One of the forgotten wars - TCP/IP versus the OSI stack - was reminded by Lynn Wheeler in a discussion of history. It was so forgotten that it took me a while to recall that indeed it was a war.
For a while, in about 1988, I was part of the standards committee for OSI, or at least the national chapter. My logic for joining was that I would learn about the future, and get a jump on everyone else.
Also, I got the protocols for free which made for a significant benefit, as no-one could afford to buy them. Every week protocol drafts turned up for review, and I read most of them diligently, looking for issues and recommendations. I didn't have much to say, because by the time they got to me, the protocols were either pretty "correct" or too big to absorb. I think others felt the same way, as I never came across any evidence that any one else was reading them.
After a year of this, I got a bit disillusioned. I'd submitted a bug in one of them - the sliding window calculations in the core TCP-like protocol were wrong - and nothing had happened. Mysteriously, the bug is still there, as far as I know. I'd noticed that the stack of protocols was exploding - upwards, sideways, backways, every way you could think of - but there wasn't that much evidence that anyone except big corporates was doing any implementation.
Meanwhile, those who wrote protocols were chugging along and connecting people, as if ISO didn't exist. And those networks were delivering value, whereas the standards bodies were not. So I left, and got back to real work.
My time was well spent. It taught me the value of real work as opposed to time wasting committees. Today's alliances and cartels and these or those initiatives are no different. Quick test - which Internet alliance has deployed a real successful product? Longer test, what percentage do you get dividing that number into the total number of cartels?
For old timers, Lynn's words are spooky. I was there, I was!
some of this was replayed in alt.folklore.computers a number of different times over the past 8-10 years. i have a little of it archived at http://www.garlic.com/~lynn/internet.htm as referred to in the above ... there is big difference between technology and operational service. while tcp/ip came into deployment in 1/1/83, a big change-over was the deployment of NSFNET1 & NSFNET2 backbones. These were NSF contracts with commercial entities. There is something to the effect that commercial entities sunk into NSFNET1 between four to five times what NSF actually paid in the contract (as part of promoting an operational, viable infrastructure ... and even more in NSFNET2).
misc. additional refs at:
http://www.garlic.com/~lynn/rfcietf.htm#history
in the late '80s and the early 90s the federal gov. had switched was actively trying to kill tcp/ip and was mandating OSI ... there was a bunch of stuff about GOSIP ... the federal government mandate to eliminate tcp/ip and move to OSI. some specific posts on gosip
http://www.garlic.com/~lynn/99.html#114
http://www.garlic.com/~lynn/99.html#115
http://www.garlic.com/~lynn/2001i.html#5
http://www.garlic.com/~lynn/2001i.html#6
at Interop '88, well over half the booths were demo'ing some sort of OSI related stuff. misc. past post/thread talking about OSI
http://www.garlic.com/~lynn/subnetwork.html#xtphsp
misc. posts discussing interop '88
http://www.garlic.com/~lynn/subnetwork.html#interop
one of the issues (during this period) in the ISO & ANSI standards bodies was there was a mandate that no standards work could happen on anything that didn't conform to the OSI model. That represented a big problem. The "IP" or internetworking layer doesn't exist in OSI and therefor nothing could happen that involved internetworking (strong assertion that the internet is successful precisely because of the internetworking layer ... which is forbidden by ISO & ANSI as not existing in the OSI model. Another thing that doesn't exists in the OSI model is LANs. I was involved in some work with high speed protocol to standardize in ISO/ANSI. However it was rejected for 1) it went directly to the LAN/MAC interface (forbidden because LAN/MAC doesn't exist in the OSI model) and 2) it wen directly from layer 4 to MAC layer (bypassing the layer 3/4 interface, also a violation of the OSI model).
random other arpanet, nsfnet, internet postings
http://www.garlic.com/~lynn/subnetwork.html#internet
concurrent with the tcp/ip in the early & mid '80s was bitnet in the US and earn overseas
http://www.garlic.com/~lynn/subnetwork.html#bitnet
specific post of NSFNET announcement
http://www.garlic.com/~lynn/2002k.html#12
misc. additional NSFNET references
http://www.garlic.com/~lynn/2000e.html#10
--
Anne & Lynn Wheeler | http://www.garlic.com/~lynn/
Said to say, you are correct. The wasted time in table items kills any effort to achieve. It is however required that once a path has been blazed finding a means of taxing it is next, and in this area the table groups work perfectly. So for new and exciting services that deliver value regardless of its purity in approach, table groups are at their worst. In the area of defining a blazed trail and getting a portion of the revenue, table groups can not be beat. So if taxing people becomes an exciting thing that requires new and interesting approaches we may have some hope of never paying a dime.
Posted by: Jimbo at July 29, 2004 11:33 PM