Quotes from "Oral history of Alan Kay" - PDP-10/maxc GitHub Wiki

Kay: [...] it's a great story, that we tried to buy a PDP-10 in the first few months of PARC and Xerox was selling Sigma… what used to be Sigma… yeah, Sigma. They used to SDS but Xerox had bought them and so they didn’t want DEC out there taking pictures of these black PDP-10 boxes going into the Xerox then using them in ads because they were already contending. So Xerox said no and so PARC almost broke up at that point.

Spicer: Really?

Kay: Yeah, because our thought was, and there was probably only 12, 15 people at that point. But the thought was, "Well, it isn’t up to them. We need a PDP-10 because that's what the ARPA… is on the ARPA net and we're part of that community and blah, blah." And so in these meetings that were held, it turned out that the reason Xerox had bought SDS at all was because of Butler and Project Genie at Berkeley.

Spicer: Oh yes, timesharing.

Kay: Yeah, and so SDS was not responsible for the SDS-940. It was actually done at Berkeley and Taylor forced SDS against their will to make 940's. So, the first round of timesharing computers in the ARPA community were those things. So, this is an irony. And because of that, Xerox, without really looking too clearly into where all this technology came from, bought SDS.

Spicer: Right, at quite a premium too.

Kay: Yeah, and then there it was when we showed up at PARC. And so we had one of the best operating systems designers in the whole world there and so part of the meetings were something like, well, it takes about three years to do an operating system and I don’t think Butler wanted to do another one right then anyway. But then the idea came up: but we could build a PDP-10 in less than a year. I don’t know who uttered that thought first.

Spicer: Yeah, it seems so implausible almost.

Kay: Everybody started… well, not really because we had Thacker and the BCC-500, which was the project that maybe 10 of these guys had come from, had done a micro-coded timeshared computer. So the idea is, you know, we'll emulate a PDP-10. We won't make… and by the way, we'll test out the first integrated circuit memories. We won't use cores, and that led to a whole bunch of interesting interactions with Intel about the 1103 [memory] chips. But to get to the point of this story, and there are many wonderful stories about this machine, but the cool one was that Chuck was going to do the memory and the memory architecture and the console and everything else and he just really didn’t think he had time to do the CPU. And Butler said, "Well, if I don’t have to use an oscilloscope, I'll design the CPU." And Butler never had done anything like this before. But, of course, he knew…

Spicer: Because it's too much hardware for him?

Kay: No, he didn’t want to do anything analog.

Spicer: Oh, I see.

Kay: He was a very confident guy and with good reason.

Spicer: That it would work from paper?

Kay: Yeah, because Butler was very confident. And so his idea was, you know, what's basically sort of what's the big deal, it's just a bunch of logic gates. And so Thacker set up the design rules for Butler so he never had to use an oscilloscope. But it was complicated. It was 14-phase clock. And by God, Butler designed the CPU for this thing and it worked.

Spicer: Wow, yeah, he's in a league of [his own]…

Kay: That was one of the most amazing things I've ever seen because the learning curve he… and the whole MAXC was done within that first year.

Source: https://archive.computerhistory.org/resources/access/text/2016/08/102658340-05-01-acc.pdf