Departures: Among the Yankees: Part 6

Summary: The Mark I... Free concert... Moshing for codgers... Notice the word "CANS" written on the bottom of the box... Making sure the trains go the right way... I got all excited about pre-transistor logic... Graph theory... Light at the end of the tunnel... Things left undone... At loose ends in the park...

Turn the Other Cheek

1999.05.01 SUN Cambridge (Harvard)

I'd had a great time wandering the halls of MIT. I hadn't been so impressed by Harvard, but it occurred to me that I'd given MIT more of a chance--I'd wandered its halls, beheld its activity. Maybe I should give Harvard a similar chance.

Once again, I entered Harvard Yard. This time, I was looking to see how people were getting into buildings. I'd look for a building that was obviously unlocked, and then I'd enter it. I'd look for a busy building.

I sat on the grass, pulled out a book, and did some surreptitious observation. Not a lot of people seemed to be entering or exiting buildings. At least some of these buildings seemed to be residences. Most people seemed to be just passing through. A large fraction of them seemed to be tourists.

The more I watched, the less I sure I was of what was going on. It was a Saturday, so I wasn't expecting class to be in session. But these people weren't really acting like students. Some of them were, but not many.

There were signs up announcing "FREE CONCERT TOMORROW". The Violent Femmes would be playing. I made a note that perhaps I should drop by. Then I double-checked the date on the signs. The signs' "TOMORROW" was my "today"--they'd been posted the day before. Soon they'd be playing at some festival in a place called MAC QUAD. I didn't have much time to find this "MAC QUAD". Other signs let me know that there would be dunk tanks there. So now I had a landmark to look for.

I walked out of the Quad, through a gate. There appeared to be a festival going on. Dancers were dancing in a circle, singing hey-uh-hey-uh. I looked around. I didn't see a dunk tank. I was looking for the Spring Festival but had stumbled across the Harvard Pow-Wow. I kept walking. I walked past science-y looking buildings. I wandered into what turned out to be law buildings.

There were students here. They didn't look like the ultra-snooty people I'd seen in Harvard Square. Unlike the people in the Gato Rojo, these people didn't inspire me to mayhem. They seemed pretty normal. I was beginning to re-think my earlier dislike for Harvard students.

Then I stumbled into a science building. They had the Mark I on display there. The Harvard Mark I (a.k.a. the Automatic Sequence Controlled Calculator (ASCC)) is one of the world's oldest computers. (In fact, I thought that it was the oldest, and said so in this travelogue, but a kind reader has since set me straight.) I goggled at it for a while. I didn't find the Cruft Lab, though a map indicated it was close by.

Photos: the Harvard Mark I (a.k.a. the Automatic Sequence Controlled Computer)

I exited the science building, crossed Mass. Ave., heard the sound of guitars and amplifiers. I walked down a narrow street which opened up into a wide square. Sunshine beamed through the smoke of cooking fires. Many large inflatable play areas towered and bobbled over the scene. And the Violent Femmes were playing.

What's History? Whose History?

Photo: You can almost see the Violent Femmes. I circled them in red, like that helps.

People come to Boston looking for history. Most of them, as near as I can tell, are looking for the USA's War of Independence. I'd come looking for some pieces of maritime history. But here was a piece of my history. I remembered seeing the Violet Femmes play at the Warfield theater in San Francisco. They had done a great show. The crowd had been great. The crowd had kept calling out for more music--and the Femmes had already played all their songs (one of them twice). They had ended up playing the theme from the Batman TV show. Even that seemed great.

Harvard: It's Not Cal, but It's Okay

Maybe I was in a good mood from the music, but looking around, I was once again struck by the lack of snootiness in this crowd. These people looked like normal college kids. They weren't dressed to impress; they weren't dressed like frat boys sweatshirting in lockstep; they were--they were diverse or something. I wondered if some of these people were party crashers like I was. But a lot of them looked smart enough to be Harvard students.

I bobbed my head in time to the music. I decided against picking up any of the free food. I wasn't really supposed to be here. Stealing a listen of some music seemed okay. I didn't want to steal any atoms, though.

