The problem with the iPad and Facebook
April 17, 2011
I loved Napster.
The original Internet was designed as a peer-to-peer system, like Napster was. Up until around 1993, the Internet had only one model of connectivity. Computers were assumed to be always on and always connected. The goal of the original Arpanet after 1969 was to share computing resources through integrating networks and allowing every host to be an equal player. Any two computers on the Internet could send packets to each other. Firewalls were unknown and communication patterns were by default symmetric.
Reach together with symmetry and equality were the things that made the Internet such a radical social innovation.
The explosion of the Internet in 1993 – 1994 was largely the result of the web browser and a different logic: the client-server protocol. The client initiates a connection to a known server, asks a question, downloads the answer and disconnects. The device running the client doesn’t need to have a permanent address. It does not even need to be always on. This is the reason why broadband providers gave us asymmetric bandwidth. More bandwidth is offered when getting data from the Internet than when sending data to it. The assumption was that the majority of users want to download and consume, not upload and produce.
It was not about symmetry and equality any more.
The client-server model was not the only development that changed usage patterns. The original model was transformed even more as a result of firewalls. Now the hosts of the network could not talk freely to other hosts because of firewalls creating obstacles to communication.
One of the most common and widely spread social developments is people being able to be their own authors and publishers. What Napster did was a different and likewise revolutionary social innovation. It came up with a third alternative, a new logic between producing and consuming: every computer in the network was used as a re-publisher and curator.
The assumption that there were few publishers and many consumers did not hold any more. Napster changed the flow of data.
The real genius of Napster was the way it made collaboration automatic. By default, a consumer of files was also a producer of files for the network. Once somebody downloaded a file, her machine was available to pass along the file to other users when needed. A central addressing authority connected the nodes of the network and then after that left everything else to take place by itself.
The totally transparent architecture produced value as a by-product of people getting what they wanted. No altruistic sharing motives were needed
Napster was a very decentralized system with some important centralized elements. In a decentralized system every host in the system is an equal participant. No hosts have facilitating or administrative roles. But Napster was also a search engine. It maintained a master song list adding and removing songs as individual users came online. This created redundancy and led to a high probability that a given file could be found although the probability of a given user being online is very low. As a result the contribution of one individual is very small but the collaborative interaction of the group creates tremendous value.
In a centralized, hierarchical system, coordination between peers is controlled and mediated by a central server, one host. A modern version of a hierarchical system transfers some coordination responsibility down from the centre to a tree-like architecture of coordinators. In this model, peers are organized into groups, where a local manager/host mediates communication between peers in the same group, but communication between peers in different groups is passed upwards to a higher level manager. This is essentially the way firms operate today.
Ronald Coase developed the concept of transaction costs. These are the costs of coordinating actions and the costs of interacting and contracting. When it is cheaper to do this inside a formal organization than as a network of more or less independent parties, organizations will form and prevail.
The reverse side of the Coasean theory is even more interesting. As transaction costs outside the organization fall as a result of technological and societal advance, the reasons for formal coming together dissolve. This leads to the organization becoming outdated, unless it can simplify its processes significantly. The big challenge for many organizations is to do things in a much, much simpler and more responsive way. The sad truth is that it is easier for managers to grasp the threat of competition than the risk of simply becoming obsolete.
In theory, if transaction costs in society at large become low enough, there will be no hierarchical, formal organizations as we have known them. The transaction costs of forming and maintaining these types of organizations are higher than the transaction costs of the alternative ways of creating the same value. The traditional hierarchical and formal organization is just too complicated, slow, and far too costly as a system. Unfortunately, the mainstream business schools haven’t figured this out yet. They still keep on teaching yesterday’s pricey way of doing things.
Peer-to-peer is an architectural model that is much more interesting, but also much more demanding, than the dominant client-server models. I believe that Napster gave us a glimpse of the future. The architecture it pioneered is going to be a viable model for the agile value constellations of the very near future.
Client-server is not the only truth and Facebook is (just) a modern version of a Telco. Facebook is not the same as the Internet.
Thank you Larry Lessig, Clay Shirky and Andy Oram
More on the subject: The early history of the Internet. Blog post by Doc Searls. Blog Napsterization.org. On personal data. Personal leverage for personal data by Doc Searls. On user-centric identity. Blog post by Venessa Miemis.