The folks at Stat clearly have some good sources inside Google’s Verily startup. They ran a story back in March about problems with the CEO (blogged about here), and now they’re back with more opinion from within. This piece has a number of people expressing dissatisfaction with how Verily’s three big projects are actually going, as opposed to how the press releases for them are going:
It’s axiomatic in Silicon Valley’s tech companies that if the math and the coding can be done, the product can be made. But seven former Verily employees said the company’s leadership often seems not to grasp the reality that biology can be more complex and less predictable than computers.
They said Conrad, who has a PhD in anatomy and cell biology, applies the confident impatience of computer engineering, along with extravagant hype, to biotech ideas that demand rigorous peer review and years or decades of painstaking work.
Verily said in response to this criticism that it has hired “many seasoned and respected industry, academic, public health, and regulatory veterans who understand the complexity of biology and how long it takes to move from idea to device and/or therapy.”
One of these ideas is the so-called “tricorder”, a medical diagnostic device deliberately named after the Star Trek gizmo. That’s inviting trouble right there, if you ask me – you may recall that the show’s equipment was more of a plot device than anything else (as was the transporter), allowing Dr. McCoy or his successors to immediately establish that a leading character was going to pull through or that some extra in a red shirt was a goner. We’re not going to see that one on the shelves any time soon, but by using the name, you invite comparisons with something that isn’t even close to real. The project comes out of Google’s previously expressed interest in nanoparticles for diagnostic applications, but the problem is that Verily’s CEO told reporters that the technology was proven and that the device itself would be rolled out in just a few years, because everything was so safe that animal testing could just be skipped.
We use Star Trek as our guiding force around Google because there used to be a computer called Tricorder —you’d talk to it and it would answer any question. That’s what we’re really looking for at Google X. We want to have a Tricorder where Dr. McCoy will wave this thing and say “Oh, you’re suffering from Valerian death fever.” And he’d then give some shot in a person’s neck and they’d immediately get better. We won’t do the shots—our partners will do the shots. But we’re hoping to build the Tricorder. . .
. . .We’ve done a lot, to be quite humble about it. Enough to give us great confidence that this is all likely to work. . .
We know that much of this works: we’ve become very good at nanoparticle decorating, we’ve become very good at concentrating them and understanding how they behave in magnetic fields. There’s still a million crazy things that happen with people, and there’s a long journey to put medicines into people, and it has to be done in the open because we’re going to do experiments— people will be wearing these devices at our Baseline Study. But I think we have years to go, not decades.
You’re not going test on animals first?
You don’t have to in this case, the medicines are well known. I think we have pretty demonstrable evidence that this concept is plausible, maybe even probable. . .
Uh-huh. As a note for the future, any time someone tries to tell you that a new biomedical technology works so great that it doesn’t even have to be run through a rat, you should probably start heading for the exits. To be sure, Andrew Conrad does talk about a “long journey” and so on, but that (to me) looks just like a sprig of parsley on the side of a big steaming plate of We’ve Got This. Verily’s statement to Stat about the project sounds a lot more grounded in reality, though, as opposed to being grounded in whizbang PowerPoint slides: they said in response to a query that it’s “very early stage”, “ambitious”, “difficult” and so on, and I have no problem believing any of that, as opposed to believing that it all pretty much already works, y’know, to be humble about it. The people who spoke to Stat were even less complimentary, saying that it was generally assumed now within Verily that the whole project now existed mostly to get good press coverage.
The other big projects – a contact lens that reads off glucose levels, and a large “Google Baseline” study of human health in a standard population – aren’t doing all that well, either, as the article details. And I think that one reason for the problems, a big one, is described up in that first quoted section: we’re very smart, and we can do hardware and software very quickly and make it all work, and since everything in the world comes down to hardware and software, we can do most anything. De mortuis nil nisi bonum, but I attached the late Andy Grove’s name to this particular fallacy some years back, and it seems to have stuck.
The Stat article goes on to say that some of Verily’s smaller and less-hyped projects do seem to be working out better, but those aren’t the ones that grab the headlines and generate the noise. Vaporware is a Silicon Valley tradition, and it’s not unknown in the biotech world, either, but it doesn’t go quite as far over here as it does out by San Jose. There’s no softwarebetatrials.gov web site, there are no NDAs when you decide to release a new app, and there are no advisory committee meetings with public minutes where experts grill you about about your user interface or how quickly your algorithms really converge. It’s a different world, and it has to be a different world, and it’s worth keeping that in mind.