This was different from the concerts of my youth. There wasn't a mosh pit. These kids were very well behaved. I was pleased. By the time I'd arrived on the scene, it seemed like most mosh pits turned into something ugly. Yes, I surely didn't miss the presence of a mosh pit. I fidgeted. Really, I did miss the pit. Being a massive kind of guy, I was often placed as a bulwark between my friends and the pit. I was thinking this when I saw, out of the corner of my eye, a student getting shoved away from a wall.

It wasn't really a wall. It was an inflatable wall, one of the barriers of one of those inflatable bouncy play areas--this particular play area was called "Star Walk." Inside this thing, people were jumping around, bouncing off the floor and walls. With each bounce, the walls would bulge. When someone bounced off a wall, the wall would bulge out quite a bit. One of these walls was facing the stage. Some people had been leaning against it, and had been shoved away.

It was sort of like a mosh pit, but encased in plastic. I walked over and stood next to the wall. Each time it bulged, I got shoved. Each time someone inside tossed themself at a wall, I got shoved harder. It was sort of like being at the edge of a mosh pit, but with more padding between me and the elbows of the moshers. I started thinking of the inflatable thing as the Mobile Inflatable Advanced-Safety Moshing Area.

But mostly I just listened to the music, enjoyed the not-too-warm sunshine. They played "American Music". They played "Kiss Off," the Sesame Street counting game of my generation of college students. They played "Old Mother Reagan" and I tried not to think about whether or not these kids had been politically aware back when Reagan was in office.

At one point, Gordon Gano (the singer) pointed out that it was May Day. He wondered if these Harvard students were down with the cause. Many students cheered. I was reminded of the wise words of another band of my youth, the Dead Milkmen:

I do not like you radical
Hate you and your fancy school
You're wrong about the working class
I hope they kick your Harvard ass.

(emphasis mine). I liked Mr. Gano better when he was playing than when he was talking.

I listened to music, I shoved back at the Mobile Inflatable Advanced-Safety Moshing Area. I clapped, whooped, yelled. I realized that my sore throat must be completely healed if I was able to whoop like this.

Not Construction Workers

1999.05.01 SUN Cambridge (Mass. Ave)

When the concert was over, I headed South. I walked past one of the dorms. There didn't seem to be anyone around, but I heard voices. I looked around, and then I saw them. There were some college kids sitting on the porch. I hadn't noticed them at first because they were sitting on the porch--on its roof. But I noticed them now, and they noticed me noticing them. A woman laughed, sheepish. I went back to looking ahead, started walking. Behind me, I heard a wolf whistle. A good thing about being a geek is that people are more likely to think that you're wolf whistling because you're sitting high up like a construction worker; you're probably not whistling because you're a sexist pig who objectifies men.

So I didn't go back and kick her ass or anything.

Glory

1999.05.01 SUN Cambridge (MIT)

I arrived at the building which housed the MIT Museum. This time, I wasn't going to the museum. I was going to see the Tech Model Railroad Club.

It was on the floor below the museum. It was in a big room. There was lots of stuff in it. There were college-age students in TMRC t-shirts. There were some older men, alumni. There were a few students, come to look at the open house. There were parents with children, come to look at the trains.

I don't know what to say, exactly, about what I did there. I didn't pay much attention to the tracks or trains because I was mostly interested in the control system, which had indeed been wired up from a bunch of old phone relays. The gates didn't seem to be controlled by rotary dial, but it was still fascinating (if you're into that sort of thing).

It was a lot to take in. I was pretty wiped out from headbanging out in the sun for an hour and a half.

Photo: Part of the Control System

I don't know what to say about the control system, exactly. John McNamara, one of the nice TMRC people, agreed to send me some of the documentation for the control system, and he did. They were in Word format, and I converted one to HTML, offerred to convert the others, figuring that this stuff belonged up on a web site, where the public could read it. I mean, it was pretty cool. But he told me that the stuff wasn't really polished well enough for public consumption, that he and his colleagues would want a chance to edit it first. So I don't know exactly what to say about the control system. If I were to talk about how it worked, I'd just want to quote from a bunch of papers which aren't ready for public consumption.

