Francesca Taolini asked him what he thought of the algorithm that the proposal hoped to develop.
“It’s too sketchy to be sure! That’s really what he’s hoping to find, as far as I can tell. There would be a final toolbox with a software environment and language, then a gene grammar to makes sense of palindromes in particular, he seems to think those are important, but I think they’re just redundancy and repair sequences, that’s why the palindromic structure. They’re like the reinforcement at the bottom of a zipper. To think that he could use this to predict all the proteins a gene would produce!”
“But if you could, you would see what proteins you would get without needing to do microassays,” Francesca pointed out. “That would be very useful. I thought the line he was following had potential, myself. I know people working on something like this, and it would be good to have more people on it, it’s a broad front. That’s why I gave it a Very Good, and I’d still recommend we fund it.” She kept her eyes on her screen.
“Well yeah,” Thornton said crossly, “but where would he get the biosensors that would tell him if he was right or not? There’s no controls.”
“That would be someone else’s problem. If the predictions were turning out good you wouldn’t have to test all of them, that would be the point.”
Frank waited a beat. “Anyone else?” he said in a neutral tone.
Pritchard and Yao Lee joined in. Lee obviously thought it was a good idea, in theory. He started describing it as a kind of cookbook with evolving recipes, and Frank ventured to say, “How would that work?”
“Well, by successive iterations of the operation, you know. It would be to get you started, suggest directions to try.”
“Look,” Francesca interjected, “eventually we’re going to have to tackle this issue, because until we do, the mechanics of gene expression are just a black box. It’s a very valid line of inquiry.”
“Habib?” Frank asked.
“It would be nice, I guess, if he could make it work. It’s not so easy. It would be like a roll of the dice to support it.”
Before Francesca could collect herself and start again, Frank said, “Well, we could go round and round on that, but we’re out of time on this one, and it’s late. Those of you who haven’t done it yet, write down your scores, and let’s finish with one more from Alice before we go to dinner.”
Hunger made them nod and tap away at their consoles, and then they were on to the last one for the day, “Ribozymes as Molecular Logic Gates.” When they were done with that, Frank stuck its Post-it on the whiteboard with the rest. Each little square of paper had its proposal’s averaged scores written on it. It was a tight scale; the difference between 4.63 and 4.70 could matter a great deal. They had already put three proposals in the “Fund” column, two in the “Fund If Possible,” and six in the “Do Not Fund.” The rest were stuck to the bottom of the board, waiting to be sorted out the following day. Pierzinski’s was among those.
That evening the group went out for dinner at Tara, a good nearby Thai restaurant with a wall-sized fish tank. The conversation was animated and wide-ranging, the mood getting better as the meal wore on. Afterward a few of them went to the hotel bar; the rest retreated to their rooms. At eight the next morning they were back in the conference room doing everything over again, working their way through the proposals with an increasing efficiency. Thornton recused himself for a discussion of a proposal from someone at his university, and the mood in the room noticeably lightened; even when he returned they held to this. They were learning each other’s predilections, and sometimes jetted off into discussions of theory that were very interesting even though only a few minutes long. Some of the proposals brought up interesting problems, and several strong ones in a row made them aware of just how amazing contemporary work in bioinformatics was, and what some of the potential benefits for human health might be, if all this were to come together and make a robust biotechnology. The shadow of a good future drove the group toward more generous strategies. The second day went better. The scores were, on average, higher.
“My Lord,” Alice said at one point, looking at the whiteboard. “There are going to be some very good proposals that we’re not going to be able to fund.”
Everyone nodded. It was a common feeling at the end of a panel. Rate of funded proposals was down to around ten or twenty percent these days.
“I sometimes wonder what would happen if we could fund about ninety percent of all the applications. You know, only reject the limited-discussions. Fund everything else.”
“It might speed things up.”
“Might cause a revolution.”
“Now back to reality,” Frank suggested. “Last jacket here.”
When they had all tapped in their grading of the forty-fourth jacket, Frank quickly crunched the numbers on his general spreadsheet, sorting the applicants into a hierarchy from one to forty-four, with a lot of ties.
He printed out the results, including the funding each proposal was asking for; then called the group back to order. They started moving the unsorted Post-its up into one or another of the three columns.
Pierzinski’s proposal had ended up ranked fourteenth out of the forty-four. It wouldn’t have been that high if it weren’t for Francesca. Now she urged them to fund it; but because it was in fourteenth place, the group decided it should be put in “Fund If Possible,” with a bullet.
Frank moved its Post-it on the whiteboard up into the “Fund If Possible” column, keeping his face perfectly blank. There were eight in “Fund If Possible,” six in “Fund,” twelve in “Do Not Fund.” Eighteen to go, therefore, but the arithmetic of the situation would doom most of these to the “Do Not Fund” column, with a few stuck into the “Fund If Possible” as faint hopes, and only the best couple funded.
Later it would be Frank’s job to fill out a Form Seven for every proposal, summarizing the key aspects of the discussion, acknowledging outlier reviews that were more than one full place off the average, and explaining any Excellents awarded to nonfunded reviews; this was part of keeping the process transparent to the applicants, and making sure that nothing untoward happened. The panel was advisory only, NSF had the right to overrule it, but in the great majority of cases the panels’ judgments would stand—that was the whole point—that was scientific objectivity, at least in this part of the process.
In a way it was funny. Solicit seven intensely subjective and sometimes contradictory opinions; quantify them; average them; and that was objectivity. A numerical grading that you could point to on a graph. Ridiculous, of course. But it was the best they could do. Indeed, what other choice did they have? No algorithm could make these kinds of decisions. The only computer powerful enough to do it was one made up of a networked array of human brains—that is to say, a panel. Beyond that they could not reach.
So they discussed the proposals one last time, their scientific potential and also their educational and benefit-to-society aspects, the “broader impacts” rubric, usually spelled out rather vaguely in the proposals, and unpopular with research purists. But as Frank put it now, “NSF isn’t here just to do science but also to promote science, and that means all these other criteria. What it will add to society.” What Anna will do with it, he almost said.
And speak of the devil, Anna came in to thank the panelists for their efforts; she was slightly flushed and formal in her remarks. When she left, Frank said, “Thanks from me too. It’s been exhausting as usual, but good work was done. I hope to see all of you here again at some point, but I won’t bother you too soon either. I know some of you have planes to catch, so let’s quit now, and if any of you have anything else you want to add, tell me individually. Okay, we’re done.”
Frank printed out a final copy of the spreadsheet. The money numbers suggested they would end up funding about ten of the forty-four proposals. There were seven in the “Fund” column already, and six of those in the “Fund