At the moment, it is not intended to alter the approval process. We are first trying to impact our internal decision making. The data packages provided for a drug that's being submitted will be the same. If anything, there will be more understanding about the biology and the mechanism of action. Someday, this may provide us a way to create additional surrogate measures that can streamline the process, but it is not a near-term goal here.

Given all the work going on at Merck in the R&D division, is there a parallel wholesale innovation taking place in other areas of the company?

Our CEO, Dick Clark, talks about "one Merck" thinking. The initiatives that included use of experimental medicine go from end to end—from how we pick targets all the way to patent expiration. So we're breaking down the silos. There's a lot of communication and collaboration.

It typically takes about 10 years from target identification to filing. With Januvia, there were a number of elements that allowed us to cut that time in half. One of them was that we had the right people who worked seamlessly across boundaries. Discovery and Clinical and Chemistry worked together as one Merck.

That's how innovation occurs: You bring diverse, talented people together and put the problems you are trying to solve on the table. We had real alignment; we had our strategy worked out ahead of time. We knew what we would do if we saw a positive outcome, what we would do if we got a gray-zone outcome, and what we would do if it didn't work. And everybody was all ready to go.

We relied very heavily on that early single-dose experiment in type 2 diabetes. And that saved us more than a year because we went from Phase I directly into large dose-range finding trials in patients. And, in fact, the dose that we predicted from that early Phase I study was, in fact, the clinical dose. We went into man in 2002. We were in Phase IIb in the early part of 2003, which is very, very fast.

So it requires some luck, the right people, and an experimental-medicine mentality. It also helped that our Phase III program had very good alignment between the clinical and the commercial sides to make decisions about what was essential to be part of that package. And you make choices. We didn't do every single study that could be done. We did the essential studies. We focused on safety and tolerability and efficacy in the most critical setting.

I guess the devil's advocate argument against this whole experimental-medicine approach is the Viagra story: a compound for diabetes that wasn't looking so hot but was in Phase III anyway, and then that surprising side effect turned up, so to speak. Do you have to address this argument?

You know, you can keep hoping for the winning lottery ticket, but we're scientists, and we have to have think about our research from a portfolio-management perspective. There's always going to be feelings like "Wouldn't it be great if..." or "We might miss something unless...." That's just part of the scientific process. But if each scientist thinks in aggregate and in terms of meeting our mission, we're still going to be much better off.

When will we be able to see evidence of this model's success by looking in your pipeline?

This is going to take a little while given the cycle time. We're developing metrics internally so that we have leading indicators: Are we making more decisions early? Are we stepping up and showing the courage to make the early decisions? Our initial goal is making more early decisions. But, ultimately, our goal is getting more drugs to patients.

And if you say one isn't working based on certain information and that you're going to take it off the board, then there are always other candidates, right?

Absolutely. There are lots of interesting, compelling targets—many more targets than drugs—so you have to find the one that has the greatest potential to help people. And each one always look good on paper.