Chapter 20

SANDY - July 1961 Since there was no reason not to, and nothing else useful that I could do, I joined the team that was assigned to the cabling work that would hook the Reaper box up to the Stretch computer. I was in a foul mood most of the time, thinking of where I had hoped to be over the long weekend and the following week, but managed to learn more than I ever wanted to about cabling. I checked through everything in painstaking detail, and found several errors in the rework instructions.

I guess that because of my frame of mind, I was short-tempered about the errors, and insisted that they be quickly corrected before the process could continue. While the corrections were made, I'm sure that I didn't win any friends with my attitude.

Finally, the test began. It started out well, and we were able to find and write up several problems with the interconnection logic between Reaper and Stretch. Fortunately, we had a maintenance console for Reaper, which allowed us to see the contents of the critical registers, so we didn't have to probe and scope too much. We got to a point where the Reaper startup information was being transmitted and registered perfectly. However, the Reaper program wasn't sequencing, and we couldn't figure out why for a couple of days. Finally, Aaron came up with the answer. He discovered that several logic cards in the control circuitry used the same transistor type that had been so troublesome on Stretch. When the cards were tested, several of the transistors failed the test.

I blew my stack! We had enough to do without having to revisit problems that should have been fixed. I picked up the phone and called Paul, telling him that something needed to be done fast to resolve this issue. He said, "Sandy, settle down! I'll take care of this. Right now, the best thing to do is to test every one of the cards that use that transistor."

I was still fuming when I hung up, and we spent the rest of the day testing the cards and replacing transistors on about a third of them.

It turned out that there was also a serious logic error in the control circuitry, and it took another two days for the designers to come up with an Engineering Change (EC) that they claimed would fix the problem. I was not happy when we discovered major errors in the EC rework instructions; it took another day and a half to straighten that out. I called Paul and asked him to set up a meeting with the design manager.

By the time we met, I had calmed down somewhat, but was still quite blunt about my displeasure with the quality of the rework instructions, and insisted that they be checked more thoroughly in the future. Fortunately, Paul backed me up on this.

After this, things picked up, and we started to make some real progress, although I was still concerned that we were identifying and documenting problems faster than we were getting the fixes. Jody was a godsend with the paperwork! Every morning when I came in, I first met with Miguel and reviewed what had happened on his shift. By then, Jody had the daily status report on my desk. She had set up a simple tracking scheme that showed a concise picture of the status of each problem. The status categories were: new; assigned to engineering for a fix; fix available but not yet applied; or fix applied and verified so the problem could be closed. In many cases, we were able to come up with a solution to the problem ourselves. For these, we applied and verified "temp fixes" (temporary fixes), which we included as suggestions in the paperwork that we sent engineering. I found that I was working at least nine, and usually ten to eleven hours a day. The pressure was beginning to wear on me, although every time another one of Sarah's diagnostic programs ran without error, I was quite elated. I just wished that I could have been able to share my triumphs and frustrations with Steve. I was really missing our phone calls, and not even knowing when we would be together again made it even harder.

STEVE - July 1961 - Somewhere in Germany

Jake Goldstein was the Intelligence Officer who had been assigned to work with us. Fortunately, he was fluent in German and even had a fair knowledge of Russian. He was easygoing, but didn't seem to know much about ciphers and codes. I asked him, "What can you and Stan tell me about the traffic you've studied so far?"

Jake answered, "Well, we have some fixes on the various points where messages originate. Based on the volume, we've identified one site in particular that we think is the command center."

I said, "OK, then, let's look at that traffic first."

Sam, Stan, and I spent the rest of the day looking at messages. By five o'clock, I had thought of a possible plan of attack. It turned out that Stan had been programming the 7070 for a while, and we started coding an analysis program. He was a demon programmer, and I learned a lot of useful tricks from him. The next day I punched the program on cards, and Stan and I assembled and debugged it. When we ran the program on the real messages, it showed me some interesting patterns, and some sequences that repeated. We sat down with Jake again. I asked him if he had any ideas as to the contents of the messages. He was able to do better than that; he wouldn't tell me how they got it, but he did let us look at the German cleartext, along with a translation, for one of the messages; Sam told us that this was called this a "crib." I should have realized that the German military mindset was very formal and authoritarian, since the message started off with what appeared to be a standard header, and the phrasing of the message body was interesting, to say the least!

I looked at several encoded messages from the same source, and started to see a clearer pattern, at least at the beginning of the message, presumably a header section. I went back to the program and made some changes to have it look for new variations in the header area. After two more days, I found the first clue to how the messages were encoded, and, by the next week, the program was able to decode the message headers. Jake and the Colonel were elated about this, since the headers identified the recipients. They told me that alone was extremely valuable information.

We continued to make slow but steady progress. One day Jake brought us another cleartext message, and this led the way to finally being able to recreate individual words in the messages. I wrote another program to decode messages using the correspondences we had identified, and we were finally able to read message fragments. When Jake saw this, he started giving me some ideas for the words we didn't have yet. When we used some that worked, after another two weeks, we were able to read about 75% of the text of some of the messages. Everyone was quite satisfied with what I had done so far, but apparently it wasn't enough to get me sent back to the States yet.

I sat down with Sam one day, and told him about Sandy and our relationship, and how devastated she was when I had to tell her about my sudden trip. I pleaded with him to let me go back home, but he said, "Steve, they still need more. You've done wonders with your program, but let's come at the problem from the other side. We need to apply everything we know about ciphers and cryptography to these messages."

I was torn. Part of me wanted to get back and re-establish contact with Sandy in the worst possible way. However, the problem intrigued me, and I reluctantly agreed that I wouldn't raise the issue again for a while, at least until we had either cracked the code completely, or knew that it couldn't be done.

Sam and I spent many days looking at the output of my analysis program for all the messages we had by now. I was falling into a routine, where I'd rise in the morning, have a quick breakfast, and then walk over to the communications center and the office that I shared with Sam. Usually, Jake and Stan joined us, and we went over the traffic that had arrived in the past 24 hours, and that had been run through my programs. While we were still picking up additional words from time to time, it wasn't enough.

When the others left us, Sam and I would look for useful patterns in the cipher. Usually, we gave up around five o'clock, and headed for dinner, but one day, Sam suddenly got excited, saying, "Look, Steve! It's right here, in front of us! Look at these consecutive characters, and now look at the bit permutations that might be applied from one character to the next."