arts college, mentioned the other day that store customers have been drastically cutting back on the amount of meat they buy, and have been shunning the top names in favor of cheaper store brands. His wife's commissions [she works as a travel agent] have plummeted and he worries he's going to lose his main job, which he already lost a few months back before finding a similar position elsewhere. And then Quake writes about academics worried about food on the dinner table - what kind of bubble does he live in? There's always plenty of food on my dinner table, and organic one to boot. At least tenured faculty members like Quake have job security, and if academics don't like the size of the mortgages in Northern California, they can move elsewhere.)

Quake's post doesn't enhance the image of academics, unless whiny is the new cool. Yes, transformative ideas are difficult to secure funding for, and anyone who has been reading my blog for a while knows I'd love to see basic research get more money from the government. I wish that Quake had emphasized the sense of lost opportunities instead - who knows whether the ideas that didn't get funded will be implemented one day? what if their inventor loses hope and decides to leave academia? what about the lives of the people such ideas would have helped? This staggering sense of lost opportunities is what motivates taxpayers and their Congressmen to put more money into science and engineering. It is truly sad to go into scientific research to help change the world, and be prevented from making a contribution because no one agrees to give you money.

A recent article in Nature emphasizes this point (Research funding: closing arguments, February 5, 2009, p.650-655). (Sorry, I can't post a link. You have to be a subscriber to view it, or at least your university has to.) It describes the case of two academics with stellar track records who were recently denied funding and are considering leaving science as a result. The message I took away from the article is that there are many scientists with good ideas and good grant proposals out there, and, although highly deserving, they are not getting funded because of the agencies' limited budget. This is not exactly breaking news, but it helps putting faces on the problem and makes the issues of closing down an experimental lab much more real.

In an update at the bottom of the article, one of the researchers did secure last-minute funding for the one grant that kept her lab running, and the other doesn't seem to be seriously considering quitting academia; however, this does vividly illustrate scientists' predicament. Another excellent website advocating an increase in the NIH [National Institutes of Health] budget is brokenpipeline.org, which asserts that "An unprecedented five consecutive years of stagnant funding for the National Institutes of Health is putting America at risk—slowing the pace of medical advances, risking the future health of Americans, discouraging our best and brightest researchers, and threatening America's global leadership in biomedical research. Unfortunately, President Bush's budget proposal recommends a sixth year of flat funding for the NIH in 2009." I particularly recommend the "A Broken Pipeline?" report.

The need to improve the funding situation in academia has nothing to do with faculty members worrying about their mortgages. It has to do with discoveries within grasp and yet not being made because agencies don't have enough money.

Science and the Stimulus Plan

Obama's first weekly video address ("fireside chat") as President has drawn over one million viewers on YouTube. In the five-minute-long address, the President provides a summary of his Recovery and Reinvestment Act, nicknamed "stimulus bill" in the press (a five-page overview of the bill is available here, on the website of the Senate Appropriations Committee).

I first read the New York Times article about the boost the bill would provide for education ("Stimulus Plan Would Provide Flood of Aid to Education", January 27, 2009), so I was surprised to learn the plan seeks to "triple the number of science fellowships" (see here for a transcript of the address) - the NY Times makes no mention of that part, maybe because fewer readers care about scientific research than about other very worthy programs such as Pell grants for needy college-bound students, maybe because post-baccalaureate education in science and engineering is sometimes considered closer to professional training than to traditional education, or maybe because the amounts involved are tiny compared to the size of the whole package. It represents, however, a sizable increase in the money devoted to federal research.

The summary of the bill on the Appropriations Committee bundles science with infrastructure in the second item by size ($165 billion out of $888 billion in investments and tax cuts), although science only represents $4.1 billion, to be shared by the National Science Foundation (NSF), NASA and NOAA (the National Oceanic and Atmospheric Association, presumably to study global warming) - this amounts to less than 0.5% of the total stimulus package.

The AAAS R&D Budget and Policy Program, which analyzes research and development initiatives in the U.S. budget for the American Association for the Advancement of Science, has come up with a much higher figure, including R&D facilities, capital equipment, and research efforts in the health and energy sectors: it estimates that "the just-released Senate draft of the 2009 economic stimulus appropriations bill contains $11.9 billion in federal research and development (R&D) funding, $9.7 billion for the conduct of R&D and $2.2 billion for R&D facilities and capital equipment. The recently approved House version of the bill contains more R&D funding, $13.2 billion, with slightly less for the conduct of R&D ($9.5 billion) but more for R&D facilities and equipment ($3.7 billion)."

This fact sheet from the Office of Speaker Nancy Pelosi places the price tag for scientific research at $10 billion; it also states that the package puts "the NSF budget on track to double over the next seven years, as called for under the America COMPETES Act". Better still (and underlined in the original document!), the $2.5 billion for NSF would "support an additional 3,000 new NSF research awards and would immediately engage 12,750 senior personnel, post-docts, graduate students and undergraduates." While the actual numbers will certainly change again before the bill is passed, the plan gives significant hope to the scientific community that the administration is serious about investing in research for the long term.

Predictably in light of the amounts involved, The Washington Post mentions that the stimulus package "has come under attack from Senate Republicans and some Democrats alarmed by its roughly $900 billion price tag" ("Obama Defends Stimulus in Effort to Get Bill Through Congress", February 4, 2009), and science could become a victim of lawmakers' price-cutting efforts. (From the Post: "Moderate Republicans are trying to trim the bill by as much as $200 billion," or about 25%. Good luck.)

Science spending would be an easy item to eliminate, as it is not tied to any specific constituency with clearly identified representatives in Congress; furthermore, the link between tackling the current economic crisis and maintaining America's standing in terms of scientific research is tenuous, at least from a short-term perspective (hence, the prominent place given to R&D infrastructure in the bill, such as lab construction). Senator Olympia Snowe from Maine, "a moderate who has been considered the most likely GOP vote in favor of the plan, said [on February 3] that she cannot support it until items that would not do enough to stimulate the economy or create jobs are dropped" (also from the Post). Academic research will not stimulate the economy, and it will not create jobs (with the exception of research assistantships, although those are not generally considered as employment). But it will support innovation, which is critical to ensure a robust recovery.

The Post article ends on a positive note for scientists: "The chamber ended the night by unanimously accepting an additional $6.5 billion for research at the National Institutes of Health, pushing the cost of the Senate legislation -- for now -- to more than $900 billion." Are lawmakers finally making the funding of innovation and scientific research a priority? Let's hope it stays that way when the time comes to pare down the bill.