One thing this photo makes clear is how low-powered the #LINC was. You can clearly see at the bottom that it's fed off of one 15A/115V 60Hz US wall outlet. The power goes to the scope (which was very dim, if the DigiBarn videos are any guide) and the #LINCTape drives, and then big cable bundles come off the back to carry power and logic levels to the tall cabinet that held the CPU and #core memory.

This makes sense, as it was designed to be one piece of kit in a lab full of other equipment, all running off of ordinary US mains power. If they'd required 220V or three-phase (as many #dec machines did), that would have ruled the device out for many institutions.

But this is how it was possible to run a LINC in a residential home that had pre-war electrical setups. You could theoretically have powered this off a light socket.

Here's the 1964 sales brochure from #dec (who ended up building the things and selling them for a while).

I may have got the chronology of the photocells a bit wrong here, and they may have been there from early on. The story Wes Clark tells is that the things used mirrors as part of the light path for the photosensors, and they were looking for a positive signal (light makes it past the tape, indicating that it's off or misaligned).

And apparently one of the people at Lincoln Labs was a heavy smoker, and he would rest his arm in such a way that the ash from his cigarette would fall on one of these mirrors. So when the tape slipped its moorings, the dirt on the mirror prevented the photosensors from picking up anything!

Wes Clark also tells stories (midway through the 1986 talk I posted, I think) about demoing the resilience of the #LINCTape systems to funders by tipping an ashtray over the equipment, brushing it off gently, and then loading up a tape. Apparently it got some complaints, but it proved the point that they were making something durable enough for the end user to keep their data on their person and load it up as needed.

I highly recommend watching the video above. Wilkes wrote an OS for the scope that provided a full-screen text editor at a time when most luxury OSes only had teletype-oriented line editors at best. The interviewers are kind of pushy, and seem to be getting her to give her talk to the camera when she wants to ensure they get certain documents, but she's game and gives a lot of good history about the machine's operational lifetime.

In particular, she worked first on a LINC simulator on the TX-2, but they kept changing the architecture so that's partly why they had to ship her a LINC in Baltimore while they all set up shop in St. Louis.

Note that she also stressed the behaviour of system sounds as an indication of smooth operation, much like the technicians who preferred Hamilton's code for SAGE.

Anyway, LINC showed that you could build a small inexpensive computer from #dec flipchips. So what did Digital themselves do?

Well, they tried to make their own, and that was the #PDP5. This ultimately became the #pdp8
But they did something clever: they made a change to the #LINCTape format that let it read data backwards as well as forwards.

And they called it #DECTape. In one of the videos I pasted above, Wes Clark mentions that he specifically asked that they take all LINC branding off of this, because he didn't want to get called doing support requests for Ken's customers!

OK, so here's the earliest known photo of a working LINC, showing the OS software author, Mary Allen Wilkes, sitting at the console in her father's home. The cabinet on the right side of the photo is the main set of flipchips and the core memory assembly (and power supply), and you can see the console, display, and dual #LINCTape drives next to her.

She moved away from Massachusetts, but they begged her to write the software for the thing, so she just had them send her the computer. This is widely cited as the absolutely very first "home computer" in the sense we know it today (Konrad Zuse had his Z1 in his apartment, but that was such an esoteric device compared to this that I think the claim is still valid to our modern sensibilities).

OK, so that's great, but what's this LINC?

Here's a talk by Wes Clark on the TX-0, TX-2, but mostly about LINC: https://youtu.be/l9YBZo30Ses?list=TLPQMDYwMTIwMjNntcAHDint6g

Everyone at Lincoln Labs built this amazing computer in the TX-2, but they found a LOT of people kept wanting to use the TX-0. One of those people was Ken Olsen, who left to found #dec, and you can see how my theories about the history of the PDP machines fits. Olsen wanted to "churn out computers like television sets" (which DEC eventually did with the PDP8), but at first the board forced him to make lab bench equipment.

From there, DEC made modular electronics boards called "flip chips" (no relation to the term we use today: that's a theme with early DEC stuff. Ask me about "microcoded" instructions in the PDP-8 instruction set...). It was pretty clear this was similar to the modules the TX-0 and TX-2 used, and the goal was a computer, and in 1959 the PDP-1 was made with these modules.

But Wes had moved on a little, and thought about the TX-0 and how some folks preferred it. MIT had purchasing rules for new computers that required a board of bureaucrats to approve any new computer purchase. But he noticed that you could buy anything you liked so long as it cost less than the total amortised cost of one full salary for office admin staff. So he wondered how much computer he could build for this magic number. What could sneak under the RADAR?

So using lessons from the TX-0 (people will work around a limited instruction set) and the TX-2 (fast tape drives will hide a lot of limitations in a computer), he designed the #LINC as a laboratory instrumentation automation and data-collection computer.

This was a system with a 12-bit architecture, a two-bit opcode with 10-bit operand (yes: 4 instructions, but *so* many side-effects!), a whole bunch of analogue integration ports (to plug into your lab gear), a keyboard driving the CRT-based console, and 1kW of core memory taking up most of the equipment. All built with DEC flip-chips.

And yes, that tape drive was what we now know as #LINCTape. It's a scaled-down version of the TX-2 tape drive, with reels you can fit in an overcoat pocket.

The terminology for this machine has everything pushed down a level from what we expect. What the docs refer to as "registers" are nearly always locations in core (the accumulator is always called just that). And often operations we think of as happening in main memory are pushed to the tape. After all, it was fast!

While @tastytronic is setting up the new more powerful #dec #flipchip tester (in its snazzy #pdp12 livery!), I went to re-watch the videos about the tester that the #umnpdp12 project uploaded to YouTube. The last one ended with a classic fix (CLEAN THE CONTACTS, LADS) while diagnosing the LINCTape, and they didn't even test the cards before making progress: https://youtu.be/aiHSRF_2bmk

I did a bit of digging into #LINCTape and #DECTape over the past couple years, and so here's a #VintageComputing thread with some of the backstory on these tape formats.

Spoiler: The answer is *always* MIT Lincoln Labs.