Comment by badpun

This is why I'm a bit skeptical about the global warming predictions. AFAIK they're all based on multi-million lines of code models. Not only it's monumentally hard to have such large code base without bugs, but also, in case of a software implementing scientific models, finding bugs is extra hard (compared to software which works in a visible way and has millions of users, such as Linux kernel, games, car's embedded software etc.). An effort to have such model verified (let's say to a NASA standard) to be bug-free would probably cost billions of dollars. And that's only bugs, let's not forget that all the models are approximations, plus the numerical methods used all have their quirks and limitations etc. All in all, the problem seems too hard to tackle for humanity right now.