NASA made an interesting decision with the Artemis II computer: instead of trying to prevent errors at all costs, they accepted that errors will happen. In space, radiation can corrupt data and cause unexpected failures. You can’t guarantee 100% stability. So they changed the approach. The system was designed to keep running even when something breaks. They use redundancy, multiple modules doing the same job at the same time. If one fails, the others take over. The architecture is modular, which allows faulty parts to be isolated without bringing everything down. And maybe the most interesting part: the computer can reconfigure itself in real time. A lot of this is possible because they use FPGAs, which allow hardware to be changed through software. In the end, the lesson here isn’t about space. It’s about engineering. Truly robust systems aren’t the ones that never fail. They’re the ones that fail well and keep going. Worth thinking about when designing distributed systems, critical backends, or anything that needs to scale for real. Source: https://cacm.acm.org/news/how-nasa-built-artemis-iis-fault-tolerant-computer/