Tyler: What’s the last thing software will eat?
Marc: I think its fundamentally — the term that you used was project selection — and so basically its this question of how do you organize a small number of people to do something new, and by the way that could be a startup company, that can be many other kinds of efforts…
Then how do you pick — if you’re going to go finance or donate or fund those things — how do you know which ones to donate to? And then how are those things actually going to run, right? The new part that’s really important, the new little known thing about venture capital is there’s a term in hedge funds called carry — carried interest — which is the profit participation that the VCs or hedge funds make.
But the term carry comes from whaling in the 1600s off the Atlantic Ocean. Literally, the term carry was the people would finance a captain and a crew of a boat — the boat would actually run like a startup with equity participation for all the people on the boat — and you would pick a captain and raise money in town, and the boat would go off and try to take down a whale. About three-quarters of the time the boat would come back, and the other quarter of the time the whale would win — Moby Dick was not a joke — and carry was the portion of the whale that the investors got.
So if you think about it, the process — if you’re in Boston in 1675, with this ship, this captain, this crew, and in these waters, with these weather patterns, and how are they going to behave under pressure, and what’s going to happen when things go wrong; are they going to make any money doing this? — its such an intricate puzzle and it revolves around people.
Tech startups are the same way. And by the way green-lighting any kind of movie or TV show in Hollywood is the exact same kind of process. It’s so intangible, and so much is riding on the actions of a small number of people who are going to be under extreme pressure.
If we could figure out a way to automate that, then we’d fund that company and then retire. But we don’t know how to do that.
That was a great a16z podcast episode with Marc Andreessen called, “Beyond Software, to Talent and Culture.” Certainly project selection is one of the most important aspects in defense acquisition. Yet the fundamental uncertainty of the process — which is reflected in the rationale for why it will be the last thing automated — is often not recognized.
It is expected that the best possible project can be selected beforehand without many mistakes. Critics often bemoan the failure of a handful of projects in the DOD as a sure sign of incompetence. Yet no one in the venture capital world goes in expecting a 90 percent hit rate, much less a 100 percent. They often come in on the underside of 50 percent of investments that turn a real profit (i.e., were technically and economically feasible).
The defense acquisition process, in a sense, is an algorithm for outputting decisions on project choice. There are routine analyses from more than 50 separate offices which provide input. I think that the most pernicious part of the process is that it outputs project choices largely independent of the people who will actually work on it.
The VC model, by contrast, starts with individuals who are passionate about a project idea, and often it is the person and not the project which is selected for. It is expected that the project may pivot at some point. But the VC is not shopping its own project ideas to entrepreneurs.
Leave a Reply