Saw a very interesting post this week over at VentureBeat Life Sciences taking Intel Chairman Andy Grove to task for his views on medicine and the pharmaceutical industry. Grove's view is that medicine should be like semiconductors and that the only reason it's not is that the industry isn't trying hard enough:
Is Grove really under the impression that pharma and biotech executives spend their days sitting around twiddling their thumbs and inventing new roadblocks to toss in the path of life-saving therapies? Or that drug companies aren’t eager to take apart clinical-trial results in order to salvage some shred of success from failure?
The fact, of course, is that chip development and drug development are so different that you have to take great care with direct comparisons lest your argument collapse in a ludicrous mess, and there’s no evidence that Grove has taken that sort of care here. Semiconductors are based on well-understood principles of solid-state and quantum physics a century or more old, whereas our fundamental knowledge of human biology is still in its infancy, notwithstanding the advances of the past several decades. Engineers know more about how even the most complex Intel CPU works than anyone does about how biochemical signals direct the activity of individual cells — much less tissues, organs, or, God forbid, networks as complex as the immune or central nervous systems.
Chips can be improved by tinkering, whereas the same is rarely true of a drug, since even minor changes can cause toxic side effects. (A commenter at Derek Lowe’s blog In the Pipeline likens drug discovery to writing software for a buggy operating system — only you can’t test your code on the OS until the very end, and can never, ever let it crash.)
The software/pharma comparison has actually occurred to me before, but from the opposite perspective. It occurred to last year when reading this post on software bugs by Vinnie Mirchandani. Basically, as a software package goes through successive iterations, it becomes buggier. It's not just Microsoft either. At every incumbent software vendor, development cost and times continue to spiral, while the number of bugs only proliferates. When I read that post, it struck me as being similar to what's going on right now at any big pharmaceutical maker.
There is a countervailing force in software, which is based on some combination of on-demand, lightweight or open source,. Though none of this is a panacea, all this stuff, I think, puts us back on the right direction. I'm not sure, though, what an open source pharma would look like. It's a great idea, in theory, but obviously the software business and the drug business aren't the same. Anyone can think of a million reasons why the Linux model wouldn't work for drugs, particularly given modern regulatory structure.