(I've since heard from John McNamara again--some guy apparently got the notes into shape for publication. That guy hasn't been in touch with me. Oh well.)

What can I say? This control system was made from electro-mechanical parts that were part of phone switching networks--a complicated network. These parts were from back in the days before transistors. A transistor is kind of like a switch or a faucet. If you've got phone lines from 15 houses coming into a switch and you want to allow up to three of them to have access to the three data bus lines you have, there's a few ways you can handle the problem. You can have a human operator who physically makes the connection, connecting a wire from a house line to a data bus line. Or, using much less energy, you can put a little power into a transistor, turning "on" its switch, so that it will allow data to flow from the house line to the data bus.

The Grids

Photo: Grid

There was a time between human operators and transistors. This equipment was from that time. Do you want to know how it made and broke connections? Do you? (Author's note: If you don't, suffice to to say that I'm talking about a "crossbar switch." I didn't know the gadget's name when I first wrote this (only learning through a kind reader), so I ramble on for a while here. Fair warning.) Okay, try and picture this. You've got 10 lines running vertically parallel, held about 3cm apart in a metal frame. You can think of these as the house lines. Closer to you, there are five horizontal bars. Each of these bars holds a line; you could think of it as a data bus. You want a way to connect the vertical house lines to a data bus. The horizontal bars are out a little ways from the verticals--there's no connection made. If you stand right in front of this thing, it looks sort of like a grid. Okay. At each vertex in the grid, you put in a little springy bit of metal. This bit of metal is attached to a horizontal bar. It's angled and long enough so that it could touch the vertical that it's closest to.

Now if all these little springy bits of metal connected at the same time, then all of the horizontals would be connected to all of the verticals. Every single house line would be garbling and shoving onto every data line. This would not be good for communication--you just want one house line to one data line.

So you allow each springy little piece of metal to swing a little bit. You set up two notches for the springy bit to rest on--if it's resting on one, it's making a connection. If it's resting on the other, it's not. Most of the time, the springy little bits of metal are resting in the "off" position. But then what do you do when someone wants to make a connection?

You use magnets and big forces. In the "off" notch, you use a magnet that repulses the little bit of metal. In the "on" notch, you use a magnet that attracts the little bit of metal. But this isn't quite enough to get the little bit of metal to change positions, so what do you do? You use big force--you use a motor or a magnet or something to twitch that whole damn big horizontal arm. That jiggles every one of its springy bits of metal. This pulls the springy bits of metal out into space. One bit of metal gets pulled into an "On" notch, makes a connection. The others don't.

You've got a bunch of these frames. You've got a timer. Once a second, the timer goes off. Each time the timer goes off, those horizontal bars which want to change connection give a twitch. If you're at the TMRC, once each second, a light turns on with a click. By watching the bars move, you can get an idea of what's going on--each bar corresponds to a moving train. If a bar twitches, that means its train is moving from one section of track to another. If your back is turned, you can still get an idea of a level of activity--how many connections are being made and broken--just from the sound.

Something This Cool Needs a Purpose?

I talked with Alan Kotok, who designed a lot of the sytem. A lot of what he said was beyond me. But he explained how the predictive part of the control system worked. What is the control system for, after all? It provides power to the tracks. It provides power only to those tracks where a train needs them. If there's no train on a section of track, that track isn't powered. If a train is heading one way along a track, it will need power; if it's headed the other way, it will need the power reversed.

Photo: L to R: Alan Kotok talks with someone I don't remember and John McNamara. From his posture, you might think Mr Kotok was a wack MC busting mad rhymes, but I think he was just talking. In the background, the System provides rhythm.

Suppose you've got a train on track section 23. You know it's heading towards track section 24. Should you have to turn on the power to section 24 by hand? No. The system can tell when a train occupies a section of track, so it could wait to power up section 24 until such time as it detected a train there. But meanwhile, the train would be on unpowered track, screeching to a halt. So the system figures out what section the train's going to next and powers up that section, too. It predicts the future.

