Talk:CASINO

Medidata's previous name

According to Special Follow-up Feature: Nuclear Family:

Scientific Engineering Institute. The SEI was .. eventually rebranded as Searle Medidata Inc., a subsidiary of G.D. Searle.

So if the Bell quotation from Computer Engineering:

DEC also never built a PDP-3, although one was designed on paper as a 36-bit machine. [...] In 1960 a customer (Scientific Engineering Institute, Waltham, Massachusetts) built a PDP-3

is correct, this seems to be the machine he was talking about. Which fits with things in the letter from Rawson to Lane, which speaks of:

the basic 4K memory. This memory consists of two PDP-1 memories placed on top of one another to provide 38-bit words

At first I was thinking 'if the PDP-1 was 18 bits plus a parity bit, making its memories 19 bits wide, that would check - if CASINO was 36 bits, plus a parity bit - or a parity bit for each half-word, if the machine was prepared to do half-word writes (the way PDP-11 memories have separate parity bits for each half-word, because they can do byte writes)'. If the PDP-1 had parity, that would all check, and explain why the paired PDP-1 memories were 38 bits wide - but I decided I should check to make sure the PDP-1 had parity. It didn't! But, in the PDP-1 Maintenance Manual (pg. 8-7, 214 of the PDF), I found:

The core bark actually includes an extra core plane which is completely wired in. .. The extra 19th plane is not ordinarily used, but is provided in case it is wanted for some special application

So either CASINO had 36 bit words, with parity (or it had no parity :), or 38 bit words... But still, I'm fairly certain this was the 'PDP-3'. I will leave the PDP-2 and PDP-3 articles as they are, for the moment, but if you think I have correctly identified it, I will fix them. Jnc (talk) 01:48, 8 January 2024 (CET)

Very interesting theory! I'd like to mull over this for a while before announcing my edict. And pass around the word. Larsbrinkhoff (talk) 09:27, 8 January 2024 (CET)

Sure; no rush.

In looking for info about Scientific Engineering Institute, I found some other sources which indicated that it became Medidata (with Homer Oldfield, he of the GE ERMA project, involved). I decided this morning that I wanted to find better sources on that connection, because a lot of the ones online are kind of conspiracy theorists (and sound like they are all quoting each other, which does not increase the weight of actual evidence).

I did find a public record (bottom right corner on pg. 21), which gave SEI's address (140 Fourth Avenue), and that is indeed the same address as the one printed in the Medidata letterhead of the letter from Rawson, so I think it's now pretty well nailed (as the English-language expression goes) that they were the same.

And that means, if Bell was right about SEI in his book, that CASINO is the machine it spoke of. (The 'other' PDP-3 he speaks of, in his oral history, is not an actual machine, but just an order for one.) Jnc (talk) 12:22, 8 January 2024 (CET)

Having slept and mulled and put out the word, I can't raise any strong objections. It seems likely that building a 36 and/or 38-bit computer would be a major effort and cost, so SEI doing it twice over in the same time period seems improbable. Larsbrinkhoff (talk) 08:14, 10 January 2024 (CET)

The 19th core plane is an interesting find by itself! No PDP-1 aficionado I mentioned it to was aware of it. Larsbrinkhoff (talk) 08:14, 10 January 2024 (CET)

One wonders what the extra bit could/would be used for, if not parity! But the earliest DEC computer with parity (that I could find in a quick search) was the PDP-7, on which it was an option. Jnc (talk) 15:52, 10 January 2024 (CET)

Original use

A lot of Web-sites seem to say something like this: The only PDP-3 was built by the CIA’s Scientific Engineering Institute (SEI) in Waltham, MA to process radar cross section data for the Lockheed A-12 reconnaissance aircraft in 1960., but I haven't yet tracked down a reference to the A-12 usage, other than this. Jnc (talk) 13:15, 8 January 2024 (CET)

