Before the internet, we sent email by punching the message on Hollerith cards, and shipped the deck by pony express.

But really, Judy Fairey, Development Office analyst 1991 - 2012, had worked for Westinghouse in Pittsburgh in the early 1960s, programming in Fortran. Her card deck was flown from Pittsburgh to Philadelphia to execute.

Only three computers needed. Attributed to Thomas Watson 1951, president of IBM. “I went to see Professor Douglas Hartree, who had built the first differential analyzers in England and had more experience in using these very specialized computers than anyone else. He told me that, in his opinion, all the calculations that would ever be needed in this country could be done on the three digital computers which were then being built—one in Cambridge, one in Teddington, and one in Manchester. No one else, he said, would ever need machines of their own, or would be able to afford to buy them.”

Chuck McClenon writes: my father worked for the National Bureau of Standards, in the mid-50's. They had the Southeastern Automatic Computer (SEAC), the second real computer in the US, after the ENIAC, at Princeton. It was so powerful that it would never lose at tic-tac-toe.

Louise Nelson: My father worked for IBM from 1960 to 1992. In the very early 70's, he often brought me with him to his office in mid-town Manhattan, and to keep me occupied while he was in meetings, he set me up to play “computer games.” His office had what looked like a dot-matrix printer attached to an IBM selectric typewriter set up in the corner, with green-bar continuous feed paper rolled into it. My dad would enter a few keystrokes on the keyboard, and in response the printer printed out a line-by-line picture of a Monopoly board, with 2 small icons on the Go block and a summary below the board showing how many of each denomination of money I had. He taught me to “roll the dice” by typing in a command, and when I did, the printer would first print out an image of the dice with my roll, then would print out the board itself, showing my figure advanced to the property indicated by my roll. If the property was still available to purchase, it would ask if I wanted to buy it: “Buy Marvin Gardens?” By typing Y or N, I could purchase property.The computer acted as the banker, subtracting the price of properties purchased and rent paid automatically from my money tally. When my turn was complete, the computer would roll the dice for itself, print out the board showing where its token had advanced, then would ask (and immediately answer) questions about its own options. The business rules for the computer's turns seemed to be that it always purchased properties and it always built houses or hotels if it had the money. These Monopoly games took hours to play since each move required printing out two views of the monopoly board line by line. At the end of a day, I had a thick pile of green-bar paper piled up on the floor behind the printer.

Chuck in response to Louise: My dad worked for the RAND Corporation in the 60's, in their Bethesda, MD office. He spent a couple of weeks at their headquarters in Santa Monica, CA, learning FORTRAN. The main computer there, the Johnniac, soon became obsolescent so it was retired from its main duties, and made available for timeshare use by employees. RAND always encouraged its employees to experiment and have fun, see he wrote a football strategy game in Fortran, where the two players could call their offenses and defenses, and enter them on a teletype machine. Then the random number generator would determine how the play developed, and would print out “Caught the pass at the 40”, “Still running at the 35”, “tackled at the 33.” One time – this is probably about 1967, he brought the teletype home with a modem. A modem at that time was a box large enough to set the phone handset into, a purely acoustical connection. He dialed long distance to the computer in California (that call probably cost $10 an hour, at discounted weekend rates). The modem was probably 300 baud – a moderate typing speed. The teletype printed wide green-and-white paper. Around 1985 my dad got a PC and wrote games in DOS. He stayed up late and didn't get enough sleep, so he retired in 1988 to have more playtime.

Read how Curtis Pew came to write webAgent, UT Austin's very own web scripting language that echoes the syntax of Natural, in his own words.

One of our more high-tech systems in the late 1990's was TEX, the telephone registration system. It consisted of a rack of PCs in the main building, connected to the mainframe via Token Ring. Each PC was connected to a phone line and ran a 3270 emulator, which were logged into a Natural application in COM-PLETE. When students dialed into TEX they would be assigned to one of the PCs, and in response to what they said commands would be entered on the 3270 screen and responses would be read back. From the mainframe’s point of view it was just a normal 3270 application.

In 1972, Robbie Simpson brought in the TOTAL database (Cincom Systems) system.

IBM introduced the 3270 terminal in 1971. http://en.wikipedia.org/wiki/IBM_3270 A group of terminals were connected by coax cable to a Control Unit, which was in turn connected to the mainframe. The role of the control unit was to bundle up the I/O interrupts of the keystrokes, and send them to the Mainframe and receive them back in blocks. On a 3270, nothing happens until one hits enter or a PF key. We were never distracted by look-ahead searches.

Programmers with 3270 terminals could log in to the mainframe under IBM's TSO (Time Sharing Option). Only a limited number of users (32? 64?) could be logged in at one time, and the system got slower with more users. UT systems programmers wrote the TP (teleprocessing) system in assembler code, circa 1974, and TP was the means for deploying applications to end-users, until it was replaced by Software AG's Com-plete. Application programmers, in COBOL, designed layouts. Security was generally controlled by the terminal ID (TID) and the known location of the terminal. 3270 terminals in the Registrar's office could use RG TP commands, and 3270's terminals in the Payroll office could perform PR commands.

When IT staff realized that the card catalogue would take 50 TOTAL files, we looked into alternates. See also Memorial Day Flood.

Data processing in 1981 was a 2-week 5-challenge process in Natural 1.0. COBOL training was 4 months.

Natural 1.2 allowed screenmaps.

UT served as a flagship site for Software AG products and hosted site visits for potential customers, earning us a substantial discount.

Starting with Natural 2.0, we started to write programs for more than ad-hoc purposes. Each shop developed its own sets of standards. Curtis Pew was the first trainee to go through training with Natural 2.0, to which he switched partway through training. Natural 2.0 had date and time fields, which was new and exciting… but they didn't work on Tuesday and Thursday. And this is where sacrificial goats entered our culture.