Ideas + Resources are enough

Today I participated in JPL’s Research Day, where researchers from across the Lab presented their work. I was genuinely impressed, and honestly a bit shocked, by the sheer number of posters. And this was only a fraction of the research happening at JPL; many teams weren’t even represented.

What struck me most is this: ideas are everywhere. But one lesson the AI community has taught us is that ideas alone don’t move the world—ideas + resources do. When a field receives focused investment, attention, and infrastructure, it accelerates dramatically. The same could happen here. Many of the projects presented today, spanning thruster technologies, power systems, sensors, electronics, radiation testing, Earth observation, autonomy, and much more, are brimming with potential.

What they need is a clearer path from idea → prototype → flight heritage. You might be thinking, “Space is expensive—how could we possibly afford this?” I believe it’s absolutely possible, but it requires intentional design and deep strategic thinking. It would also require people to not be lazy, and put in the effort.

Here’s some steps:

  1. Cluster related ideas and encourage cross-team collaboration. Often, people in different sections are working on similar problems without realizing it. Bringing these efforts together could create stronger research programs. Yes, this can be tricky, ego sometimes competes with scientific progress, but we should still try to structure incentives around collaboration. Researchers must take initiative to push their work beyond the poster stage, and management must invest the time to understand the underlying technologies so they can make informed decisions about where to focus resources. I’ll re-iterate: put in the effort.

  2. Invest in maturing ideas. A promising concept should not stop at a poster. There should be support to turn ideas into research output: 2–3 high-quality papers, open-source software, open-source hardware (internally or publicly), validated datasets, internal testbeds, clear documentation, shared documentation. This requires dedicated resources and dedicated people, not just researchers doing this “on the side.”

  3. Give ideas flight heritage. If a piece of software, hardware, or algorithm is mature, we should test it. Fly it. Even cheaply. JPL could adopt a “fast heritage” model: small, frequent, low-cost CubeSat missions, perhaps every month, to validate propulsion concepts, comms hardware, onboard autonomy, radiation tolerance, electronics, and more. Nothing accelerates innovation like actual data from space.

This is a long-term effort, but it’s worth it.