[iDC] history of protocols

[Jim Piccarello]

IP, or Internet Protocol, by its very name could not be the original protocol for computer networking. When it was introduced there were many different proprietary protocols sold by a variety of vendors. “Internetworking” was the term of art for how to get these different vendor products to work together. IP  was intended as a kind of digital inter lingua but it was so successful that it became ubiquitous and replaced virtually all of the proprietary protocols. The concept of packet switching goes back to at least the early 60s in the work of Leonard Kleinrock. He provided a good deal of the underlying mathematics that made packet switching possible.

In the analog world of telephones and telegraphs the word “signaling” would have been used before the term “protocol” came into vogue. Jennifer Cool mentioned “ground start protocol” but the term “ground start signaling” would have been used. The current protocol used to control large telephone systems is call Signaling System 7 (SS7) even though most of it is digital. A standard text issued to Bell Labs engineers in the 70s and 80s called *Engineering and Operations in the Bell System* maintains a not too strict distinction that voice and analog systems are referred to as “signaling” but data communications are referred to as “protocols.” It sounds like you’re interested in when and why the word “protocol” came into use not when the technology was deployed. The concept was there in the analog world for a long time in different terms.

You may want to look for a Scientific American article by Gerard Holzmann. He’s an expert on the mathematics of protocols but I read his article on the history of data networks. It’s about 10 years old but I vaguely remember him writing about networks using mirrors to flash messages. I don’t know if he wrote anything about the history of the term “protocol".

Also be aware that there are different categories of protocols. IP, TCP, and UDP are transport protocols, i.e. how to get data from point A to point B.  But there are other types.  NTP (Network Time Protocol) is used to coordinate the synchronization of clocks among servers in a network. DNS is a protocol for looking up IP addresses based on names. 

A general definition of a protocol that I’ve come across (and if forced I could probably dig up a citation) is that there is a set of agents that can exchange messages. The protocol is the rules for exchanging those messages and the rules are designed to achieve some goal. In the case of transport protocols there are two agents and the rules are designed to get data reliably exchanged. NTP involves a larger number of agents (computers).  I would suggest you also investigate the related concept of “distributed systems” where protocols are used to make a collection of computers look like a unified entity. The way these disparate entities can appear as a unified whole might be described as “digital diplomatics". 

One particular protocol you may want to look at is the Border Gateway Protocol (BGP) even though it’s relatively recent. The idea that the Internet consists of a collection of computers connected via a system of routers is too simplistic. The Internet consists of many collections of routers owned by a variety of corporate, governmental, and non-govermental entities. Each such entity may have one or more large collections of routers organized into what are called Autonomous Systems. An AS can connect to another AS only through a router called a Border Gateway. BGP is designed to allow business policies to control how one AS can talk to another AS. If you are on one side of a continent and you want to communicate with a computer on the other side then you will get a better quality of service if your ISP has an AS the spans the continent. Otherwise your data may have to flow through a number of different AS links. So the Internet is really a collection of entities where the exchange of data is governed by political and economic rules baked into the software that routes packets. 

So BGP seems like it merges the engineering sense of protocol with the political and economic sense of protocol. I’m too ignorant regarding economics and politics to understand the consequences of this but the topology of the Internet reflects politics and economics as much as it reflects engineering and that topology is defined by protocols in the engineering, political and economic senses.

Hope the rambling was of some use.
Jim





On Oct 24, 2013, at 3:18 AM, Simon Biggs <simon at littlepig.org.uk> wrote:

> I'd be looking at people working at places like the National Physical Laboratories in the UK and the ARPA programme in the States. This is where packet switching was developed. TCP/IP, which was the original protocol for computer networking and which has the term protocol in it, was developed by Bob Kahn and Vint Cerf at Stanford in the early 70's. I don't think there's a precise date for when it was launched.
> 
> best
> 
> Simon
> 
>  
> On 29 Jun 2013, at 11:12, Claus Pias <claus.pias at univie.ac.at> wrote:
> 
>> 
>> Dear all, 
>> 
>> I am curious if anyone did some historical research on WHY protocols were called "protocols". From the existing literature and old RFC's I vaguely know WHEN transmission protocols emerged and how the structure of packages was defined in the times of early online-systems. There are also a few texts on the history of protocol engineering (i.e. Computer Networks 54(2010) 3197-3209). But as far as I see, no one yet asked the questions why the term "protocol" was chosen. 
>> 
>> The background is that I am working on medieval and early modern documents (deeds) whose structure is called "protocol" in diplomatics (in the sense of Mabillon). In fact, the structure of digital data packages very much resembles the structure of deeds, that follow a highly formalized framework of invocatio, intitulatio, inscriptio, narratio, sanctio, corroboratio, eschatocoll  (to use the latin rhetorical terms) that are equivalent to time stamp, sender, receiver, message, 'checksum' or authentifier etc. etc. Questions of security of transmission were crucial for that kind of structure.  
>> 
>> Was anyone aware of this historical notion of "protocol" when the term was introduced to computer networks in the 1960's?
>> 
>> My apologies for such an esoteric question -- it's my first post here.
>> 
>> Best wishes, 
>> Claus
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> --
>> Claus Pias
>> Leuphana University Lüneburg, Wallstr. 1, 21335 Lüneburg / Germany
>> Professor for History and Epistemology of Media (ICAM)
>> Director, Institute for Advanced Study in Media-Cultures of Computer Simulation (MECS)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.thing.net/pipermail/idc/attachments/20131105/f8d52a0d/attachment.html>