2009/2/19 Luke Palmer <lrpal...@gmail.com>
> 2009/2/19 Rick R <rick.richard...@gmail.com> > >> I think the capabilities community including E and Coyotos/BitC have >> extensively addressed this topic. Coyotos is taking the correct approach for >> trusted voting platform. Since, even if your software is trustworthy, it >> can't be trusted if the OS on which it runs is suspect. > > > Woah, that's a pretty interesting question! How do you write software > which is protected against a malicious operating system (mind -- not > erroneous, but rather somebody detecting the software you're running and > changing your vote). Maybe some sort of randomized cryptographic technique, > in which, with high probability, the OS either runs your program correctly > or causes it to crash. > > It gets worse. Even if you write your OS in Haskell, how do you know your compiler hasn't been compromised? Or the hardware? The solution necessarily involves a social component, e.g. Haskell, with the development practices of OpenBSD (continuous re-auditing of everything including tools, complete openness, etc.) IOW, it'll never happen, but it might end up better than paper ballots.
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe