NASA

The Story

NASA's contribution to software engineering is measured in two currencies: the systems that worked and the failures that taught.

The Apollo Guidance Computer — 72KB of memory, running software written by Margaret Hamilton's team at MIT — landed humans on the moon. When the computer overloaded during the Apollo 11 descent, Hamilton's priority-based scheduling system shed low-priority tasks and kept the navigation running. That software saved the mission and established the principle that error recovery must be designed in, not added later.

The Space Shuttle's Primary Avionics Software System was the most rigorously tested software ever written. The team at IBM's Federal Systems Division achieved an error rate of 0.1 defects per thousand lines of code — at a time when the industry average was 25 per thousand. The cost: $1,000 per line of code, a hundred times the commercial rate. They proved that near-perfect software was possible. They also proved that near-perfect software was economically impractical for anyone who wasn't launching humans into orbit.

Why They're in the Hall

NASA is in the museum because it produced both the gold standard and the cautionary tales. Mariner 1 (1962) — destroyed by a transcription error in FORTRAN guidance code — became the most-cited software failure in engineering education. Mars Climate Orbiter (1998) — lost because one team used imperial units and another used metric — became the most-cited integration testing failure. Apollo 11's 1202 alarm — survived because of Hamilton's error recovery architecture — became the most-cited example of defensive programming.

NASA proved that software could be reliable enough to trust with human lives. Its failures proved that reliability requires verification at every boundary — between teams, between systems, between unit conventions. The museum's entire Temporal Coupling pattern class echoes NASA's hard-won lesson: assumptions about the environment, no matter how obvious they seem, must be stated, tested, and enforced.