spacer image
spacer image

Welcome! You're looking at an archived Snarkmarket entry. We've got a fresh look—and more new ideas every day—on the front page.

August 5, 2009

<< A Fine Vintage In the Kitchen | The Name is Shrdlu... Etaoin Shrdlu >>

Beyond Starbucks: Physical APIs

Some great ideas are sparking here, helped along by Robin’s notion of a “Starbucks API.” Noah Brier calls it a “physical API” (see also the smart comments) and Kit Eaton at Fast Company extends the concept (tongue-in-cheek) to Microsoft, Apple, and Twitter. But I like Drew Weilage’s proposal at Our Own System the best:

The idea: create a “physical API”… of the Mayo Clinic or Cleveland Clinic. Copy their entire way of doing business and paste it into hospitals around the country. In a nicely wrapped package deliver their systems for decision-making, integration, coordination, and expertise. Include their human resources practices, innovation efforts, and technology. Import their employment model, their bargaining power, and of course brand recognition. This is a beta release so if anything is left out, it can be included in a later version.

Mix with water. Implement. Poof! Great health care!

Just think about it, Local County Hospital, powered by the Mayo Clinic or Our Lady Health Care System, supported by the Cleveland Clinic; it’s a definite brand extender.

Seriously — this has, potentially, amazing public policy implications. My dad, who’s worked in the government for-practically-ever in Wayne County/Detroit (first at the jail, then in public health, then in lots of places), always used to stun his bosses, co-workers, everybody, because whenever they ran into a persistent problem or one they couldn’t solve, he would get on the phone to people he knew in Oakland County, or Chicago, or Denver, to see how they handled it, who would in turn refer him to other people, etc.

You can get these information bottlenecks even when there’s no competing interests, and nothing proprietary — it’s just hard (without an API) for people to know where or how to look.

(Also note - at this point, whenever I hear terrific health care ideas, I do a little dance, with a chant that kind of sounds like this: a-TUL! Ga-WAND-e! A-tul, Ga-Wande!)

Posted August 5, 2009 at 7:46 | Comments (6) | Permasnark
File under: Briefly Noted, Cities, Medicine, Snarkpolicy


Wow, way to track & synthesize the conversation, Tim. The only one of those posts that I'd seen was Noah Brier's.

Seems like you're mixing in a new idea here -- entirely appropriately, and entirely for the better -- so now maybe the suite of products is:

1. The physical API -- a set of material procedures, services, & supplies that are as easy to access as Amazon's web services are today.

2. The franchise-as-network -- so it's not a top-down "this is how you do X" franchise but rather a big horizontal knowledge-sharing system. And we define "franchise" broadly so it can be a coffee shop, a school, a hospital, & more.

a-TUL! Ga-WAND-e! a-TUL! Ga-WAND-e!

I keep Google Alerts, Technorati pings, and Twitter and Friendfeed searches for "Snarkmarket," "New Liberal Arts," "Digital Humanism," "Blood and Treasure," etc... This is the new way to read.


Checklists turn out to be ridiculously useful for porting procedures around between institutions.

Here's a NY Times story about checklists in use in hospitals along with the infuriating reason that they were shut down. Which serves well to illustrate some of the issues with the Beta attitude when it comes to health-care.

People will die.

Now, the thing is that people will die anyway now with the current bad procedures, but at least these are kind of known bad procedures. New procedures could be better, could be worse but until you test them out (which requires patients' permission) you don't know which ones will work better and which will kill more people, or cause some other undue harm.

I've been thinking about this same problem in terms of architecture. If you make a mistake in code, it can take weeks or even months to fix. If you mess up a building, that's a 50-year error. You can't build and use beta-architecture for safety reasons. Ditto for beta-healthcare.

On the other hand, see also: books like A Pattern Language, which is essentially APIs for city and building planning. Which is what led to design patterns in all sorts of disciplines. Including computers.

Yeah, good call on Pattern Language. Re: regional APIs, I'd love to know what the procedures/protocols are for mega-ports like Singapore and Oakland. Talk about packet-switching. That would be a REALLY fun article to read -- the too-detailed technical exegesis of how a port really runs.

Actually, I'd love to see a David Macaulay book called: PORT.

Some of this is tied up w/ the question of scale, too. If we're talking about a giant, Promethean skyscraper in downtown San Francisco: Yes, let's not "beta-test" that new steel support structure.

But, I mean, is that really the kind of architecture we want, anyway? That's like the Windows Vista of buildings.

Think of some of the classic buildings at MIT that have been hacked, re-hacked, and hacked again -- some of them were /designed/ to be hacked.

Think of Steward Brand's How Buildings Learn.

Yes, we're still talking longer time-scales... maybe even 50-year time-scales. But I don't think the perceived "seriousness" (or, by the way, the professional tribalism) of things like architecture & medicine should make us afraid to experiment, to play, to hack 'em a bit.

And anyway, aren't clinical trials just beta tests for new drugs and procedures? ;-)

P.S. Did A Pattern Language really influence/inspire the concept of design patterns in software? Really cool, if so. I'd never heard that.

Actually, now that you mention it, the only place I've ever heard it was Wikipedia. The book came out in 1977 so there is some plausibility to the claim. Still, I think it's fair to put in a little [citation needed] next to those,

spacer image
spacer image