Thesis #4

The Internet is an architecture for ‘permissionless innovation’

The Internet we use today grew out of the Arpanet, a Pentagon-funded network constructed between 1966 and 1972. After the network was up and running, an “Internetworking” project was funded to devise a way of seamlessly linking the various packet-switched networks that had evolved as the Arpanet was being built.

The architects of the new ‘internetwork’ (which is where the term Internet came from) decided to work with two fundamental axioms for the new network:

1 It should have no central ownership or control.
2 It should not be optimised for any particular application.

This philosophy was informed by political’ realism (Axiom 1), and by bad experiences in the past (Axiom 2) when earlier networks devised for one purpose proved unable to cope with new purposes. Thus the telephone network — which was optimised for voice calls — had proved woefully inadequate as a way of enabling digital computers to talk to one another.

So the Internet we got — which was switched on in January 1983 — was designed to do one thing and one thing only: to take in data-packets at one end and do its best to deliver them to their destination at the other end. The network was completely agnostic about what was in those data-packets. If an engineer or an entrepreneur had an idea that could be realised using data-packets, and the talent to write an application that could do that, then the Internet (which, remember, had no central ownership or control) would do it for him (or her), with no questions asked. All the ingenuity was left to the edges of the network.

This led to an explosion of creativity, as programmers dreamed up network applications that spread like wildfire. It proved to be an architecture for what one scholar termed “permissionless innovation”. The Internet became a global engine for springing surprises. Some of these surprises — like Skype and the World Wide Web — were pleasant. Others — like P2P file-sharing of copyrighted music — were more problematic. And some — like viruses, Trojans and malware generally — were deeply unpleasant and problematic.

Further Reading

Barbara Van Schewick, Internet Architecture and Innovation, MIT Press, 2010 Amazon UK:

J.H. Saltzer, D.P. Reed and D.D. Clark, “End-to-End Arguments in System Design”, ACM Transactions in Computer Systems, 2(4) 4, November, 1984, pages 277-288. Online at:

Lawrence Lessig, “Sorkin v. Zuckerberg”, New Republic, 1 October, 2010. Online at:

David D. Clark, “The Contingent Internet”, Daedalus, 145(1), Winter 2016.

Back to list