The Brits, not once but twice, the last time being in the Falklands, couldn’t hit anything with naval gunfire because they forgot to reverse the sign of the coriolis effect in the southern hemisphere. If look at the Vasa, the Swedish ship that turns turtle in the 1600s, you’ll see they had a model for vertical stability but they linearly extrapolated out into an exponential area. So models are great, and I put a lot of my professional life developing models and stringing them together. It’s risky business.
That was Vice Admiral Joseph Dyer on the Preble Hall podcast discussing the development of the F/A-18E/F, The Birth of the Strike Fighter. I previously blogged on this podcast here.
I think Dyer’s point is that models can be useful, but they must always be put to empirical testing to assure their validity. This is particularly true for extrapolations outside the range of historical data.
Nature is non-linear. The form of an ant cannot exist at the scale of an elephant. It would collapse on itself. Phase transitions from solid to liquid to gas show how there can be abrupt discontinuities. Resonance between frequencies can drive oscillations to even higher amplitudes.
Usually regimes outside the range of historical data have not been explored because there is some nonlinearity at play that prevented past attempts. The sound barrier is a good example that required redesign of the horizontal tail to maintain stability. The C-5 was another example of a “simple extrapolation” to a larger aircraft size which turned out to be wrong.
These things matter for specific decisions related to technical feasibility. In many cases, he results cannot be known for sure in advance of trail-and-error testing. This is because the whole point of advancement means anticipating results outside of the realm of existing data. But the benefits could be 10x or 1,000x gains. So it is a worthy strategy to take risk — you cannot afford to miss out on a potential discovery.
VADM Dyer mentioned the willingness to take risks in the post-war years that seems to be missing from government today:
[On the century series] It’s a time of great engineers who tried a lot of exciting things. And it coupled with a government who at that point in time was willing to try and fail pursuant to breakthrough capability. American aviation in the time of the century series, philosophically, looks a lot like venture capital today.
One major element of venture capital is the ability to recognize talent, build scalable teams, and evaluate technical feasibility. Wall Street is good at getting capital to companies that have proven product-market fit. They can rely on their financial data and don’t need to be subject matter experts.
Perhaps DoD used to have more technical knowledge which allowed them to act more like VCs than Wall Street. If you have perfect models of the world and can compute feasible outcomes, then DoD perhaps didn’t need so much organic effort to connect users, testers, and developers. Here’s Dyer on something similar:
Another piece of magic for naval aviation is the reach-back between individuals and communities across the years. Those [test pilots] who went to the fleet were always in close contact with the R&D community. It was a time — frankly we let slip quite a bit — It was a time when requirements were quicker and better understood in their genesis because of that working relationship.
Leave a Reply