One of the posts I wrote was a case study of the transistor. The transistor was invested at Bell Labs in the semiconductor research group. Clearly depended somewhat on the science of semiconductor physics. We have to know they existed and have some notion of their properties before we could go do those investigations. A lot of semiconductor physics necessary to invent the semiconductor had not yet been figured out.
In fact, it was figured out by the researchers at Bell Labs in the course of inventing the transistor. It’s this fascinating story where they shuttled back and forth between science and invention. On the basis of science, they tried to make a transistor, and when it didn’t work, they’d go back to the drawing board to figure out a new theory for why it didn’t work, and with the new theory they tinkered again with the experiments, and with the experiment they got an unexpected result. ‘Oh, that’s interesting, why is that happening?’ And then they go back to theory again and explore the physics. They were pushing the frontier of semiconductor physics in conjunction with the experiments.
This throws a simple linear model, that goes first we discover scientific principles, then we make inventions based on them, then we create products based on the inventions, and then we build a business based on the product — boom, boom, boom. It’s like the waterfall model of software development. First we decide on our product requirements, then we create a design, then an architecture, then we implement it, then we test it, then deploy it — boom, boom, boom. Anyone who’s been in software development knows it doesn’t happen that way, and can’t happen that way. And yet, there’s some truth to it. It’s not completely wrong. There is some truth to the notion that design precedes implementation. But it’s not as clean and simple as that.
I think it’s very much the same in the linear model of innovation. It’s true in a certain sense that science underpins technology and invention, but it’s not a linear thing. It seems to be what happens is there’s a lot of iterative invention that happens at the frontier. By that I mean tinkering that happens on the basis of previously established science, and could not happen without that previously established base of concepts and ideas, but which then pushes the frontiers of the science that explores phenomena that the science cannot yet explain. Great inventions come out of this boundary-zone tinkering.
That was another excerpt from Jason Crawford on the Venture Stories podcast, “Progress Studies in 2020.” Can I get a hallelujah?
The DoD’s entire acquisition system was developed on the basis of the linear model of innovation. All programs require that same “waterfall” method of development, starting with well-defined requirements and moving forward from there. All programs must move through linear stage-gates of funding. First, the program gets basic research funding in RDT&E BA 6.1, and the moves on through the remaining six RDT&E budget activities until it goes through test and evaluation. Then, it goes into procurement. Then, it goes into sustainment. There is no room for iteration. All knowledge is supposed to be completely ascertained before the program begins, usually around BA 6.4 or BA 6.5 in RDT&E.
There has been a trend to a new “colorless” appropriation for software where funding isn’t broken up into the linear model of innovation. (Of course, it’s awkward that the “colorless” mechanism sits in RDT&E budget activity 6.8.) But that doesn’t get to the heart of the matter. A software program ostensibly within the “colorless” appropriation still needs all sorts of documentation as if it were completely planned out as a waterfall series of events. That defeats the purpose of the “colorless” appropriation. And so it is my assertion that the DoD must first address it’s overall structure of program governance to allow tinkering at the frontiers, not just in S&T, but at scale. Never will program risk be reduced to zero before it hits an acquisition program. Until that time that the DoD ditches waterfall processes of lifecycle planning and risk reduction, the “colorless” appropriation cannot do it’s job.
Leave a Reply