There is a considerable amount of confusion involved in the reporting of today's FCC network neutrality decision. Rather than getting into it and my bias, I suspect it might be more useful to spend a bit of time talking about what the Internet is and isn’t.1
A few weeks ago I wrote a post noting there were myths about the Internet and hinting I needed to say more. That’s a good starting point. I don't have much time tonight, but it is the right time to talk about the basics. Don't think of this as a history. It just highlights a few bits of brilliance.
Packet networks were invented in the 60s. The basic notion is information could be encoded into a digital form and grouped into packets that could be guided around a network. In the late 60s the spiritual predecessor to the Internet, ARPANET, went live linking (sort of at first) UCLA, SRI, UC-Santa Barbara and the U of Utah. A few other networks were established and a few services came along (email in 72 from BBN). By the mid 70s there was a Cambrian explosion of network types. There was a lot of freedom in what you could do and funding from the military (in the US at least). Engineers *love* to optimize. They would understand the problem at hand and go out and invent a protocol and network type optimized for that particular problem. The big engineering universities were supplying professors and grad students with dozens of Ph.D.s being generated on very lovely specialized solutions.
An exciting time, but two events changed the world. Bob Kahn and Vint Cerf proposed a very simple protocol and ARPA did something unexpected and wonderful.
The folks at ARPA worried a tower of Babel was being constructed. They were charged with coming up with something ultimately useful to the military and sought commonality. The simple solution was to declare a winner, but ultimately technology would have obsoleted it. Instead they formed a working group of the best computer scientists looking for something that would scale across technology, size and time. A serious rethink of networks was launched.
The realization was the heart of the matter was simple - just messages and addresses. A network sends a message from one address to another and a reply is returned telling the sender the message has made it. The protocol for what a packet should look like is published and it really doesn’t matter what the physical network is - it could be coax cable, a radio signal, a flashing light or even sound waves underwater. You could connect groups of these networks into something larger as long as everyone agreed to the same protocol and someone would construct a gateway to let the messages flow across boundaries.
Technically it was an overlay protocol. For specific applications it generally isn’t efficient. Engineers hate this sort of thing. AT&T pulled out sensing it was ugly and, of course, it violated AT&T’s core DNA of centralized control.
It may not have been locally efficient, but it was pure beauty if you wanted something that worked for almost everything. They had realized a great truth:
Optimization that becomes standardized kills innovation.
ARPA had created a working hourglass model. It didn’t depend on the technology underneath (the physical part of the network and its operation) and it doesn’t control or anticipate the types of uses that appear on top. It is close to a minimal set of what can work fairly well while allowing technology to constantly change and new service types to be created.
A mechanism where proposals for protocol changes and new service types that were low level was made.
People began to play. It suddenly became very useful to stitch networks together. In the late 70s this was ugly (I was involved in bringing two together as a hobbyist). By the early 80s a protocol was agreed upon and peering arrangements were negotiated between groups - ‘I’ll carry your traffic if you’ll carry mine’. Lots of innovation and technical change, but that thin middle layer was largely untouched.
It started to be known as the Internet. Noncommercial and mostly researchy those who used it in the day were building our own little clients and services for their own needs. The late 80s saw rules from Congress (largely Al Gore who was trying to fill his father’s TVA shoes) that led to explosive growth on campuses. Later commercial traffic was allowed...
AT&T and probably other carriers were happy to sell T1 and T3 lines to Internet service providers and colleges to support this quirky little non-theatening network. They had their own plans for a centrally controlled network - as late as '93 or '94 they believed the big packet network would come from specs to be written by the ITU in Switzerland.2 The carriers were sideswiped and stunned when the commercial Internet finally became available to normal people using this nifty world wide web that came from physicists at CERN - folks who had a problem and figured a solution could easily ride on the Internet.
That’s it - the history isn’t terribly important. It is just addresses and messages. Optimization was the enemy. There are very few examples of technologies that scale across time and technology. The Internet is probably the most important example.
__________
1 There are many legal issues in the FCC's rule. Barbara van Schewick is a Stanford law professor who specializes in the topics. Check out her blog to stay up to date for a deep dive away from the ISP bias.
2 It should be noted that most of research at Bell Labs was convinced the Internet would ultimately rule. David Isenberg's dumb network paper was somehow allowed to be published on the outside, but it was an articulate and thoughtful summary of the feelings of those of us who were in play mode.
__________
Recipe corner
In the spirt of a quick post just a simple idea. Winter squashes are still good, but dealing with big hard ones can be frustrating. Here's a trick. Wash the squash and find a pot large enough to hold it fully immersed (more or less). Put the squash in the pot, add water and now remove the squash. Heat the water to a simmer and now add the squash. Let it simmer for two or three minutes. If it wasn't fully immersed, take it out and flip it letting it simmer another two or three minutes.
Now lift it from the pot and let it drain. You need to be careful as having it fall back in the water can splash and scald. - I use a couple of big spoons.
It should peel very easily with a normal vegetable peeler and the flesh slices easily.
Comments