Andy Kessler has worked in Silicon Valley for a long time. He's seen the way that improving technology can lower costs and increase capabilities in all sorts of areas, and now he says that it's time for silicon to do for medicine what it's done for so many other fields.

That's the topic of his new book, The End of Medicine: How Silicon Valley (and Naked Mice) Will Reboot Your Doctor. And he thinks there's plenty of room for improvement. I agree.

Right now, technology isn't doing much. Kessler spent time with a radiologist reading mammograms, and discovered that it's done the same way it was done in the 1950s -- with a light box and a dictation setup. But there's one improvement. Where once two radiologists read each mammogram, now only one does, with the human opinion being compared to that of an expert system. That cuts the manpower needed in half. But it's just the beginning of what Kessler wants to see:

"Doctors hold the expertise. It's imbedded in their brains. ... But in other industries, the expertise is increasingly embedded elsewhere -- in software, in silicon, in routers, in cell phones, in iPods, in Xboxes, in search engines. That's what made Silicon Valley what it is today. You can take intellectual property and embed it on a chip -- to handle telephone calls, move email around, display 3-D graphics for video games and on and on. A dozen guys with no life design the chip and then workerless factories in Taiwan stamp them out by the millions to be shipped in products you and I can buy for under $100. That's scale. ... Yet here is R2. Cancer-identifying expertise is embedded in an algorithm you can buy for $29. Well, you can't buy it. Some weird system of service and reimbursements pays for it. But it's the first crack in the armor."

I hope he's right and that, as he says, "the Geeks are at the gate." Because he's certainly right that medicine up to now hasn't really benefited much from the electronics revolution.

Kessler thinks that things are about to change, and he makes a compelling case, based largely on technology that's already in existence rather than on Kurzweilian extrapolations of technology that ought to be available down the line. He notes that scanners are becoming far more effective, and much cheaper; that DNA-sensing chips are making blood tests for biomarkers indicative of cancer or other diseases much more effective (and cheaper), and that the improvements in these technologies will allow much earlier treatments: Zapping a tumor when it's at the ten-cell stage rather than the million-cell stage, for example.

As I said, Kessler makes a persuasive case, and I certainly hope he's right. (In fact, we interviewed him for a podcast recently, which you can listen to here or in dialup form here and found him pretty persuasive in person, too). But I wonder if the technological progress he describes will occur as rapidly as he thinks, for reasons that are organizational, rather than technical. That "weird system of service and reimbursements" may prove slower to change than other markets.

Kessler's expertise is in Silicon Valley electronics. But those products are usually paid for by consumers. Health care is paid for by intermediaries: government agencies and private insurance companies that are as big, and as bureaucratic, as government agencies. Will they be as swift to adopt improved medical technology as consumers are to pick up the latest Xbox or iPod? I'm not sure. In fact, I almost wonder if health-insurance companies might prefer to avoid diagnosis of tumors at the ten-cell stage, since they might then be on the hook for much more in the way of treatment expense than they would be if they didn't discover the tumor until just before it was fatal...

Kessler recognizes this problem, to a degree -- he jokes that you'd be better off having your health bills picked up by your life-insurance provider, or your death benefits paid by your health-insurance provider, because then their incentives would be aligned with your interests -- and he notes that the tradeoff between treating existing disease slightly more efficiently, and funding research and development that might lead to drastic improvements, is currently balanced pretty strongly in favor of the near term. I think he's right, but I fear he underestimates the difficulty of bringing about the kind of change that would produce really dramatic new capabilities. It's certainly the case that we're a lot more likely to see that sort of thing if health care consumers demand it. Which I guess is a reason to hope that his book sells well, particularly among policymakers and healthcare consumers. Perhaps you should send a copy to your Senator.

Glenn Reynolds is a TCS Daily contributing editor.