Date: 5 May 2007
From: <name withheldt>
Subject: Re: scratch monkey

The dead monkeys are real, but the story has been garbled. How I remember this — and it’s been over 20 years now — is like this:

I was working for the <institution> Department of Zoology. We didn’t do any experiments with monkeys. You need special permits for that in <country>, which are very hard to get. We didn’t have any. We did, however, have permits to allow experimentation on rats, including neurological experiments on the brains of rats. Those rats had sensors — electrodes — directly implanted in their brains.

This is all very crude and ugly work. Whoever coined the expression ‘this isn’t brain surgery’ has a much higher opinion of the skill required for brain surgery than I have based on what I learned at this point of time.

We read data from the electrodes into a CAMAC crate and there was a complicated way to get the data from the CAMAC crate into a pdp 11/05. From there we could get the data into the 11/44 that was our timesharing system at that time. Note that the computer problem we were solving only involved reading data from the electrodes. Nobody wanted to control the electrodes so that they could administer shocks via the computer.When the electrode manufacturer came out with a new version of the equipment, it looked as if we could do away with the CAMAC crate, which would have been a real blessing. I wrote a driver for the thing, and this was based on a v6 unix disk driver. I can no longer rememebr why I thought that a disk driver was a good model for an electrode sensor driver, though I remember that the sensors had a block mode and a character mode. For 18 year old me, that may have been enough. It could also have been curiosity — I had written 2 tape drivers and a printer driver, and had hacked on a driver for DH-11s, and may have wanted to do something else.

The driver worked well enough, but for the defect that it was un-interruptable. The electrode equipment had no buffering. When it wanted to write something, you really had to give it the whole unibus and lock out everything else including the clock until it was done. This was unacceptable for a time sharing system, and that was where the problem stood. Naturally, we never tested it on any animals.

At this point the professor who was doing the rat experiments decided to take his research someplace else. We all breathed a sigh of relief and forgot all about him. In particular, I did not know that he had taken a copy of my driver with him when he left.

Skip forward a bit. Somebody, I forget where, who was doing primate brain research, with emplanted electrodes in monkey brains got a copy of my driver and modified it to work with a dedicated small Unix machine they had. The electrode manufacturer had released an even newer model, but as far as I know, even that model was not designed to deliver electic shocks on computer command. It did, however, have a self test mode and in running self tests it could deliver shocks.

And one day, a DEC service repairman really did try to run a diagnostic which involved copying data from a test diskpac into what he thought was a different drive which he had loaded with a scratch pack. But he got the volumes wrong — because they were mislabelled at the computer centre — and ended up trying to copy his diagnostic into the electrode driver, which was connected to some monkeys. This ought not to have done anything, but for some reason it did trigger the electrodes self-test mode, which was the cause of the catastrophe.

Some of the monkeys received shocks and at least one of them died. As I recall some of the electrode equipment was made unusable as well, which means it may have been the heat rather than the electricity which was fatal. To make matters worse, the tests were being done at a time when the researchers and the animal technicians who most closely worked with the monkeys were not around. And the monkeys, of course, weren’t in the machine room, but on a different floor of the building. Thus the fact that the monkeys were in distress was not noticed until people came to work the next day. At which point, and rightly so, all hell broke loose and everybody involved in the project was in for a lot of blame, including me because my name was in the comments for the driver.

I’m still pretty embarassed about the whole thing.

<name withheld>

So there you have it.

I hadn’t even heard of the SAA, or I’d have joined many years ago, as I’m very interested in archiving computer history.

The next step will be to try to get the programming manual for the AP-101 computer, and perhaps the source and object code for the HAL/S compiler, which it might be possible to run on the Hercules mainframe simulator.

On April 21, 2006, I submitted this FOIA request to NASA:

I: HAL/S Language Specification

This is a manual providing a specification for the HAL/S programming language, which is the language used for the software for the Space Shuttle on-board computers.

This document was produced by Intermetrics under contract to NASA; I believe that is is either in the public domain, or that the United States Government owns the copyright.

Multiple editions of this specification exist. I am aware of editions with dates ranging from 1974 through 1980, though I expect that there are probably more recent editions. The most recent edition available would be preferred.

II: HAL/S Programmer’s Guide – Volume 1 & 2

This document was produced by Intermetrics under contract to NASA; I believe that is is either in the public domain, or that the United States Government owns the copyright.

