Diablog 001: Kevin Starr

Hi Bleeps-
So I have had this idea for a while that it might be interesting to start getting other people involved in this blog. First, I have had some great guest posts from folks like Carl and Teju. But I also wanted to try something I called a "diablog"... a mashup of a blog and a dialog. Kevin Starr and I were emailing back and forth about my recent post on Small Batches, and I asked him if he minded if I used it as an MVP of a diablog. One of the many great things about Kevin is his willingness to play guinea pig, so here goes.  I kept editing of the email thread to a minimum... rather than add links to the thread, you can find links to the organizations Kevin mentions here. Let me know what you think... who/what else would you like me to probe or provoke with this format?
Happy holidays, Paul
++++++++++++

Kevin sent this slide from a talk he'd given: 

Paul: ... thanks for looking over my blog on small batches/experiments. As I said to someone else, it's smart to keep an organization "quick and crafty." And you know, I think you are a madass. I liked your slide showing stages... idea -> r&d -> proof -> replication -> scale. And I liked your line under scaling companies that "all good organizations continue to have a lab to help them innovate and evolve"... what is an example you've been impressed by? One Acre's experimental farm popped into mind...

Kevin:   Yeah, One Acre is a perfect example.  They are continually honing their model, and trying out new things.  They know that you need to separate your work at scale from your experiments until you have proven stuff to integrate into operations and model.  One Acre is in fact planning to expand their r&d as they grow, which i think is a really good way to go. 

All labs should be "crafty and quick."  That doesn't mean that organizations should stay small, just that they should continually improve their model and operations via that part of the organization designed to "fail fast" and obsessively iterate.

Paul: One of the challenges I see, and what I liked  about the New Belgium article, is that these labs are separate, yet part of, an organization. In a lot of the work on innovation inside existing organizations, there is a tension between getting the main business done, and doing the r&d need to position for the future, offer new services, etc. Another tension is bringing experiments into the mainstream when they look promising.  Do you have any stories from Mulago organizations on bringing things into the mainstream, or of scaling them back into the main business? What should organizations think about when an experiment works?

Kevin: Well, they shouldn't do experiments unless they are ready to do something with the result. Nuru, One Acre, Proximity Designs, Kickstart, Kiva - and I could add a lot of other names, but these are all Mulago organizations that have ongoing trials of new ideas, technologies, and products.  They are eager to make use of the products of their experiments - that's why they did them in the first place.  When I was doing medicine, there was a useful saying that more should have adhered to: "don't do a test unless you know what you're going to do with the result."  The same should be true for field experiments.

Paul: As usual, you have better examples than I. I usually tell the story of the dog that chases the school bus every day. Then one day he catches it and says "now what?"  So, building on your point, one thing that concerns me about a lot of early social ventures is they haven't thought this through, and sometimes their experiments have consequences or risks that fall on the people that can least afford them. In early days of Envirofit, we took great care to compensate taxi drivers for the day that their motorcycle was in the shop for the retrofit, and we were worried about what happened if the system failed. Not just because we wanted it to work, but because that would be a day's lost income for the driver, plus having to push the vehicle home, etc. In medicine, there is a lot of early pre-clincal work and testing before one tests safety in humans... it doesn't seem that the same thought goes into our sector. The best work i've seen on this is the BOP Protocol that Stu Hart, and his crew at Cornell, put together. I'm a little worried about the lean startup, go do experiments approach in the social sector... 

Kevin:  Well, it’s hard to do market trials with monkeys…..really, though, few of the “experiments” we’re talking about have much potential for harm – things like different marketing approaches, trying out new crop varieties on test plots, better recruiting strategies.  You’re not trying out new therapies on unwitting patients or having a bunch of farmers bet their farms on a crop with an unproven market.  You’ve got a firewall between your experiments and the interventions that you take out into the world.   If and when people are part of your experiments, they should know what they are getting into and be enthusiastic participants in a mutual adventure.