[ih] IEN Notes and INWG

[John Day]

Well, that will teach me to do things too early 
in the morning!  Sorry about that.

However, I will stand by my account and while it 
may not sound like I really don't take any of 
this personally.

In fact, quite the opposite.  I am fascinated by 
how forces (economics and politics) essentially 
outside our control were probably the primary 
determinants of the outcome.  It is safe to say 
that most of us were new the global game we found 
ourselves in.  The Europeans were probably more 
experienced than we were.  In some sense, we were 
individuals each responding locally to a much 
bigger process that none us could see the whole 
thing.

As to the details, no OSI didn't get it right and 
the Internet get it wrong.  In my book, I 
detailed what was wrong with OSI technically and 
why it was destined to fail.  It was too early 
for any of one to get it right.

TS was simpler than TCP, but still not the 
answer.  That honor goes to delta-t, which is not 
only simpler but also has better security 
properties than both.  (Watson's proof that 
bounding 3 timers is necessary and sufficient is 
one of the most important results (as well as 
surprising and astounding) in all of networking 
and little understood or recognized. An 
incredible intellectual achievement.)

Not doing CLNP was a major set back.  Had we 
adopted it in 1992, we would not be staring the 
current IPv6 crisis in the face.  And no, that 
crisis is not well in hand and still suffers from 
post IPng trauma:  loc/id split.  That problem is 
still to be dealt with.  The implications of 
which are scary.

If the Internet made any mistake, it was to not 
keep moving and leap frog where OSI was.  But as 
I have argued elsewhere, it essentially stopped. 
The bunker mentality set in on both sides of the 
connectionless/connection debate.  The Internet 
should have let OSI be tied down with the 
political debate which it couldn't escape and 
keep moving.

None of us did the right thing every time.  Least 
of all me.  Again, taking the blinders off and 
really looking at what was happening is a far 
more interesting story, with far more interesting 
implications for not making the same mistakes 
than the narrow focus of most of the work in this 
field.  It wasn't pretty.  History seldom is. 
But it is edifying.

Sorry if I upset some people.

