Cost: Teletype 33 vs. IBM Selectric Terminal 2741

Actually, there were a bunch more: - CRT terminals, async as well as sync block-mode - dot matrix printing terminals - daisy-wheel printing terminals

Most of these started proliferating from about 1972 onwards.

A number of other companies made terminals based on the IBM printer mechanism. I first encountered Selectric-style typewriters as papertape punches (Friden Flexowriter) used to prepare input to GIER computers at the computer science department at University of Copenhagen in 1970.

I think it was in 1972 I headed up a project to develop a 2741 terminal driver for Univac EXEC-8 operating system. We had had a driver (developed by ITEL, a vendor of 2741 clones) but it was being obsoleted by e new OS version where the internal interfaces had changed.

Let's face it: In 1971-1972 minicomputers were very new and still hard to afford. The successful vendor had to create the perception (and the reality) that they were aggrewsively driving down prices to deliver the most machine for the least money. The TTY-33 was ugly and looked robust (although it took a TTY-35 to survive extended duty cycles). The Selectric was elegant but quite finicky. And did you really want people tying up the computer while producing nice-looking printouts?

On the mainframe side, there were different clbuttes og systems, with only the PDP-10 covering the whole spectrum: - batch mainframes had very poor support for interactive terminals. Synchronous block-mode terminals were the only practical means of supporting transaction processing such as banking and airline reservations. They reduced the need for buffer storage for in-process transactions by allowing most of the work queue to sit on the terminals until polled. So when the load was rising, the system could reduce polling rates to maintain stability. Asynchr5onous terminals could not do that - time-sharing systems ran a multi-threaded BASIC interpreter. Modems were new and not very good yet. The Selectric terminals has several modes and a lost byte could cause them to get locked up in a wrong mode causing a need for user support. TTYs were much simpler, and a lost byte just caused one wrong character to be sent or printed.

Lars Poulsen