You've probably already heard by now that NASA engineers have reestablished an intelligible link with the the Spirit Mars rover. They believe their problems are related to one or more of the rover's flash microchips, a solid-state memory chip you might find in your digital camera, game console memory card, or USB pen drive. The rover has 256 megs of this stuff, apparently used as a filesystem.
NASA got the rover back online by using a "RAM drive", consistent their bad flash theory. (A RAM drive creates a filesystem in the computer's RAM, which will be lost on reboot.) One thing to note—flash chips are fast when it comes to reading data from them, but very slow and often timing sensitive when writing to them. NASA is currently talking to the rover at 120 bits per second, a slothful communication speed not seen here on Earth in many years, which I have to imagine would make it very difficult to try to reprogram one or more of the chips. When writing data to flash, these chips expect data at a certain rate; if NASA couldn't feed data to the flash fast enough—at 120 bps I would consider that likely—the chip would probably "time out" thinking the computer was prematurely done programming it. This would cause the chip to stop accepting data and go back into a "read-only" mode. You can see the problem here. If they can perhaps identify which chip may be acting up, if the problem is indeed a bad flash chip, they might have to update the rover's software with logic to avoid using the misbehaving chip.
Sheesh. I think I have problems debugging embedded software. I'd hate to have the job of the NASA techies who have to fix this mess. Good luck, guys.