Bitcoin and the Agency Problem
Whereas for earlier moneys, it was generally the case that most people could expect to be better off storing the bulk of their money with agents, for Bitcoin the opposite is true. The Bitcoin network obviates traditional financial institutions and there are really only two agents that Bitcoin users need: exchanges and payment processors. The exchanges are temporary; they are only required because fiat currencies require them. Once the fiat currencies have died, centralized currency exchanges will no longer be required. As for payment processors, they are only required over time scales of about an hour or less to ensure that a transaction gets into the block chain. With Bitcoin there is no need for anything like a bank that would store bitcoins over a long term.
Being nearly superfluous is not the only problem with Bitcoin agents. To whomsoever a private key to a Bitcoin address is known, the bitcoins it holds are far easier to control than dollars or gold are under almost any circumstance. This is because controling bitcoins is simply a matter of keeping a number secret and requires neither securing physical matter (as with gold) or the permission of issuers (as with dollars). Thus there is a fundamental agency problem with Bitcoin. Agents can disappear or simulate a heist upon themselves. They cannot rationally be trusted without extreme costs being imposed on them which are more stringent than traditional banks. Furthermore, agents must advertise their services and consequently become the targets of hackers. They must take extreme measures to achieve a level of security similar to that which an individual person achieves with much less.
The inherent agency problem in Bitcoin is borne out by Bitcoin's extraordinary history of hacks and thefts. Bitcoin institutions seem to be worse than the banks that Bitcoin replaces. As Bitcoin proponents must point out repeatedly to naysayers, people lose bitcoins not by an inherent flaw in the protocol, but by misuse. They stored them with Mt. Gox, Atlantis Market, or one of many other companies that imploded or disappeared. Bitcoin has had a problem with amateurish entrepreneurs, but I expect that we will see continued failures even when established institutions attempt to use Bitcoin.
Despite this, there are still a lot of Bitcoin agents. The practice of storing bitcoins in web wallets, like those provided by CoinBase and Circle, abounds. People are still used to dealing with banks as a necessary evil and are not yet comfortable with the skills required to safeguard a private key. This is a problem because it means Bitcoin adoption requires a real change in peoples' habits. Everything that helps people to learn new habits is greatly appreciated.
The only solutions to the agency problem are either to remove agents or to remove their agency. In other words, people must either take control of their own bitcoins or, if they do not wish to, they must cede control of their bitcoins in a way that does not give control to an agent. The only way to do this is to spread the control over many different parties who are not expected to collude. This could be done with multisig wallets or Open Transaction voting pools. This is a fairly drastic solution because it means no one can do anything with the bitcoins on his agency alone.
Bitcoin Institutions
The agency problem becomes especially interesting when it comes to Bitcoin organizations because an organization has no autonomy of its own and all its members act as agents to it. All, whether employees or owners, have a similar incentive and ability to steal that other Bitcoin agents have. This is not a new problem, but Bitcoin gives it a new character.
An organization inherently cannot take control of its own bitcoins, so its only solution is to remove the agency of its own employees and owners. It is not enough for an organization to distribute keys between its own people, although that will help. However, it is much easier for people within an organization to collude than without, and furthermore within an organization people will tend to have similar characters so the probabilities that each is a bad actor are not independent of one another. Votes must come from different organizations on order to maximize the security of the wallet.
Criminal Organizations
Whereas individuals can do whatever they want with their bitcoins, organizations must seek outside approval. They must subject themselves to constant auditing in order to deserve any measure of trust. It has long been understood that Bitcoin gives more control to individuals, but the complement to that is not yet well enough understood. I believe history has shown it to be an important effect, but it is not yet clear how important it will be. Maybe people will be able to work around it or maybe they will have to replace most organizations with distributed systems.
Of course, the requirement of being constantly audited and losing control over the funds it holds is less of a problem for honest organizations. It is much worse for criminal organizations or for organizations that want to keep their internal operations secret. Mafias, cults, and governments will have a greater difficulty adapting to Bitcoin than will publicly-traded companies.
All organizations tend to evolve so as to resist change, but governments, being subject to the problems of socialism, suffer much worse from this because government operations lack a clear concept of efficiency their overall success, the relative importance of any of its parts, or the relative merits of alternative organizational structuring. Consequently, governments can more easily evolve into labyrinthian structures that nobody understands without anyone realizing what is happening.
An eye-opening article called Sinkhole of Bureaucracy describes a surreal example of this phenomenon in an outstandingly incisive way. In an abandoned Pennsylvania mine, which is now an office containing 600 federal employees and endless filing cabinets, process all the federal retirement pensions on paper by hand. The system is widely understood to be insane and dysfunctional, but despite repeated and ceaseless attempts to automate the process beginning in the early 80's, the system has not changed. It is not a problem of will, but of knowledge: there is no one available with the skills to carry through the transition successfully, no one who knows precisely what those skills would be, and no one who can evaluate anyone else for them. As a result, the attempts to develop an automated process failed because the software engineers did not understand the laws and the bureaucracy well enough to design something correctly, and the bureaucrats did not know how to tell if the software engineers knew what they were doing.
Will the federal government be able to adapt to Bitcoin? This would require building an a system not just for the one department, but for the entire organization, and it would have to be built properly—it must distribute decisions enough so that bitcoins cannot be stolen easily by employees. After reading that article, I think it is reasonable to think that the government may not be up to such a task at all.
http://www.youtube.com/watch?v=mu1iND6vtcE