16 thoughts on “SpaceX And Palantir”

  1. This is hardly a new problem. J. Walter Christie tried and failed to sell his tank designs in the US facing stubborn opposition from the DoD establishment only to find eager clients in the Soviet Union and the UK. The Soviet T-34 tank and British Cromwell tank used a Christie suspension.

    A couple of years back The ‘Innovator’s Dilemma’ was a popular book. Much like there in here the case is the DoD has a firmly established mindset and is simply not interested in doing things any differently.

  2. Lobbying costs being justified should alone be enough evidence that government should be hands off business. Consumers should regulate business with government limited to assisting transparency. Even there the media has a more important responsibility (which they fail.)

    It’s sad that lawsuits are required.

  5. In the period 1945 – 1965, computing was in its infancy, personal computing meant a slide rule or perhaps a Marchant calculator, mechanical drawings were still rendered by hand on large tables and automated machine tools were things like analog tracer lathes and pantograph mills at the beginning of this era, segueing into paper tape NC machines by the end. Yet the U.S. developed dozens of military aircraft designs that made it into serial production during this period.

    These days we have server farms and supercomputers, computational fluid dynamics, computer-aided design and engineering including parametric modeling of components and assemblies, finite element analysis, computer numerical-controlled machine tools, assembly robots and much else besides. Yet in the last 25 years, we have produced only a handful of new military aircraft designs that have seen serial production.

    To put matters bluntly, the paperwork requirements of the federal procurement bureaucracies, especially FAR, and the perverse incentives to stretch out and inflate the costs of programs pursued on a cost-plus basis have more than kept up with advancing design, engineering and manufacturing technology – indeed have swamped them. The F-35 Lightning II, which was supposed to be the “cheap” and numerous 21st Century stealth air superiority fighter, is even later and will probably wind up costing more per unit than its now-predeccessor “expensive” and limited-production 21st Century stealth air superiority fighter, the F-22 Raptor. This is, in a word, nuts.

    The entire edifice of FAR and cost-plus business-as-usual has to be swept away and replaced by fixed-price milestone-based contracting. No weapons development project – including the damned software – should take more than five years from a clean sheet of paper to serial production. The best air superiority fighter of its day, the P-51 Mustang, went from sketches to a flyable prototype in something like 90 days. That we cannot remotely replicate this in our modern era is a scandal and a disgrace.

  6. “The best air superiority fighter of its day, the P-51 Mustang, went from sketches to a flyable prototype in something like 90 days. ”

    Well that’s the legend but in actuality it was under consideration and development for a year. Schmued had done lots of preliminary design work prior to the Brits asking NAA if they would build the P-40. He and the design team got really serious once the Brits showed up.

          1. Large codebases have been rewritten plenty of times in the civilian economy. It’s often done piecemeal, using an emulator to keep much of the old code as-is while re-doing the really foundational and time-critical stuff first. Then you incrementally convert the rest on a priority basis.

            This is especially a good idea if the underlying hardware is also changing in ways that are significant to the codebase in question, like running natively on a new processor with a different instruction set. Apple took such an incrementalist approach to converting its original Motorola 68000 Mac codebase to run on Power PC chips, then to do something similar to the NexT codebase (also originally Motorola 68000) to run on Intel processors. These conversions weren’t without hiccups, but they went fairly smoothly and, by the standards of large, present-day defense projects, fairly quickly.

            Now, having settled on Intel processors, it seems Apple may be at it again at the source language level, converting legacy Mac Objective C code to a new language they’ve invented called Swift. That’s pretty much directly analogous to the Ada-to-C++ switchover you describe.

      2. Granted. But planes got more and more electronics and jet fighters were routinely supersonic within 10 years of V-J Day. And the tools, both software and hardware, we have these days to produce and prototype designs have advanced by leaps and bounds. Still, the bureaucratic blob has overtaken them.

  7. I strongly believe you’d often get faster and better results if you have 10 guys working independently on a design than 10 guys working together. After a certain amount of time you take the best of the ten, make that person the team leader and then let the team work on it.

Comments are closed.