[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ih] Fwd: Internet History - Commercialization

I'll apologize for my dyslexia before I start.   I've been less active on
this list so many of you are not aware of my issues with communications in
my email.   I hope my message is parsable, but frankly, I don't see the
typo's in my email until a day or two later or have someone else check
them.   I wanted to get this message out while they were fresh so I did not
sit on them for a day in hopes to clean them up a little more.

Anyway ...

I agree with most everything Jack and Vint have mentioned but ....  I think
HBS Professor Christensen explains what happen more succinctly.  The other
networking systems were examples of what he calls *sustaining technology *in
his book* "The Innovator's Dilema"*.   IP/TCP got traction in two places
which we not being served by the traditional computer firms:   Workstations
(and the rise of the technical Unix system) and eventually the PC.

Simply it won because it was the most economic solution.

Hopefully, you have all read his books completely, but sadly many folks
have not (although I hear people using his terms without really
understanding what he said).   To remind you/recap his thesis, Christensen
observes a few important facts about successful disruptions:

   1. The Leaders (the sustaining systems) are doing excellent engineering
   of their products *listening to their customers* and providing what *those
   customers are telling them.*
   2. The disruptors technology when viewed by the sustainers is a 'lessor'
   or *inferior technology* that does not provide something(s) that the
   leaders consider valuable to those customers so they are able to ignore the
   3. The new technology* is cheaper *and easier to access to some group of
   4. And this the new technology is *embraced by a new market*  made up of
   those people that does not care about the 'issues.'
   5. The new market *grows much faster rate* than the old market, to the
   point that it eventually passes the old market.

I personally loved the IP/TCP technology and was a 'true believer' in it at
the time, but looking back on the experience (*i.e.* as an old guy that
lived it) why I believe the commercial success of IP was based on
economics, not as much as the technical success.  So let me explain my

IP was developed by folks like Vint and Jack for the ARPA community and
made available to folks to like me (what Jack talked about those bring
young engineers coming out of school that new the technology).   I
certainly was brainwashed with it at CMU in the mid to late 1970s and would
go on to write the first VMS implementation of it for Tektronix
upon graduation.   In our case, Stan Smith and I picked it because we had
nothing that supported UNIX, VMS, RT-11 and the CDC/Cybers which were the
bulk the cycles at Tek at the time.  None of DECnet nor SNA were going to
solve our problem and we needed something.  IP's specs were published. We
had examples of the code from MIT for Multics and I had worked on the CMU
Front End, so I had seen IP. Plus Tek was certainly not completely ignorant
of networking/interconnect technology.   FWIW: Tek (i.e. Maurice Graube)
was chairing what would become IEEE 802 - so we knew about LANS (in fact
folks in my lab had created something we called the NIBB - Network
Interface Black Box using 75-ohm coax because there was nothing like
Ethernet chips yet. 3M Xerox ethernet existed but you could not commercial
buy the boards).

The truth is that implementations like Jack's at BBN or mine at Tek were
not 'supported' product like DECnet, SNA or the like.  The commercial folks
did not trust it - it was an 'inferior' under those definitions.  Hey, it
was referred to as a 'research experiment' in the RFCs and the like.   So
it was easy for sales types to tell managers - You should use DECnet (or
whatever) its a 'real' product with DEC/IBM/Wang ... behind it.

But roll time forward to the early 80s.   By the time the Unix Wars on the
workstations, I'm at Masscomp, we needed something for networking.   This
is true across the workstation industry.   IP was cheap (i.e. point 3 - it
was free with BSD Unix, plus there were example implementations from a
number of places by then).   BTW:  I took a lot of heat at Masscomp who was
mostly ex-DEC folks for that choice.   DECnet was the installed base at our
customers!!!   DECnet supported remote files, remote devices *etc*.   We
had none of that with IP [yet].   But IP worked good enough for us.  We
needed file copy, remote login, and email.   It was super at solving that
for us.   And we ended up with lots of UNIX boxes, far more than DECnet
systems in the wild.   What is happening we all are selling a lot of UNIX
systems, far more than the mini-computers.  The 'inferior tech' is 'good
enough' for us, i.e. point 4 comes into play.

Let time go even further.  The PC's did not start out with IP.  They
started with Novell (over ARCnet to start, transitioning to Ethernet when
Nat Semi creates 'CheaperNet.').   But each of these was local networks.
 As the need to interconnect (i.e. Metcalf's law) begins to come into play,
the cheap solution for them is IP.  Eventually, Microsoft joins the fray.
 What is interesting is when the PC displaces the UNIX workstation, by the
PC is based on IP, not Novell.

This is when point 5 becomes important, the new market has now exceeded the
old market in growth.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://elists.isoc.org/pipermail/internet-history/attachments/20190216/dba91ff7/attachment.html>