Financial Cryptography

What is FC? (ii) - Debunking the 'Bank' View

Yesterday I outlined 3 views on "what FC is," yet there were some obvious differences. If I can call a token money system using big random numbers as FC, and the academics say that FC is crypto applied to banking problems, then we have a conflict. Banks don't buy random numbers! Which then is the more useful view?

I shall answer this by elimination, and in the process, debunk the 'bank' view. Mostly because it is easier to do that than to prove the alternate!

Firstly, banks do not buy crypto. What they buy is security. They buy secure systems from reputable suppliers, all of them called IBM. (Remember SET?) So whatever it is that banks do, that could be FC but it isn't crypto.

Banks buy systems and processes. Sometimes those systems have some crypto in them, and sometimes not. IBM are experienced in creating a complete process, and if their architecture calls for crypto, then that's what will be in there. But, it's entirely optional, it all depends on the requirements.

This is because applications are driven top-down. Requirements are laid out and they are driven down through various layers, with each layer attempting to cover as much as possible. By the time we get down to the murky depths of crypto, most all decent architects would be somewhat depressed if their crypto needs were exotic. As well as potentially out of a job; any architects who started out thinking about crypto would be totally irresponsible. Systems are built to requirements, not cool formulas.

Further, even when we get to adding in some crypto, the quantity in any given system is generally quite low. Maybe about 1% of the lines of code in a software system. (If the key exchange stuff is done properly, a couple of pages in the manual.) So, what's all the fuss about?

Finally, there is a further problem with banks - they don't do anything that other banks aren't doing. This chicken&egg problem is called 'herding' and it's an economics term of art. From this, we can conclude that banks don't do exotic crypto.

Which leaves us where? If FC is crypto as applied to banking systems, we've got some real issues in making that 'interesting': It's small. It's bottom-up. It's not sellable and it's not wanted. It's also not professionally responsible.

(Which all to a large extent explains the rise and fall of companies like DigiCash. But that's off the topic.)

It is in fact far far better to say that FC has nothing to do with the banks. The key is to turn all that around and look at other opportunities. Payment systems run by non-banks are fruitful applications. (And by non-banks I mean not-anything-like-banks.) Non-payment systems run by anyone are also important - think about trading, contract formation, reputational systems, identity, and new methods of accounting.

Or, consider it this way. The reputational auctions that Paypal helped, otherwise known as eBay, were as interesting as the payments. A large part of that success was because there wasn't a bank involved, as seen by the consistent exits by the banks.

(Next: Start from the top.)

Posted by iang at March 1, 2005 01:14 AM | TrackBack