Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
Computing in the Old Days
No, not back to Babbage, but some vignettes from the mid-’60s.
When registering for classes in college, after lining up all your classes, you were sent to the “computer” to have the tuition calculated. The “computer” was an upper-classman with a calculator sitting under a large banner which said “Computer.”
Working at a high tech company of the time:
The programmers were women and the engineers were men. Once projects started which depended on mini- or micro-computers, engineers had to learn to program. I started working in the industry during that transition. One of the women programmers (Freida Robey, thank you) helped me a lot by mentoring me and helping me write a program that my manager said they didn’t need. I finished it and he used it all the time.
The computer room was:
- A fairly large room filled with lots of large cabinets with fans.
- It had a raised floor.
- It was the only place in the building where smoking was not allowed and it was always cool.
- The input was a punched card reader and, when not used, the cards were kept in large cabinets with a press for each set of cards to keep them from warping.
- The main output was a line printer which printed each line with a particular ripping sound on the familiar striped wide paper with the perforations on the sides.
- Initially, there was no time-sharing. One person “owned” the computer at a time. My career was boosted because I was willing to take the night shift.
- The main panel had lights showing program location, data and so on. It also had lots of switches to control things. You could even single step it through the code, although that was usually not very useful.
Nerd Humor of the time
When inputting a program to the main computer, if an error was found during the pass, an error message would be printed on the line printer. I don’t know what it meant, but the message always started with “FRSOPN,” followed by an error code and a long line of asterisks. Because most of the line was the same character, all positions printed at the same time and there was a “whap” instead of the usual rolling/ripping sound. That distinct sound would cause the programmer who was trying to run the program to leap up and run over to see what the error was.
The trick was to slip a punched card with a comment character in the first position and followed by FRSOPN, a reasonable error code, and the asterisks. Usually, it would take a while for the programmer to notice the comment.
Our computer technician had a “trick” of taking a programmer’s card deck and tossing it from hand to hand. Unfortunately, he sometimes dropped the deck. Because of accidents like this, all program card decks had a diagonal line across the top so you could see immediately if any were out of place.
The computer room temperature was critical and was monitored with a recording thermometer. When I was working the night shift, for several nights in a row (it was very boring work) I would breathe on the sensor so the recorded temperature would go up about 20 degrees. This drove the computer manager nuts. He was planning on coming on to see what was happening when I confessed.
Minicomputers
When minicomputers first came in, they were lots of fun. Like the mainframe, there were lots of lights and switches, but it seemed much more personal. Initially, everything was done in assembly language and the input was the paper tape reader on an ASR-33 terminal and the output was the ASR-printer. The minicomputers were 16 bits, so there were 16 switches for the registers. The initial “language war” was if code and data should be expressed in Hex or Octal. Octal used 3 bits and could express numbers from 0 to 7. The problem was that a 16-bit value took 6 characters to express — that is, a one or zero for the first bit and then 5 sets of digits for the next 5 sets of 3 bits. Hex expressed the value for 4 bits and was expressed as 0-9 and then A-F. Hex took only 4 characters to express a 16-bit value.
I was a Hex guy. One of the early minicomputers I worked on needed to have a “bootstrap” loader entered manually before it could be started in the morning. I got used to using a 4-finger button push to do it quickly.
The equivalent to today’s “Hello World” program as a first program in a new language was to write code that would rotate the data bits (visible on the console) at a slow enough speed that the action was visible.
Even though the programs were written in Assembly language, the compiler took two passes. This meant that the paper tape needed to be run through the ASR-33 twice. As the program grew longer, managing the paper tape so it didn’t get hung up was a major issue. I remember taping pencils at strategic points through the lab to run the tape over so it wouldn’t snag. High-speed input was when mylar tape and optical tape readers were introduced.
Memory was extremely limited. On one project, we needed to add more — something like 64K bytes (thousands of times less than in a current cell phone) and it came in a separate 4-foot high cabinet. The salesman would bring customers from miles around just to see it.
At one point, I was working with a Raytheon 703 or 704. Raytheon had just come out with a Fortran compiler and I tried it, but it always crashed at a certain point in the compile. I called our Raytheon tech support guy who said, “I don’t know what you are doing wrong, the University of North Carolina is using it with no problems at all.” I gave up on it. A month or so later, he stopped by and I showed him what happened for me. Of course, it crashed again. His response: “Damn, just like at Carolina!”
In spite of all the limitations, you could do an amazing amount with them. The company where I worked did pattern recognition and had been working with speech recognition. This was back in the ’70s where speech recognition had to be trained to a specific user and could not deal with continuous speech — words needed to have a clear separation. I developed an Operating system which could handle 4 simultaneous speakers, process the input and drive up to 4 printers and 4 speech synthesizers. My name for it was VOICE for “Voice Operated In-Core Executive”, but marketing didn’t like that and changed the name to something like VDETS.
