The Internet was alive and well before the web came along. It was certainly a different beast at the time but the author is spot-on when observing that the distributed, anarchic nature of ARPANET lies at its foundation.
There is also an alternate scenario that might be interesting to speculate about. Before wide-spread internet availability a lot of people were connecting to (mostly) privately run BBSes. Inter-BBS networks like FIDONET were developed for passing email between FIDONET nodes with gateways into other systems (including, IIRC the internet).
In a world without ARPNET based internet it's not hard to imagine commercial services being built around this. Given enough volume these services would start setting up dedicated links to each other. Enough services start doing this and it makes sense to set up exchange points. All these services rely on the same protocol to communicate over exchange points and that would make proprietary protocol changes difficult so a more-or-less open engineering consortium would be the next logical step to push such changes. Real time communication over this network would be a real commercial selling point (hello chat systems!) so next thing you know you have 'special' emails with high priority and some extra flags that don't end up in email boxes but are sent to end-user programs that only send and receive the 'special' emails. (Hey there protocol selection!).
This wouldn't get us the web any time soon (this kind of network wouldn't have room for arbitrary data floating around), but market forces + social momentum could conceivably build an alternative version of a global data communication network.
There is also an alternate scenario that might be interesting to speculate about. Before wide-spread internet availability a lot of people were connecting to (mostly) privately run BBSes. Inter-BBS networks like FIDONET were developed for passing email between FIDONET nodes with gateways into other systems (including, IIRC the internet).
In a world without ARPNET based internet it's not hard to imagine commercial services being built around this. Given enough volume these services would start setting up dedicated links to each other. Enough services start doing this and it makes sense to set up exchange points. All these services rely on the same protocol to communicate over exchange points and that would make proprietary protocol changes difficult so a more-or-less open engineering consortium would be the next logical step to push such changes. Real time communication over this network would be a real commercial selling point (hello chat systems!) so next thing you know you have 'special' emails with high priority and some extra flags that don't end up in email boxes but are sent to end-user programs that only send and receive the 'special' emails. (Hey there protocol selection!).
This wouldn't get us the web any time soon (this kind of network wouldn't have room for arbitrary data floating around), but market forces + social momentum could conceivably build an alternative version of a global data communication network.