There were eight of them sitting around the long cluttered conference table.

Dr. Frank Vanderwal, moderator, NSF (on leave from University of California, San Diego, Department of Bioinformatics).

Dr. Nigel Pritchard, Georgia Institute of Technology, Computer Sciences.

Dr. Alice Freundlich, Harvard University, Department of Biochemistry.

Dr. Habib Ndina, University of Virginia Medical School.

Dr. Stuart Thornton, University of Maryland, College Park, Genomics Department.

Dr. Francesca Taolini, Massachusetts Institute of Technology, Center for Biocomputational Studies.

Dr. Jerome Frenkel, University of Pennsylvania, Department of Genomics.

Dr. Yao Lee, Cambridge University (visiting GWU’s Department of Microbiology).

Frank made his usual introductory remarks and then said, “We’ve got a lot of them to go through this time. I’m sorry it’s so many, but that’s what we’ve received. I’m sure we’ll hack our way through them all if we keep on track. Let’s start with the fifteen-minutes-per-jacket drill, and see if we can get twelve or even fourteen done before lunch. Sound good?”

Everyone nodded and tapped away, calling up the first one.

“Oh, and before we start, let’s have everyone give me their conflict-of-interest forms, please. I have to remind you that as referees here, you have a conflict if you’re the applying principal investigator’s thesis advisor or advisee, an employee of the same institution as the P.I. or a co-P.I., a collaborator within the last four years of the P.I. or a co-P.I., an applicant for employment in any department at the submitting institution, a recipient of an honorarium or other pay from the submitting institution within the last year, someone with a close personal relationship to the P.I. or a co-P.I., a shareholder in a company participating in the proposal, or someone who would otherwise gain or lose financially if the proposal were awarded or declined.

“Everybody got that? Okay, hand those forms down to me, then. We’ll have a couple of people step outside for some of the proposals today, but mostly we’re clear as far as I know, is that right?”

“I’ll be leaving for the Esterhaus proposal, as I told you,” Stuart Thornton said.

Then they started the group evaluations. This was the heart of their task for that day and the next also the heart of NSF’s method, indeed of science more generally. Peer-review; a jury of fellow experts. Frank clicked the first proposal’s page onto his screen. “Seven reviewers, forty-four jackets. Let’s start with EIA-02 18599, ‘Electromagnetic and Informational Processes in Molecular Polymers.’ Habib, you’re the lead on this?”

Habib Ndina nodded and opened with a description of the proposal. “They want to immobilize cytoskeletal networks on biochips, and explore whether tubulin can be used as bits in protein logic gates. They intend to do this by measuring the electric dipole moment, and what the P.I. calls the predicted kink-solitonic electric dipole moment flip waves.”

“Predicted by whom?”

“By the P.I.” Habib smiled. “He also states that this will be a method to test out the theories of the so-called quantum brain.”

“Hmm.” People read past the abstract.

“What are you thinking?” Frank said after a while. “I see Habib has given it a ‘Good,’ Stuart a ‘Fair,’ and Alice a ‘Very Good.’”

This represented the middle range of their scale, which ran Poor, Fair, Good, Very Good, and Excellent.

Habib replied first. “I’m not so sure that you can get these biochips to array in neural nets. I saw Inouye try something like that at MIT, and they got stuck at the level of chip viability.”

“Hmm.”

The others chimed in with questions and opinions. At the end of fifteen minutes, Frank stopped the discussion and asked them to mark their final judgments in the two categories they used: intellectual merit and broader impacts.

Frank summed up the responses. “Four ‘Goods,’ two ‘Very Goods’ and a ‘Fair.’ Okay, let’s move on. But tell you what, I’m going to start the big board right now.”

He had a whiteboard in the corner next to him, and a pile of Post-it pads on the table. He drew three zones on the whiteboard with marker, and wrote at the top “Fund,” “Fund If Possible,” and “Don’t Fund.”

“I’ll put this one in the ‘Fund If Possible’ column for now, although naturally it may get bumped.” He stuck the proposal’s Post-it in the middle zone. “We’ll move these around as the day progresses and we get a sense of the range.”

Then they began the next one. “Okay. ‘Efficient Decoherence Control Algorithms for Computing Genome Construction.’”

This jacket Frank had assigned to Stuart Thornton.

Thornton started by shaking his head. “This one’s gotten two ‘Goods’ and two ‘Fairs,’ and it wasn’t very impressive to me either. It may be a candidate for limited discussion. It doesn’t really exhibit a grasp of the difficulties involved with codon tampering, and I think it replicates the work being done in Seattle by Johnson’s lab. The applicant seems to have been too busy with the broader impacts component to fully acquaint himself with the literature. Besides which, it won’t work.”

People laughed shortly at this extra measure of disdain, which was palpable, and to those who didn’t know Thornton, a little surprising. But Frank had seen Stuart Thornton on panels before. He was the kind of scientist who habitually displayed an ultrapure devotion to the scientific method, in the form of a relentless skepticism about everything. No study was designed tightly enough, no data were clean enough. To Frank it seemed obvious that it was really a kind of insecurity, part of the gestural set of a beta male convincing the group he was tough enough to be an alpha male, and maybe already was.

The problem with these gestures was that in science one’s intellectual power was like the muscle mass of an Australopithecus there for all to see. You couldn’t fake it. No matter how much you ruffed your fur or exposed your teeth, in the end your intellectual strength was discernible in what you said and how insightful it was. Mere skepticism was like baring teeth; anyone could do it. For that reason Thornton was a bad choice for a panel, because while people could see his attitude and try to discount it, he set a tone that was hard to shake off. If there was an always defector in the group, one had to be less generous oneself in order not to become a sap.

That was why Frank had invited him.

Thornton went on, “The basic problem is at the level of their understanding of an algorithm. An algorithm is not just a simple sequence of mathematical operations that can each be performed in turn. It’s a matter of designing a grammar that will adjust the operations at each stage, depending on what the results are from the stage before. There’s a very specific encoding math that makes that work. They don’t have that here, as far as I can tell.”

The others nodded and tapped in notes at their consoles. Soon enough they were on to the next proposal, with the previous one posted under “Don’t Fund.”

Now Frank could predict with some confidence how the rest of the day would go. A depressed norm had been set, and even though the third reporter, Alice Freundlich from Harvard, subtly rebuked Thornton by talking about how well-designed her first jacket was, she did so in a less generous context, and was not overly enthusiastic. “They think that the evolutionary processes of gene conservation can be mapped by cascade studies, and they want to model it with big computer array simulations. They claim they’ll be able to identify genes prone to mutation.”

Habib Ndina shook his head. He too was a habitual skeptic, although from a much deeper well of intelligence than Thornton’s; he wasn’t just making a display, he was thinking. “Isn’t the genome’s past pretty much mapped by now?” he complained. “Do we really need more about evolutionary history?”