Microprocessors
When microprocessors first came out, I convinced my bosses to let me get one to do some testing. I got a board from Intel based on the Intel 4004 (predecessor of the 8008, 8080, 8086 and on and on). It came on a board which was about 7″ by 14″ and had a whopping 1K bytes of memory. This was divided into 4 rows of 8 chips and ran hot enough to fry eggs. If you had a question and called Intel, you talked to the guy who designed the board.
A program had to be “burned” into an eeProm — a programmable chip that could be erased using UV light. Unfortunately, the eeProms cost about $170 each. I convinced my boss to let me get 4. Then, I had to write an assembler — I used the mainframe computer and was then faced with actually programming the chips. It is amazing to remember this, but I designed and built a programmer which was driven by the mainframe. I tried the first chip and after it was programmed, it didn’t work. I rechecked everything and it seemed ok, so I tried the second chip — same results. At this point, I was starting to get worried. I rechecked again, tried the third chip and got the same result. At this point, I was really sweating, thinking I had burned up about $500 in chips. I tried the fourth and it worked fine! It wound up that three-quarters of the chips Intel sent were defective.
Personal Computers
When the Apple II first came out, it was delivered with a manual (the “red book”) which had the schematics and all the source code.
The main thing I remember about when PCs first came out was the notion that they would never “catch on.” In particular, managers would never use them, since typing was what secretaries did.
Those were the good old days. What stories do you remember from the beginning of your career?
Published in General
I started working with computers in 1973 between my senior year in high school and my freshman year in college. On the strength of one programming class in high school, I got a job operating an XDS-930 (actually an SDS-930, but Xerox had bought out Scientific Data Systems.) It was a mainframe, albeit, a small one.
The computer had germanium circuits, so the computer room had to be kept cold (like 60 degrees). I got strange looks bicycling to work on hot summer days with a fall-weight coat tied around my waist. My job was mounting 9mm tape reels on the system, and starting up a data reduction program that ran until the tape was exhausted. Then I rewound the tape, and mounted a new one. Three tapes over a 4-hour shift.
Since the machine was dedicated to data reduction of satellite telemetry, it went unused when I was not working. I got permission to use the computer after my shift. It was granted because my bosses felt I would learn more about it. I was in my freshman year as an engineering student, so I was writing Fortran programs to solve my homework assignments. Which meant in 1973-74, I had my own personal computer, even if it filled up the room.
I also programmed the thing to generate damage cards for the Fletcher Pratt naval wargame. Enter the ship’s weapons, displacement, armor, and speed and out came the cards to allow you to use the ships in a wargame. I soon had binders filled with every ship type used in WWII from every navy. (Even such obscure ones as the Netherlands Navy and the ABC South American powers.)
Your Raytheon tech support guy sounds like the world-famous Digital Field Circus guys. Why did we call them “Field Circus?” Because they were so much fun to watch.
When there were problems with the computer, they were generally some sort of timing issue. His fix was to adjust the power supply voltages on one bank of cabinets and lower them on another. Eventually, he ran out of adjustment range and moved to Canada.
* On the other hand, he gave my wife and I a nice set of steak knives for our wedding, so I still think of him fondly.
*Edit: My wife says the knives didn’t come from him, but the voltage adjustments were true.
So you came in second that month in sales?
Q: How does a Field Circus engineer fix a flat tire?
A: He changes the tires one by one until he replaces the one that is flat.
———————
Q: How does a Field Circus engineer fix a dead battery?
A: He changes the tires one by one until he replaces the one that is flat.
(They had a script and they were sticking to it, dammit!)
I started programming in the mid-60’s in IBM assembly language producing programs to operate on an IBM 360 mainframe with a 16K memory, about 12K available for the application programs. That was not much to work with even in assembly language, but there was random access storage on disk so swapping program segments in and out of memory progressed rapidly. This new experience for me followed several years managing various commercial bank operating departments including branch teller operations, proof and bookkeeping operations. This move into the computer world was what really prepared and led me into a long career in automating commercial payment operations that culminated in my eventual position as Director of Regional Financial Operations and Chief Disbursing Officer for the U.S. Treasury’s Financial Management Service (FMS).
During my time at Treasury, I was involved in the management, which included the development, implementation, and operation, of the conversion of making government payments by check to making government payments by direct deposit into financial institution accounts. First up, in the late 70’s, was the conversion of Social Security recipients to direct deposit (tens of millions of payments each month). Other payments, tax refunds, VA benefits, federal payroll and vendor payments came later. This was what we sometimes think of as a major cost reduction and improvement in operations. Some didn’t opt for direct deposit (most did, eventually) so we converted the old punched card government check to paper check form for the payments that continued by check.