There's a bunch of relays that keep track of the states of the track-switches. There's a spaghetti of wires that represents which section of track follows which other section of track based upon the current state of the track-switch-state relays and the train's direction of travel.

Photo: Spaghetti

I liked the spaghetti. There's a maneuver that comes up in topology, maybe in graph theory. The idea is that you can get the converse of a map. That is, you've got a plane that's divided into regions. The borders of the regions are defined by lines, and those lines meet at vertices. You can turn every region into a vertex and every vertex into a region. I swear, you can. The borderlines swing around like something out of Drelbs and you've got a whole new map which is somehow related to the old one. The converse of a cube is an octohedron (an 8-sided die). The converse of a tetrahedron (a 4-sided die) is another tetrahedron. You might think of the train tracks as lines; you might think of the track-switches as points--the lines meet at points. But in the spaghetti, the train tracks are points, they're just little spikes, little dots--and the track-switches are lines, wires connecting those spikes.

Tricky Bits

The system could get you into trouble. It didn't really know where every train was. It knew that trains heading North on track section 23 would hit track section 24 next. Maybe it knew that there was a train on track section 23, that section 23's voltage was pushing trains North. But maybe there are actually two trains on section 23. Maybe one of them was overtaking the other. Maybe there's no collision, maybe the one in the lead is able to pull ahead. Maybe the train in the lead gets up to track section 24. At this point, the system doesn't think there's one train in section 24 and another in 23. It thinks there's a train with its engine in 24 and its caboose in 23. If the front-runner train gets ahead to section 25 and the other train's still in 23, the system might realize that this "caboose" is actually another train. But it won't know which. The instant that train caught up to the front-runner, the system lost track of it; it was slaved to the front-runner. Now that it's separated, it's a "ghost train".

This happens in real-world train systems, too. If you read comp.risks, you'll see plenty of references to ghost trains, artifacts of control systems that don't keep track of where specific trains are--just which tracks are occupied.

Each time a train crosses from one section of track to another, the system has to think--it has to figure out which section of track to power-up next. It can only think about this for one train at a time. Its clock ticks once a second. If more than one train crosses a track-section boundary during one second, the system can't serve both of them at once. It actually has circuitry to manage a queue of think-requests. The queue looked to me like it should work; still, the thing has been rigged up to sound an alarm when it gets overloaded; this implies to me that it does overload.

I was up in the control tower with a student. The student was actually working the controls for trains--twisting knobs on control boxes, feeding more or less power to trains so that they might move faster or slower. If this were a simple system, track power would run from the control box to the tracks. The complications of the system come in trying to figure out which sections of track should receive the power from any given control box--so that each box can power up the sections associated with one train as that train moves around the tracks.

Further Down the Track

While I was up there, the student squawked a bit. There was a problem. They were going to have to reset the system. I asked what happened. He said that the system ticked once a second. Sometimes action on the track was happening so fast that the system couldn't keep up. The queue I mentioned earlier could help with situations where two trains crossed section boundaries at the same time. But other situations could creep up. If a train was crossing section boundaries too quickly, the system might never catch up with it--there were race conditions. The student pointed with his chin over at a corner of the room. Over there was the start of a solid-state control system--special boards which could plug into an ordinary personal computer. Such a system could fulfill the function of the old phone relays, but at a clock speed of 660Hz instead of 1Hz. It would take a speedy train to confound such a device.

John McNamara seemed to have mixed feelings about the new plug-in boards. "Us old timers think there needs to be a hook-up to a sound card so that we can still have this." He made a hand gesture, a wave which took in the clicking, twitching heartbeat of clocks, bars, and magnets. "You can really take the whole thing in. The new system's so quiet."

I didn't like the new system as well as the old. But maybe it wasn't my place to judge. I mean, I didn't have to maintain this system. And maintenance was non-trivial. It's not so easy to find people who know how to take care of this equipment. John McNamara said, "There are some parts of this equipment that no-one knows how it works. Even Alan [Kotok] has forgot. It's too reliable--we never have to get in there to fix it." Mr. McNamara said he had worked in MIT's telephone switch area. There were parts he'd never learned about--and parts he'd learned all too much about.