III: HAL/S Programmer’s Guide

There was an edition published in December 1981; there may be a more recent edition. The most recent edition available would be preferred.

This document was produced by Intermetrics under contract to NASA; I believe that is is either in the public domain, or that the United States Government owns the copyright.

IV: HAL/S-360 Compiler System Specification

There was an edition published in Feburary 1981; there may be a more recent edition. The most recent edition available would be preferred.

It is believed that these documents would be available at the Johnson Space Center, and would normally be used by the team responsible for the development and testing of software for the Space Shuttle’s on-board computers.

I received confirmation of the FOIA request on April 21. They should have sent me a response or a notice of a deadline extension by May 19, but I have not yet received either. It’s possible that the response or notice is in the mail to me. I’ll give it another week, and if I haven’t received anything, I’ll contact them again.

The TDC worked much better than the systems in use by our foes, at least partially because it tracked the target in real time, taking into account the submarine’s velocity, while the foes’ systems apparently did not.

Also of interest, and visible on the public tour, is the ECM Mark II cipher machine. While this was a rotor-based machine that in principle this machine performed the same functions as the famous German Enigma cipher machine, it was much more sophisticated. Instead of advancing the rotors one position at a time like an odometer, it had a second set of rotors that controlled the advances of each of the main rotors.

It occurs to me that this dual-rotor system is not unlike one of the methods used in more recent times to generate pseudo-random numbers. LFSRs (Linear Feedback Shift Registers) can be used as PRNGs, but have many undesirable properties. Sometimes one LFSR is used to control the step rate of another LFSR in order to improve the “randomness” of the results.

Many people on the tour were heard remarking about how incommodious the submarine is. Having heard for many years that they are quite cramped, I actually thought the real thing was not as bad as I’d imagined. But I still wouldn’t want to serve on one for months at a time.

The System/360 was the largest single corporate product development effort ever, at a cost of five billion dollars (back in the early 1960s when that was real money. To put that in perspective, it has been said that the cost of each Apollo mission to the moon was approximately one billion dollars. It was a bet-the-company gamble, and it paid off handsomely.

Dr. Brooks originally coined the term “computer architecture”. He and Gerrit Blaauw formalized the distinction between computer architecture (what the programmer sees), implementation (the logic design), and realization (the physical implementation of the hardware as tubes, transistors, ICs, or the like).

Dr. Brooks worked on the IBM Stretch project, then managed the System/360 project. He is justifiably famous for his book The Mythical Man-Month, one of the first books on software engineering, in which he relates many of the lessons learned during the development of OS/360.

One of the interesting observations Dr. Brooks made was that JCL is the world’s worst computer language. Its branching support is very primitive, it has no real facility for looping or subroutines (though apparently you can kludge them), the syntax was intended to be similar to assembly language even though most users were writing programs in HLLs, and there were be design only six verbs, so all manner of other verbish things are done by modifiers to declarations.

He said that JCL was originally only expected that a job would have a few control cards tacked on the front, but that it grew way beyond the expectations. It was not intended as a general purpose scheduling-time programming language, like today’s shells and scripting languages.

In college I took a COBOL programming class. We had an HP-3000 system which had a COBOL compiler, but they didn’t let us use it. Instead we had to submit jobs via RJE-3000 to an IBM 3081 at the state University and wait for the results. In theory this was because COBOL skills on IBM equipment were supposed to be more useful. But in practice, we didn’t really learn anything specific to the IBM because they just gave us the JCL statements we needed without teaching us what they meant. As far as I can tell, that’s actually how most programmers deal with JCL; they get a few JCL statements that have been known to work with other programs, and touch them as little as possible.

Someone printed up replica Green Cards for the attendees. I wasn’t able to find my copies of Dr. Brooks’ books in time to take them to be autographed, but I did get him, Bob Evans, and Gene Amdahl to autograph a Green Card for me.

I expect the museum to offer video tapes of the talks, but of course they are not available just yet.

An excellent printed account of the System/360 development, including much of what Brooks and Evans discussed, may be found in the book “IBM’s 360 and Early 370 Systems” by Pugh, Johnson, and Palmer, published by MIT Press. Truly a must-read for anyone interested in the history of computers. (For similar information on IBM’s history previous to the System/360, get “IBM’s Early Computers” by Bashe, Johnson, Pugh, and Palmer, unfortunately out of print but well worth buying a used copy.)