Quote from: phillipsjk on September 01, 2011, 08:57:34 PM

Quote from: Steve on September 01, 2011, 06:49:14 PM

Quote from: phillipsjk on September 01, 2011, 05:46:11 PM

... Modern computers are simply too insecure to handle a cryptocurrency. I predict that secure, proven correct on pain of liability, computers of equivalent complexity will take generations (or about 150 years) to develop.

I disagree. We have all of the building blocks to make a truly secure language/operating system. I actually hope the rise in popularity of crypto-currencies will spur investment in new operating systems that address these problems. Windows, Unix, and OSX are all based on 40+ year old technology. It's the huge amount of infrastructure and investment built around these old OSes that is holding back progress. Not only is a secure operating system (one free from viruses, malware and the like) possible to build, it's not even terribly difficult (but it will take a lot of effort and require a large amount of investment that might require many years to recoup).

40 years is not a long time. There was about 400 years between the industrial revolution and the printing press. The printing press allowed ideas to be widely disseminated cheaply. Copyright law was invented to curtail 'piracy'. I suspect the agricultural revolution took a similar amount of time to convert society from groups of nomads to people with permanent settlements. We are barely scratching the surface of what computers can do: for the most part, they are still emulating the older analog technologies.

The reason I think a secure computer system will take so long is that it is turtles all the way down. There exists a version of the L4 Microkernel that has been verified by an automated proof checker that it implements its advertised interface. To build a secure, proven correct system, you need to do the same for the proof-checker, compiler and any hardware the system will run on. Unless you are buying a mainframe, no computer system is guaranteed to operate as advertised. Mainframes do every calculation twice because the hardware can fail at any time.

My point bringing up copyright law is that modern consumer computers can't be proven correct. Not only are they not guaranteed to work properly, they often have undocumented features built into the hardware and firmware. We have to trust that these features will only be used for censorship and not subverting any popular crypto-currency.

I basically agree with you...especially regarding the need to build on a trusted hardware and firmware platform...open source hardware is very important. For the same reasons that bitcoin could only work as an open source project, a truly secure computing platform needs to be built on an open hardware and firmware platform (where you don't have to trust anyone, you can verify for yourself that the system is secure). As for the hardware behaving correctly, fault tolerance and error correction can be built into the software that runs on top of the cheap hardware. It might not be possible to prove that such a system is 100% correct in its implementation as a result, but a 99.999% probability of correctness may be sufficient for most applications.

Operating System research tends to take a bottom up approach to the problem...I think the community of language researchers actually have a lot more to say about how to build truly secure systems...one of my favorite quotes is:

''Operating System: An operating system is a collection of things that don't fit into a language. There shouldn't be one.'' Dan Ingalls, in an article in Byte Magazine, 1981.