That’s all folks.
My first professional job in 1987 was as a COBOL programmer on a Buroughs B1900 “mainframe” [Box about the size of a *large* refrigerator laying on its side, with a whopping one Mbyte of main memory.] This was the days of “submit your code to compile, come back in 30-45 minutes to see if it was successful”, so I got to hang out in the computer room a lot with the systems guy. They taught me how to take system dumps after a crash (involved turning a dial that had about 15 different positions and reading the registers off a set of 8 little lights), and how to reboot the system, which involved a small built-in cassette tape bootstrapper and a few rotations of that same dial.
The problem with computers today is there aren’t enough knobs, switches, and front-panel binary readouts. Takes all the fun out of.
We only had a first shift computer operation. I don’t recall exactly when, possibly memorial day weekend of 1988, the AC failed in the computer room during the three day weekend, and the temperature alarms that were supposed to signal someone to fix the problem didn’t work. At some point over the weekend it got hot enough that the system crashed, although the power stayed on. The temperature got somewhere upwards of 140 degrees in the computer room by the time people showed up to work on Tuesday. There was an exit door from the computer room that opened out into the stairwell I used to enter the building. When I came in that morning, the door was propped open and it felt like an oven venting as I walked by. If memory serves it was two or three days before we got the system up and stable again.
There was also the time when a capacitor or something in the box smoked itself, and was glowing hot. We were just about to hit it with a fire extinguisher when it went out by itself. A few days later we were having some type of system problem and had to call Unisys support. During the course of the conversation they asked if we had had any recent hardware problems, and the system programmer said “Well, it started on fire a few days ago. Does that count?”
I totally agree. The other thing that struck me as a great step backward was switching from core memory to semiconductor. With a core memory, data was retained after the power went off. All you had to do was to intercept the power-off interrupt, save a couple of registers and then start back where you were when the power came back on. Semiconductor memory lost everything, so you had to start all over again when the power came back up.
I absolutely loved the industrial design style pioneered by DEC. It was so “2001: A Space Odyssey”, and “Space: 1999”.
White. Colors. Charcoal. Clean. Molded plastic bezels. Thick rocker switches in multiple colors.
Check it out. Behold, the mighty 36-bit DEC KA10, 1968:
Oh man, that’s beautiful.
The 12-bit 1969 PDP-12:
The low-end 12-bit DEC PDP-8/e, 1970:
The fonts are really getting good.
The 16-bit DEC PDP-11/20, 1970:
In frog green.
Just awesome.
I like the first one, but “its hard being green”
Don’t forget the DEC VT-100 terminal.
My freshman dorm in 1980 had a “terminal room” with a bunch of VT-52s in it. I always went next door to the Upperclass dorm because their terminal room was half VT-52s and half VT-100s.
I took for granted when I first started in the field that we’d have a variety of computers: VAX/VMS, Apple desktops, Symbolics processors, a Connection Machine or two. Each being used for what it did best. When I retired, there was nothing except Windows and Linux boxes.
Wow! Where did you work? That sounds like a serious range. I do regret the narrowing down of the options. I have moved from Windows (needed for work) to Apple (made it easy to connect to iPods and iPads) to Linux (more in my control).
Yeah!
I worked at Symbolics for over 5 years. It was an amazing experience, I learned so much, and I got to work with some of most brilliant people I’d ever met.
Here’s a much younger me, posing thoughtfully and rubbing my chinny-chin-chin, for the 3640 brochure. Probably ’84.
My modeling career lasted for as long as whatever the shutter speed was set to.
I worked in the aerospace industry for almost 30 years, mostly classified work. When I retired, we were using a grid of Windows boxes to do some coarse-grained parallel computing. The communication layer was Jini and the studies were driven by optimization and tradespace sampling software from Sandia that ran on a Linux box.
I spent quite some time trying to figure out a work problem that I had that Jini would be a solution for. Never managed it. Sounds like interesting work.
I never had time to learn what those using the Symbolics were actually doing. They did seem to have fun, though.
We thrashed a lot of this out over in the old “Coder’s PIT” a couple years ago. I think you might be the one who asked me about it.
In high school, we all fought for the VT-100s, because they were connected to 9600-baud lines, while the VT-52s were stuck at like 300-baud.
In DECwar, someone on a VT-100 could fly into your sector, hit you like eight times, then warp out before you had a chance to type “shields”.
The downside of being on a VT-100 was that if someone opened a message window with you, they could send about a thousand CTRL-Gs to you and disconnect. Then your terminal would BEEP for about five minutes.
That was being pwned in the 80s.
And heaven help you if you tried to play DECwar on a paper terminal.
Well, the VT-52s were pretty early.