So I looked around a bit, and found a book, The Archangels, by an amazing guy, Thornton Barnes (available for download here - grab it while it's available!), which confirms the SEI was doing radar low-observable work:

the anti-radar research of Dr. Edward M. Purcell of Harvard University, who had discovered a possible means of countering or absorbing radar emanations. His discovery, led to laboratory work in techniques to blanket portions of the A-12 with radar absorptive materials to reduce radar detection. .. Project RAINBOW laboratory’s research and testing occurred under the auspices of a CIA proprietary research organization named as the Scientific Engineering Institute

This was long before Overholser at Lockheed discovered Ufimtsev's paper, and they started that low-observable program; but the A-12 had been using low-observable techniques long before (notably in radar-absorbing composite materials, and for the chine design) - which is what SEI was involved in.

This doesn't tell us exactly what CASINO was being used for - running theoretical models (as with Overholser's later work), or processing recorded experimental data (as suggested by the query above) - but it likely was some sort of low-observable work. Jnc (talk) 18:43, 9 January 2024 (CET)

Rawson seems particularly keen on pointing out CASINO's good graphics capabilities. Maybe they were put to good use in medical applications, but still, maybe that wasn't the first application. Larsbrinkhoff (talk) 08:14, 10 January 2024 (CET)

It almost definitely wasn't. I found some bio-medical papers from SEI, back in the day, so I can see how SEI get into that field, commercially, back when they apparently stopped doing engineering work for the CIA, and became e real business, Medidata. And the use CASINO was being put to later, just before it was retired ("the board testing function now being carried out on CASINO") definitely was a later re-use of a machine they had laying around.

Ben Rich's Skunk Works mentions Dr. Purcell's work on reducing the U-2's radar cross-section (pg. 152), so that Thornton Barnes guy does know what he's talking about. It's a pity there's not more available on exactly what they were using CASINO for, originally, but although the RCS-reduction work of that era is now completely antediluvian, and it would be fine to talk about now, back then it was completely hush-hush, so it was obscured. The original report speaks of "radar cross section data for the Lockheed A-12 reconnaissance aircraft in 1960", so maybe if I poke around in material on the early A-12 RCS-reduction work (which was a follow-on to previous, similar work on the U-2), I'll turn something up. SEI probably had either data or theoretical models that needed a high level of accuracy, so they needed a 36-bit machine. Jnc (talk) 11:28, 10 January 2024 (CET)

I got ahold of Radar Man, by Edward Lovick, who was a main expert on RCS reduction for the Skunk Works, and it is quite informative - a great source on early RCS work. It does mention SEI's work, but alas, no mention of CASINO. It does mention mathematical RCS models, and the correlation of model predictions with experimental results, so CASINO could have been involved in either. From the descriptions of experimental work in the book, they seem too 'agricultural' (as the English expression goes) for CASINO to have been involved. (At one point he describes a person being lifted by a forklift to read meters in a model at the top of a pole, under test!) So, I rather suspect CASINO was more used for mathematical models than for any involvement in recording experimental data. (This is particularly true as it apparently physically stayed at SEI, and sending the data cross-country in real-time in 1960 would have been fairly difficult.) The book has a lot of technical detail (in enough depth to make my eyes 'glaze over', as the English expression goes), so it's a much more reliable source on early RCS reduction work than Rich's book. Lovick was 80 when the book came out in 2010, so he's likely gone now, so unless some contemporary documents are still hiding in a secret archive, I'm not sure we'll ever get more about what exactly CASINO was doing. Jnc (talk) 15:04, 24 January 2024 (CET)

