How Lisp Powered NASA’s Deep Space 1 Mission and the Challenges of Remote Debugging

On 24 October 1998 NASA launched the Deep Space 1 probe, a technology‑demonstration mission intended to validate dozens of new techniques for future interplanetary exploration.

After traveling 240 million km, a critical process on the spacecraft stopped working, forcing the ground team to devise a way to debug software that was not in a terrestrial data centre but aboard a probe half a light‑hour away.

The mission’s control software was written in Lisp, which provided a REPL (read‑eval‑print loop) that allowed engineers to send commands and inspect results remotely, despite the long communication delay.

In 1988 Ron Garret joined JPL’s AI group, where he helped develop Mars‑rover prototypes such as the heavy “Robby” and the toy‑size “Tooth”. Lacking high‑level languages like Java or Python, the team chose Lisp for its abstraction, garbage collection, and ability to create domain‑specific languages for the rover’s hardware.

Using Lisp they wrote rover navigation and obstacle‑avoidance code, as well as a compiler and simulator that ran on Macintosh computers to thoroughly test the software before uploading it to the spacecraft.

Although the Lisp‑based rover software performed well, internal NASA politics favored C/C++ for cost and schedule reasons; Lisp became a scapegoat when projects ran over budget, leading to the eventual abandonment of Lisp at JPL.

After the Deep Space 1 incident, Ron left JPL, joined Google in 2000, and despite initially leading a Java‑based AdWords project, he continued to champion Lisp, only to encounter resistance from management.

The episode highlights both the technical advantages of high‑level languages like Lisp for remote spacecraft debugging and the organizational challenges that can prevent their long‑term adoption.