I also worked at DEC briefly. ‘Designed a modem board on the VT220 terminal which superseded the VT100.
I always thought the VT220 was adorable:
My favorite terminal was the Ann Arbor Ambassador:
(What? That thing looks like crap. There’s an unsightly ribbon cable connecting the keyboard. And you can’t tell on this photo, but that’s a green tube. Even the VT52 was white.)
The cool thing about the Ann Arbor is that, while pretty much all the other terminals displayed 24 lines of 80 characters, this puppy did 60 lines of 80 characters. That was plenty of room for doing serious programming.
Cramming that many characters required a slower refresh rate, and the green phosphor provided a longer persistence so it didn’t flicker. So that’s the green.
It was also a larger screen; 15-inch vs. the 12-inch VT100 and VT220, and the 10-inch VT52.
But this was the hackers’ terminal.
Thanks for this walk down memory lane. In the early 70s the youngest of my older sisters worked at a bank that had a card-punch-programmed computer which I thought was completely amazing back then. Later I learned to program in BASIC using a machine with a cassette tape drive and the green screens came along a few years after that. Then I did nothing with programming at all for over a decade and found myself writing parts of the U of Illinois’ Second Language Acquisition Teacher Education (SLATE) website using Dreamweaver 4 through 6. It really is amazing how much the art has changed in the last 40 years.
I remember learning MS Word, MS Excel, Filemaker, and MacPaint on a Mac Plus in high school, and ever since then I’ve been able to adapt myself to nearly any office software even if I’ve had no training in that particular software product.
Today, I have to teach my interns how to use Windows nearly from scratch.
My dad had an IBM 360 in his laboratory in the 1960s. (It ended up in the Boston Computing Museum!) When the lab had an open house for families, he figured no one would understand the work he did, so he set about to program the computer to play tic-tac-toe against visitors using its pen plotter. I got to help in the programming (FORTRAN!). I was about 11, and it hooked me!
In high school, I got bored and rowdy in Algebra II because it was mostly a review of Algebra I. The school had just gotten a teletype machine with a 110 baud (not Kbaud!) acoustic modem to a PDP-8 at the state university. My algebra teacher basically kicked me out of class and told me to play on the computer. This is the same setup Bill Gates learned to program on.
An Explorer Scouts group let us have evening access to a corporate mainframe downtown in the evenings. For kids in the mid-70s to have access to this kind of computing power was amazing! One fun remembrance from that program: we had a 7th grade science teacher who on the first day of class each year would tell the class that if anyone could turn in all the numbers from 1 to 1 million by the end of the first marking period, they would get an A no matter what else they had done. Of course, the whole point was to give kids a sense of how big a million was.
One of our group had a little brother in the class that year, so we asked the program advisor if we could use the mainframe to print out all the numbers. He agreed, so we printed it out one evening. It ended up as about an 8-inch stack of that large fanfold paper. I wish I could have seen the teacher’s face!
My freshman year at MIT, I took a computer graphics course. The work had to be done on the ME department mainframe, which was fundamentally a batch processing machine programmed with Hollerith cards, but it did have an interactive mode.
My class partner and I decided to exploit the interactive mode to create a driving simulator on the computer’s vector display. It was very primitive by today’s standards — a sky view of a stick figure car with a graphics tablet control (up to accelerate, down to brake, left and right to steer). But this was a time when Pong was still considered state-of-the-art.
The problem was that the interactive mode shut down all other use of the computer, so whenever we were running the simulator, no one else could be using the computer at all. So all the PhD students doing their computational fluid dynamics and finite element analysis were stymied by two kids playing a video game.
Luckily, they all thought what we were doing was neat. We kept them happy by letting them “drive the car” and provide us with suggestions for improvements. The only trouble was, we almost never got to use it ourselves.
My High School computer science teacher circa 1970, “No computer will ever beat a human at chess”.
During a computer science class in college in 1975 I had to write a program that had a branching search subroutine. I’d take my punch cards down to the computer center and turn them in and wait for it to run. Our programs ran on the university’s IBM mainframe, and we were allocated a huge block of 10 seconds of CPU time ( an eternity) for it to run. I knew every time my program ran as every monitor and printer in the entire lab froze for 10 seconds before resuming.
I never did manage to figure out where my program was broken.
@kozak – One of my first programs in college was to print a “3D” plot of a mathematical function. That is, each point had a density that was proportional to the Z value. I intended to do this by over printing each line several times. Unfortunately, I got the wrong code for “print this line, but don’t advance” and used the code for “print this and eject the page”. –
This was on a high speed printer, so the next day, when I went to pick up the output, I got about a 6″ stack of printer paper with one character on each page along with a very nasty note from the computer operator. It seems that the paper just flew out of the printer across the room.