Re: Frank Heart, Who Linked Computers Before the Internet, Dies, at 89 [telecom]

+********************************************************************+

Send posts to telecomdigestsubmissions.at-sign.telecom-digest.org +

+********************************************************************+

On 6/30/2018 5:30 AM, HAncock4 wrote,

> >> Mr. Heart's team built the gateway device for the Arpanet, the pre- >> cursor to the internet. Data networking was so new then, they made >> it up as they went. >> >> >
formatting link
> s-before-the-internet-dies-at-89.html > > Not to minimize Mr. Heart's contributions, but the article credits > several improvements to him which were actually developed back in > the 1950s. > > " To this day, many of the principles Mr. Heart emphasized - > reliability, error resistance and the capacity for self-correction - > remain central to the internet's robustness."

The article doesn't credit Frank Heart with inventing those things. He led the project that built the original ARPANET IMPs. Progress is built on the shoulders of others who came before. Much of the computer world doesn't take things like reliability seriously. BBN did.

I worked at BBN in the late 1970s (ARPANET era) and knew Frank, though I didn't work for him. I was corporate telecom manager. Frank was a VP, a Division Director by then. BBN had a rather unique culture, where computing was combined with both usability (human factors) and reliability. One of their products was the Pluribus TIP, a variant on the ARP that was used as a terminal server (this was the 1970s). It contained several CPUs. If one failed, the others would try to repair it. It was famously hard to turn off, as its CPUs kept trying to restart each other!

...

> "Data networking was so new that Mr. Heart and his team had no > choice but to invent technology as they went. For example, the > Arpanet sent data over ordinary phone lines. Human ears tolerate > low levels of extraneous noise on a phone line, but computers can > get tripped up by the smallest hiss or pop, producing transmission > errors. Mr. Heart and his team devised a way for the I.M.P.s > (pronounced imps) to detect and correct errors as they occurred." >

The ARPANET data lines were leased from Ma by Uncle Sam, not us. The backbone was type 303 modems on group channel (48 kHz wide) circuits, providing 50 (not 56) kbps. This was truly exotic for the day, when the Long Lines network was all analog.

The Bell System, IBM, MIT, and others were experimenting with this > back in the 1950s. For example: >
formatting link
\the_Transmission_of_Pulses_Over_Telephone_Lines_Apr54.pdf > > IBM developed a transceiver to transmit data, which included an > error detection and correction protocol.

BBN used many then-existing ideas, including error detection via CRC. IBM was already doing that. Modems go back much earlier, to the TWX network if not earlier. Humorously, when I returned to BBN in 1994, some of the new managers were spreading the idea that BBN had invented the modem. Which was laughable if you knew about them, but by then the history of the modem was not well known. BBN had in fact built a hack (in the 1960s) called Datadial, which was a sort of pre-touchtone interactive system, sort of like an auto-attendant, that remotely decoded dial pulses! It didn't go anywhere. I never heard of it in the

1970s, but somebody found out about it in the 1990s and confused it with a modem.