Choosing Unfettered General Purpose Computing over Strong Privacy

Consistency may be the hobgoblin of little minds but the Ralph Waldo Emerson quote goes on to refer to Pythagoras, Socrates and Newton and so may it doesn’t refer to logical consistency but rather to the ability to learn from new information and form a new opinion based on that. Be that as it may (any Emerson scholars here?) there seems to me a logical inconsistency in the arguments of many staunch privacy advocates. They want to be simultaneously pro strong privacy yet anti DRM and pro General Purpose Computing. These are at odds with each other.

How so? At the simplest level the requirements for privacy and DRM are roughly the same and they both entail limits on general purpose computing. If I send some information to you that I want you to read, for example my medical record because you are my doctor, you need to decrypt that record and then read it. You will either do so on a general purpose computer that is fully programmable or on a highly controlled device. In case of the former there are no definitive assurances that you as my doctor can make that my medical record won’t be intercepted by some other piece of software also executing on your machine. The only way to get to that is to have a locked down computing platform. That of course has been the exact desire of copyright holders in music and video for many years — they send you a song or video which you can play but can’t do anything else with (short of re-recording an thus losing digital fidelity).

I already described my fundamental problems with locking down computing in my post yesterday on Apple’s stance on privacy. But the issues with not being able to run whatever code you want are not limited to competition and free speech. Any lockdown means that there have to be parts of the system which you cannot inspect and hence of course you have no idea what they really do. So you wind up having to trust the vendor who provides that “trust module.” This shifts the locus of trust to entities that are less transparent (such as Apple) and less accountable. Ironically, that makes the kind of base level security — which I have compared to locking the front door of your house — ultimately harder and not easier to achieve (such as unpatched known vulnerabilities in Apple).

When you are faced with an inconsistency you have to figure out how to make a trade off. I believe that the right trade off is in favor of general purpose computing at the cost of strong privacy. Meaning: I would rather have general purpose computing and relatively limited privacy than the other way round. The best hope we have for minimizing this tradeoff (not making it go away as that’s fundamentally impossible) is to have as much open source software as possible and over time extend open source deeper into hardware design as well.