Take care,
John




At 8:05 -0400 2010/03/22, John Day wrote:
>Off list.  And in confidence.
>
>The development efforts on individual protocols 
>was always local with common discussions at 
>INWG. So to characterize TCP development as 
>being just at INWG would be incorrect.  There 
>were always ARPA sponsored meetings on TCP 
>outside of INWG meetings.  Also, INWG was not 
>CCITTT oriented at all.  In fact, quite the 
>opposite.  CCITT's decision to go with X.25 was 
>made much earlier. The first version was 1976 
>and they were committed to it well before that.
>
>The final meeting on the INWG Transport was held 
>in London in late 1977 and first published as an 
>appendix to a conference held in Leige by Andre 
>Danthine in .  The first OSI meeting was in Feb 
>1978.  The first OSI meeting was held in 
>Washington DC in March 1978.  The second in 
>October in Paris.  I don't remember but when the 
>first official liaison contribution was made. 
>You would have to check the TC97/SC16 document 
>list.  But it was probably soon there after. 
>Tom Steel of ATT was the IFIP liaison 
>representative. It would have been a 
>contribution to WG3 on Lower Layers.
>
>Now for the in confidence part.  INWG was 
>originally created to promote TCP in 74.  When 
>INWG decided to do some thing different, it is 
>pretty clear that Vint "decided to take his ball 
>and go home."  There were US meetings in 1978 
>and maybe into 1979 in which ARPA funded 
>organizations participated.  By probably 79 and 
>definitely by 80,  Vint prevented anyone with a 
>DARPA contract from participating in OSI 
>meetings.  (I don't believe there were ever any 
>ARPA funded researchers at any of the US 
>architecture meetings (which I chaired). He 
>tried to widen it to all DoD funding, but 
>inter-agency rivalry kept that from happening. 
>My funding and Hal Folts' was from DCA (Defense 
>Communication Agency) and I remember it 
>happening.
>
>But you are not starting to tread on a touchy 
>subject.  This is where the collaboraion of the 
>international research community really begins 
>to break down.  Before this, ideas were flying 
>back and forth and everyone was bouncing off 
>everyone else.  After this, it gets personal. 
>(Notice how the Americans make a big deal about 
>packet switching being the big breakthrough, not 
>datagrams.  And how everyone believes that the 
>Internet is based on the ARPANet, when in fact 
>it is based on CYCLADES.   They have to keep the 
>role of the French out of it.  One can make a 
>pretty good case that as happens often, the 
>people who adopted someone else's idea didn't 
>really understand it.  The Internet didn't 
>really understand CYCLADES and got caught 
>flat-footed.) If you want to keep your sources 
>contributing.  I would be very careful how you 
>approach this.  The implications of what they 
>didn't understand are only now coming out from 
>under Moore's Law, so the subject is not just of 
>historical interest.
>
>This is the start of the separation from the 
>ARPA side.  There was a US group that maintained 
>participation in both and were the primary 
>movers that got CLNP developed in OSI and 
>engineered the ROAD process.  However, the 
>damage was done. The isolation of the two groups 
>had created the suspicion and animosity within 
>the IETF.  So that when IPv7 was proposed it was 
>soundly rejected by the IETF.  Leading to the 
>current impending crisis with IPv6.
>
>This is a lot more complicated.  I have a 
>presentation that I characterize as the "Guns, 
>Germs, and Steel" of networking.  That makes the 
>case that we are where we are because of outside 
>forces of economics and politics, not by science 
>and technology.  Once you have a picture of 
>where we should be, it is pretty easy to see the 
>current Internet has been basically stagnate 
>since the mid-70s living on band-aids and 
>Moore's Law.
>
>You might want to look at the preface and last 
>chapter of my book, Patterns in Network 
>Architecture.
>
>Take care,
>John
>
>At 11:58 +0100 2010/03/22, Matthias Bärwolff wrote:
>>Without wanting to go into excessive detail, up until TCP-1 (RFC 675,
>>Dec 1974) the work on TCP seems to have been "part" of INWG, and by
>>TCP-2 (IEN 5, Mar 1977) the work had been split from INWG (and the whole
>>IFIP, CCITT attendant connotation) and moved to ARPA (as Vint has
>>indicated), once the final common proposal along the Pouzin TS lines
>>(INWG 96, "Proposal for an international end to end protocol", Jul 1975)
>>turned out to go nowhere given (1) CCITT's decision to go with X.25
>>
>>I wonder: when exactly did ARPA make the strategical decision to push
>>ahead with TCP? Was there any thought of sticking with INWG and try
>>implementing their stuff?
>>
>>And, what was first, INWG approaching ISO, or ARPA deciding to go with TCP?
>>
>>(Pardon if I am being lazy, those questions must have been answered a
>>thousand times already, I take it. Thanks for the additional clarification.)
>>
>>Matthias
>>
>>Vint Cerf wrote:
>>>  inwg notes were a distinct series. IENs were 
>>>produced by the DARPA contracts
>>>  while INWG was a volunteer activity that eventually become IFIP 6.1.
>>>
>>>  vint
>>>
>>>
>>>  On Sun, Mar 21, 2010 at 7:11 PM, Matthias 
>>>Bärwolff <mbaer at cs.tu-berlin.de>wrote:
>>>
>>>>  Just a quick question: Is it fair to say the IEN Notes came out of INWG,
>>>>  or were these two different games? (There were INWG Notes, too.) Put
>>>>  differently, did the "group" that produced 
>>>>the IENs have a name of its own?
>>>>
>>>>  Thanks.
>>>>
>>>>  --
>>>>  Matthias Bärwolff
>>>>  www.bärwolff.de <http://www.xn--brwolff-5wa.de>
>>>>
>>>
>>
>>--
>>Matthias Bärwolff