Erik: Let’s get into the lessons you learned from Uber that you find most interesting.
Flo: I think the first one is decentralization that exists at two levels: the HQ level and the city level. When Uber opens in a city, it doesn’t flip a switch and ‘oh, now we’re in Austin.’ It actually opens an office there and hires a team. In that team, the is a general manager, and the GM will own that profit and loss statement, and they’re going to manage the whole business. They are going to be given a ton of tools to manage the business. Data analytics tools, communications tools, payment tools, etc., etc. That makes the company multi-threaded. These teams have so many tools, they can build an entire product without HQ being involved at any point. I think a spectacular example of this is Uber Eats. It original started as a pilot in Los Angeles… Empowering people allows you to move faster. It makes people feel like owners of their own business.
What was from the Venture Stories podcast, “Silicon Valley, libertarianism, and lessons from uber with Flo Crivello.” Lots of interesting material throughout. Flo explained how business models help determine company culture, and why a decentralized culture worked at Uber:
Apple’s strategy is all about building the best products in the world. You can be first to market, you can be cheapest, or you can be best. We’re best. They don’t have a culture of thriftiness, for example, because they don’t go after price they go after quality. Amazon, on the other hand, is in retail, and it’s value is that we’re the everything store and we’re cheap. They’re in a low margins business.
The culture you see at Uber — it is a less defensible business than Google or Facebook — so it is about staying ahead of the curve and hustling. That is why a decentralized model makes sense for Uber, because you need to be close to the ground. By the time a piece of information went from the bottom to the top, and the top makes a decision — even if it is a good decision because it doesn’t have the information — the time the information reached the bottom it is too late. You need to react now.
When I think about the Department of Defense as a whole, it certainly seems like the decentralized model works best since military overmatch is often about being first to a new set of capabilities. And that should really put our attention not just on particular technologies, but on how they affect operational concepts and methods of fighting.
The DoD is more like an economy than a business, and so the decentralized nodes of the DoD will often have different cultures based on their mission or product area. Some program offices might be focused on price, like in areas of munitions and supply. Some program offices might be focused on quality, like in advanced aerospace design. Others will emphasize being first to market, like C4ISR and cyber.
What also seems important is how Uber provided its decentralized nodes with a number of tools to enable it to build. That’s a huge capital hack, dramatically increasing productivity. The DoD is starting to think about tools in terms of software development with CloudOne and PlatformOne, and also AI/ML tooling that seems to be coming out of the Joint Artificial Intelligence Center. (By the way, I’m glad the JAIC is moving to a tools-enablement model rather than being an AI-czar as Ash Carter foresaw). But there are more dimensions to enabling tools than just software and data. These concepts should be expanded into dedicated programs for advanced tooling, lab equipment, physical infrastructure, 3D printing, and so forth.
The idea that the DoD is more like an economy with each node having its own culture is thought-provoking. I’ve been wrestling with the question/idea/issue of whether AF program offices should all strive to be more innovative or if, as I believe it is suggested in Loonshots and The Other Side of Innovation, that it makes more sense to separate your innovators from those in keeping the performance engine running. E.g. AF acquisition has the RCO and Big Safari to do innovative rapid acquisition…so, rather than trying to make the Bomber Directorate a rapid innovative organization, they should just keep churning out ‘regular’ programs and mods while the rapid/innovative stuff goes to RCO and Big Safari. What do you think?
Good question. I honestly don’t know. There will always be organizational fault lines. I tend to like the “cradle to the grave” model rather than the SMC 2.0 model where you have programs transition from Dev Corps to Production Corps to Operations. I think having tech labs outside that system makes sense. But cradle to the grave organizations can also have diversity within them.
In the end, I think DoD should start with the Org structure it has, and let them drift into whatever structure makes sense. But that can only happen if they have more control of their own destiny (i.e., their own budgets).