Try this search: https://google.com/search?q=%22lovick%22+%22pdp-3%22 It got me this: As the amount of radar testing at the Ranch increased, the scientists at SEI decided that the data had become too much to process by hand. To speed things up, they decided that a computer would be needed. The difficulty was that no computer within their budget was available off the shelf. So they decided to build their own. In October 1960, the Digital Equipment Corporation (DEC) had produced a specification for the PDP-3, a new system in their line of programmable data processor minicomputers. Ed Rawson took charge of the project and with the help of Chuck Corderman and Jay Lawson designed and built a PDP-3 using standard DEC logic modules. [EG&G personnel sometimes teased Rawson that the SEI folks must have held stock in DEC. (Pendleton, Wayne E., e-mail to author, 14 Feb. 2007)] Because disk drives were not available, a tape loop running through an Ampex tape drive held intermediate results; eventually, the tape loop was replaced with a drum memory. The project was run like a homebrew computer project, with more emphasis on getting the machine and software to run rather than on making it well documented and easy to use. The design evolved so rapidly that when one of the engineers returned after a two-week absence, he didn’t recognize it (Interview with Daniel Schwarzkopf, Stow, MA, 30 Nov. 2003). The design evolved away from the original PDP-3 architecture, and it came to be called CASINO, for computer able to select internal orders. Eventually the system worked. Radar data were recorded by EG&G at the Ranch on 1-in.-wide data tapes and shipped to SEI in Waltham, Massachusetts. The data could be processed correctly, but the computer could usually only be operated with Rawson looking over the user’s shoulder. Eventually the PDP-3 was discarded; one computer engineering textbook stated that in the early 1980s it was running somewhere in Washington state, but the author of that book could not confirm it (Bell, Gordon, e-mail to author, 13 Feb. 2007). There is an unconfirmed report that it was donated to a Boy Scout troop and eventually given to Dow Chemical for disposal. It was the only example of a PDP-3 ever built. Larsbrinkhoff (talk) 14:44, 25 January 2024‎

Gnashing of teeth! That book was on my list to get, I just hadn't gotten to it yet! I didn't check to see if it was available online! (It's amazing how many books are available on the Internet these days!)

There are some useful details there; e.g. the confirmation that it was based on the PDP-3 (but perhaps only the ISP description given in the PDP-3 specification document, not an actual completed gate-level design ("a specification for the PDP-3 ... designed and built a PDP-3"). Although they may have started with a set of PDP-1 drawings, and modified them as needed.

That text doesn't address why they needed a 36-bit machine, why an 18-bit one wouldn't do; I doubt their A/D produced more than 18 bits. I'll bet they were doing some modelling too (note the use of "a tape loop [to] h[o]ld intermediate results"). Even if their budget wouldn't run to buying a PDP-1 - why not just buy the modules and build a PDP-1 from the drawings? They must have needed the extra accuracy.

I wonder if the material he cites that he got from participants is available? (Several books of recent history that I've read recently - e.g. "Wizards Stay Up Late" - say something to the effect of 'all interview transcripts have been deposited in the XYZ Library'. He gives a URL, but it 404's, and the Internet Archive doesn't have it - the only crawl is from 2022 - probably after the text was put online at dokumen.pub.) The participants are probably no longer available, alas.

Well, I will update the entries, here, and at PDP-3, to include this info. Jnc (talk) 13:18, 27 January 2024 (CET)

There are two zip files here: https://arc.aiaa.org/doi/suppl/10.2514/4.867316 But it seems they are only declassified documents. I started going through the documents in the first zip file, quickly skipping over everything before 1960. But so far nothing of interest. Larsbrinkhoff (talk) 14:40, 27 January 2024 (CET)

Yeah, I didn't see any interview transcripts, either. Under "ACKNOWLEDGMENTS", it does say "my wife ... transcrib[ed] interviews", so there were (at one point) transcriptions, but I guess he didn't include them in his 'supporting materials' collection. Oh well! I wonder if it's worth trying to contact the writer? He's in LinkedIn here; I'll send him a message. Jnc (talk) 16:59, 27 January 2024 (CET)

I already sent an email to Suhler. He sent back some additional information, but nothing substantial. I asked about interviews, but he didn't offer any transcripts. Larsbrinkhoff (talk) 18:22, 27 January 2024 (CET)