On: Challenge to scientists: does your ten-year-old code still run?

[sriku]

I would also place as a requirement that the code be comprehensible to someone familiar with the domain - .i.e. a "peer".