Tonight I'm reading "NASA at a Crossroads: Maintaining Workforce, Infrastructure, and Technology Preeminence in the Coming Decades,” the consensus report recently issued about how NASA can get its act together "to avoid a hollow future NASA.”
Boy, some of NASA's problems developing spacecraft sound a lot like the problems the rest of government has procuring the custom software that powers agency missions! I'm starting to think it's all the same problem! https://nap.nationalacademies.org/read/27519/chapter/1
The report finds that NASA keeps signing fixed price contracts for projects “when development is required and the requirements are not well established." Those contracts don't work if you don't know exactly what you need! That mismatch "nearly always ends in failure in implementation, such as cost growth, and at the worst technical failure."
Ditto for big software projects! Contracting officers love fixed price contracts, even for projects where it's a terrible idea, because it *feels* safe.
![When contracting for a product such as the implementation of a space project like the JWST or Apollo or a standard communications satellite bus, NASA has two basic contract choices, which are fixed price or cost plus. […] If the product requires little development and the requirements are well known with little change expected, fixed price is appropriate. Few of the challenging projects conducted by NASA fall in this category.
When development is required and the requirements are not well established, the appropriate contract mechanism is cost reimbursable. It is unfortunately difficult to estimate and control the cost for this class of projects. […]
A recent Aerospace Corporation study's showed that utilizing a fixed price contract when the project is better suited to a cost reimbursable contract nearly always ends in failure in implementation, such as cost growth, and at the worst technical failure.](https://files.mastodon.social/media_attachments/files/113/212/547/417/190/379/original/87f0eec0103c4771.png)
The consensus report identifies another problem that's familiar: they outsource so much work that it threatens their ability to do anything themselves. Pointing to the Artemis project, the write that ”NASA is more of a contract monitor than a technical organization capable of taking humanity into the solar system.” Vendors are _not_ as committed to the mission or its outcomes as NASA is, and the result is that projects are more likely to fail.
![NASA selected the fixed price services contract to develop and supply the Artemis HLS, in part to alleviate the aforementioned budgetary pressures. In this case, NASA is more of a contract monitor than a technical organization capable of taking humanity into the solar system. NASA's shift to this type of contract for first-of-a-kind, low-technology-readiness-level (TRL) mission work erodes the agency's in-house capabilities by eliminating NASA's participation, degrading the ability to provide creative and
experienced insights and oversight of programs, thereby putting the agency—and the country—at increased risk of program failure. […]
While the committee fully supports the goal of facilitating the development of the emerging space economy, it notes that excessive outsourcing raises serious issues that, in recent years, appear under-appreciated within NASA, successive administrations, and Congress. Among these is that NASA's engineers' and scientists' skills deteriorate very quickly when they are not performing hands-on work, and centers are thus deprived of the ability to develop future technical expertise and leadership.
[Truncated here because of the Mastodon length-limit to alt text]](https://files.mastodon.social/media_attachments/files/113/212/594/815/250/345/original/9d5a3f18b5908930.png)
An important difference between NASA and other federal agencies is that NASA still employs a lot of people who understand their technology (although they've largely relegated those people to contract oversight positions). Other federal agencies employ precious few software developers, and it's extraordinarily rare that any of them would oversee any of the software development work that is contracted out. (This goes very badly.)
This NASA consensus report names a lot of problems—there are seven core findings in all—and a lot of them have nothing to do with how the rest of federal government is so bad at procuring its mission-critical custom software. But one of those core findings is specifically about the danger of fixed price contracts for novel, custom development work. So while there are other problems, that is a big one.
I am increasingly convinced that the contracting profession’s insistence on using fixed price contracts, even for things that cannot actually have a fixed price assigned to them, is a major source of agencies' inability to achieve their missions. (In defense of contracting officers,, agencies don't actually employ people capable of seeing contracts that require proper oversight. It's a chicken-and-egg thing.) NASA's report has me more convinced than ever.
@waldoj it seems like an artificial limitation due to the strictness of the budgeting process, but honestly I’m not sure how else to approach things in a way that makes OMB happy. I don’t personally know any agency that’s found a way to justify incremental funding of software projects long-term…
@krusynth 100%, yes. And even some of *that* blame can be placed on the shoulders of the appropriations process. It’s turtles all the way down.
@waldoj Arguably it’s the ultimate absurdity that DARPA does most research on fixed price contracts. The contract I’m PI on is very much structured as a fixed price software procurement despite the fact that there’s no guarantee the ideas we proposed will work (obviously we think they will, but it’s research). Contract officers don’t seem to be able to write “we would like some research in this particular direction with these metrics” consistently.