I wasn't going to tell these people they couldn't convert over to a new system. I was talking to John McNamara about the system. He said that part of the problem was "bouncing"--sometimes a voltage doesn't go cleanly from "on" to "off", but sputters back and forth a bit. You have to filter the signal a bit if you want to only catch "real" transitions. I perked up. I'd debounced some switches for an electronics project. After the ivory-tower heights of programming a microprocessor, debouncing those switches had felt good and primitive, like working pottery. I said, "So can you debounce the switches?" Mr McNamara nodded, said, "Sure. Just hook up a couple of flip-flops." I grinned and nodded. That's what I'd done.

It wasn't until later that I realized that my flip-flop circuits, those simple, primitive circuits--those flip-flop circuits relied on transistor technology. I wouldn't have the foggiest idea how to de-bounce a switch if I didn't have transistors to use. I wouldn't have known where to go to learn such a thing. This was--what--less than fifty years after the development of the transistor? Is that how long it takes for this knowledge to go esoteric?

Those relays, for all that they were mysteries to me, had put in good service. They'd served telcos well before their tenure in academia. Some of them, it was said, had been donated by Vannevar Bush himself.

I don't know what to say, exactly about what I saw there. I mean, I'm not really ready to talk about the specifics of the circuitry without quoting from those documents. And what I have said seems rather wishy-washy, hand-wavy, short on details.

I guess I'm trying to say that I was impressed.

Maybe, despite its lack of aesthetics, it was near time to bring in digital circuitry to control this monster. Still, I hoped that the old system would be documented, recorded, preserved, enshrined, that its heart would go on beating.

Photo: part of the Control System

Man on the Street Interview

1999.05.01 SUN Boston (South End)

A man asked me if he could see my t-shirt. "I was wondering if it was supposed to be that 'American Gothic' couple as tourists."

I said, "I think it's just supposed to be tourists."

He said, "I like that American Gothic idea, though. I think I'll paint that."

I said, "Okay."

He seemed kind of disappointed that I just said, "Okay."

I said, "Uhm, that would be cool." Then I nodded and continued walking.

I was exhausted. Back in my room, I was up for nothing beyond watching TV. "War Games" was on. I watched David the überphreaker crack the WOPR.

The Next Day...

1999.05.02 MON Boston (Chinatown)

I went to the South Street Diner for breakfast again. It was mighty good.

Not Everything Goes Right

1999.05.02 MON Boston (near the State House)

I made my way to the African Meeting House. As I approached, a 4x4 vehicle was slowly driving away. The sign outside the building. made it clear that it was open on Sundays. I walked up to the door. I tugged. I rattled the latch. I tried the other door. I looked around. The 4x4 vehicle was there. Its occupants were watching me. Obviously, they were tourists who had been foiled just as I had. I slumped my shoulders. They drove off.

I didn't know what else to do. I was still kind of wiped out from yesterday's intense activity. I stumbled over to a national parks building, skimmed their brochure selection. There was nothing there that I wanted to do. I was starting to stress: what was I going to do?

Right Back Where I Started From

I shook my head, cleared it. I didn't have to do anything. I was pretty sure that when I got back, I'd be able to get a job back at Geoworks, trying to spin off again. I was sure I wanted to. This might be one of my few remaining days to relax for a while.

I took my new book on the history of IBM's System/360 computers over to the Boston Common. I flopped down on the grass. I looked around at people playing frisbee, people playing catch, people just plain walking around. I didn't move. I reclined. I flipped open my book. There was sunshine, but I was in the shade of a tree.

I sat in the park, reading, watching the crowd, forcing myself to relax. I reminded myself that I didn't have to spend every waking moment in directed activity. My time seemed very precious: tomorrow, I was taking off on a plane. Surely the best thing to do now was to sit back and relax. And that's what I did.

| comment? | | home |