On client / server architecture
Client / server architecture and exchange of information
Based on the name a reasonable expectation of client / server architecture is that it should champion the primacy of the client or user. And in some ways, these expectations are fulfilled. Functionality is provided by the server for the benefit of the client.
However, from an information perspective the reverse is the rule: server invariably has the privileged position. A client divulges information in order to benefit from the functionality provided. Though the information may be given in confidence, standard architecture puts the client in a position of having to trust the server not to break a confidence. Nor may the server give an open view of the information it has to its clients.
Information must be provided by the client in a usable format if the service is to work for the client. And this to a schema of the server's choosing, so that the client must bear the burden of translation between schemas of the various services in which it participates. There are few restrictions upon how the service provider may use the information and the few commitments that have been made in law or company policy are in many cases unenforceable.
Information is power
A dominant business model on the web at present is exploitation of information held by the server about the (user of) the client. This model tends to direct additional economic benefit to service providers who are able to elicit further information about their clients. As new information leaks into a semi-public domain, pioneers will slash-and-burn ever deeper to maintain their competitive advantage. Others will follow in their wake; setting up ranches upon what was formerly an informal commons. Clients' information thus completes its transition to an exploitable resource, and service providers become owners of the means of their exploitation.
Architecture is politics
It is not necessary to believe that any given actor does evil, in order to hold the opinion that client / server architecture can have a cumulatively pernicious effect with respect to information privacy. Because of the political significance, these are already well-drawn battle lines. The rest of this article will use personal information and privacy to frame a solution and, doing so, will establish a principle of greater generality than the application of privacy.
Subscription implies choice
It can be countered that every client is free to choose whether or not to use an on-line service and, to the extent that they are exploited, the client is therefore complicit in its own exploitation. And this argument is correct, though it does assume that there could be no alternative design of information architecture. Either my way or the highway, is the implicit assumption about surrendered information. You may either have a service and fully divulge a piece of information (perhaps constrained by a self-regulated policy on sharing with third parties), or you may choose no service and to maintain complete privacy.
The all or nothing assumption appears to rest upon a rock solid foundation. If I tell you my name, for instance, you may now provide me with services for which my name (or an ID associated with it) serves as the key. You either know my name now or you don't. If you don't know my name then, like Rumpelstiltskin, I preserve my integrity – until you discover it by some other means. But once lost, privacy is not likely to ever be recovered. The rock of atomic information granularity has been widely sought, though never actually discovered. In practice, information has proven to be a more slippery creature than logic has yet afforded us tools to work it. It turns out that natural language as an artefact of bottoms-up learning by an evolved brain over many individuals in heterogeneous cultures at many points in time is not the exemplar that it once promised to be. The best currently available is a bag of tools each of which has limited applicability within specified scope and range.
As a political struggle continues, each side of the architectural dichotomy has sought to exploit the lay of the land to its own strategic advantage. Data mining has already been mentioned as a server-side weapon in the arms race. Clients have deployed tactics such as obfuscation. Declaring a user-name instead of my real name, for instance, may help to preserve my privacy. Yet it is an ad-hoc approach that only works inasmuch as it serves to increase inefficiencies; both for the client and for the service provider. In any case, the effectiveness of obfuscation may, over time, be eroded by data mining at the server(s). Obfuscation is a blunt instrument, nonetheless, it does suggest where we should be looking for a general, scalable solution offering the possibility of mutually agreeable quality of information which would be transferred from a client in return for provision of a service.
Information degradation can take many forms. For instance, a user-name conceals the connection to a real person at the client. The server may anyway be able to infer a link from other data it now has access to, and would eventually be likely do so under some constraint of probability...
Trade not war
It is appropriate now to begin to view information as a medium of exchange. A currency, through which services may be traded. If both clients and servers have the means, then the will to