Data communications over telegraph circuits 1900

He was being *kind*! It was actually a lot worse that that.

All true. However... none of that was expected to be shared or to interact with any other company in any way. Nor was it expected to actually be a significant revenue source compared to the core business of selling long distance circuit switched voice traffic by the minute ("switched message traffic"). Data didn't begin to rival voice for revenue until the 1990's, and even then, when they *knew* it was going to become dominant, AT&T simply could not change.

The upper three-four layers of AT&T management were constantly in a battle for dominance, and the battle was *always* won by those whose history and expertise came from Long Lines.

This is the kind of thing where almost *anyone* you deal with in those layers of management can sit and talk about 25+ years experience and personal friendships with all the rest of them... and they all understand crossbar switching and things like Erlang. They understand 20 year life expectancy for equipment too, and engineering for network compatibility.

They rarely have a clue what a customer needs, but know a great deal about what they can offer... :-)

Which is to say, they were *damned good* at administration and management of the PSTN as it existed prior to Carterfone. Unfortunately that is diametrically opposed to what exists today, and with time the difference has became increasingly impossible.

Of course *all* of that was aimed solely at switched message traffic, and the fact that data could also be sent was purely coincidental until the late 1980's, by which time non-AT&T influence on the industry was becoming more significant than AT&T's influence.

Consider that until the late 1980's the vast majority of all DS1 interfaces used AMI-SF encoding-framing, and would not pbutt 8 bit data. B8SZ-ESF encoding-framing, which is aimed at data rather than voice traffic, didn't become standard until nearly 1990. That was not just AT&T either, as NTI's DMS switching system, as one example, by default was equipped with non-B8SZ T1 interfaces until about 1988-9. That, for instance, was one reason why ISDN was always "It Still Does Nothing", because to actually be effective it would have required mbuttive upgrades to the PSTN, and a "voice oriented" management could never see the gold at the end of that particular rainbow. (So modem companies came up with v.34, v.34+, v.90 and v.92, and raked in billions from the dial-up data market that the telecom industry said didn't exist.)

They did. Just not aggressively enough. WU had the same internal problem that AT&T did: stuck in a paradigm that no longer matched reality.

--