Not my memory. Heck, I forgot what I ate for breakfast.
But I do recall, like it was yesterday, jumping with glee in 1963 when my college let me have their old IBM 1620 mainframe computer when it was replaced. I couldn't wait to truck it home to my garage and start surgically autopsying my new cadaver. I methodically removed every little screw, indicator light, contact switch, transformer.
But most prized of all was the deepest inner soul of the beast. It's processing memory. It was called "core memory," named for the donut-shaped cores, made of graphite, like in pencil lead. During computations the cores would be magnetized North or South. (Or L/R, or U/D, or B/W, or yes, even the infamous 1's and 0's, all of which were untrue. Magnets are either N or S. All those other artificial dichotomies were just authors' imaginations and surrogate place-holders in our number system.) They stopped making core memory like this decades ago.
For days I excavated slowly, carefully, making sure not to break any part. Just in case one might someday want to reassemble this into a revived digital Frankenstein corpse. Finally, I exposed the sarcophagus that contained the sacred core memory. Those little bits that had kept me awake so many nights as a graduate student trying to master the special new realm of computer language that it took to talk to them. They had been so perfectly obedient. Which was the problem.
They always did exactly what I asked. And with only a beginner's grasp of their language I often asked things of them that neither they nor I wanted, or perhaps didn't even understand.
For days I excavated slowly, carefully, making sure not to break any part. Just in case one might someday want to reassemble this into a revived digital Frankenstein corpse. Finally, I exposed the sarcophagus that contained the sacred core memory. Those little bits that had kept me awake so many nights as a graduate student trying to master the special new realm of computer language that it took to talk to them. They had been so perfectly obedient. Which was the problem.
They always did exactly what I asked. And with only a beginner's grasp of their language I often asked things of them that neither they nor I wanted, or perhaps didn't even understand.
I had tried so many hundreds of times to talk in their symbols, asking them to spit back things I already knew, like the trigonometric cosine of a 60-degree angle, or the volume of a cube. But the real bonding between me and these little cores had developed when we jointly shared the intractable goal of calculating a perfect square root of a prime number, or the exact area of a circle. That took special algorithms, a kind of game plan for me to coach this little team of fella's how they would have to work together among themselves to discover something I never had known (such as the square root of 173.69, to a precision of 7 decimal places).
Sometime, later in graduate school, I mastered their language. About seven of their languages, in fact. They were so ready to go to work for a knowing master. Not just an articulate master, but one who cared to reduce their work. I spent a master's degree trying to develop methods to make computers work easier, run faster. I developed some ideas for my computer algorithms to calculate ways to speed up other peoples' algorithms. Techniques that would allow a computer to arrive a conclusion sooner, with less work. Methods that would allow a computer to serve more people in less time. Or help the nascent space shuttle recalculate its glide path in time to land safely.
It was all born of necessity. For months I had been programming multivariate statistical models that solved near-singular eigenvalue problems. And I know that you know, those rascals take a long, long time to compute. I would hand the program to the computer one day, then go home for a bit of sleep. Maybe even take children Christine and Paul to the park to fly kites. Then come back to the computer lab the next night to see what sort of mathematical construct my team of memory cores had built for me (why were those early computer labs always in the...basement?)
So I branched out, developing new mathematical algorithms whose sole purpose was to improve the statistical likelihood that they could glance over the grinding calculations of another algorithm in progress, then recommend a short cut to just leap right toward the answer. Like telling someone running around a maze to stick their head above the walls, see where the goal probably is, then just go there. Why calculate the entire path, when you can just finesse an algorithmic Gretzky?
And it worked. These memory cores were great little soldiers. We advanced some pretty crafty new strategies to accelerate iterative processes to turn near-singular eigenvector matrices upside down, (i.e., like guessing an entire song after hearing only the first few bars. Or putting a "Hurry Up!" button on the front of an old three-ton, water-cooled clunker of a mainframe computer.) They helped me get a master's thesis, and kept up the moneyflow to feed and clothe the kids.
It was all born of necessity. For months I had been programming multivariate statistical models that solved near-singular eigenvalue problems. And I know that you know, those rascals take a long, long time to compute. I would hand the program to the computer one day, then go home for a bit of sleep. Maybe even take children Christine and Paul to the park to fly kites. Then come back to the computer lab the next night to see what sort of mathematical construct my team of memory cores had built for me (why were those early computer labs always in the...basement?)
So I branched out, developing new mathematical algorithms whose sole purpose was to improve the statistical likelihood that they could glance over the grinding calculations of another algorithm in progress, then recommend a short cut to just leap right toward the answer. Like telling someone running around a maze to stick their head above the walls, see where the goal probably is, then just go there. Why calculate the entire path, when you can just finesse an algorithmic Gretzky?
And it worked. These memory cores were great little soldiers. We advanced some pretty crafty new strategies to accelerate iterative processes to turn near-singular eigenvector matrices upside down, (i.e., like guessing an entire song after hearing only the first few bars. Or putting a "Hurry Up!" button on the front of an old three-ton, water-cooled clunker of a mainframe computer.) They helped me get a master's thesis, and kept up the moneyflow to feed and clothe the kids.
Isn't it too bad, that today we hear people use the phrase "core dump" and they have no flinching what the heck they are referring to. How much grunting, grinding, grimacing and genius grandeur they are discounting. Or when someone says "core memory," how little do they get to appreciate that when one refers to a "core memory" it is referring to a phase of human developoment, a period of history, an journey of scientific discovery, an environment for raising kids, a transition to a new realm of mankind's peer-to-peer partnership with machine.
This morning I took this IBM 1620 Core Plane out of a special box I've kept it in for a very long time. Although delicate as teeny donuts of pencil lead, not a single fragile graphite core had been broken after decades of packing and unpacking in moves state to state. I was going to photograph it and put it on sale. But seeing it was more moving than photographs and portraits of the time. One doesn't have a relationship with photographs, although we sometimes do with the people in them. But with this core plane, I did have a relationship. This core plane had been a major participant in my family. It helped feed and clothe the kids, and it helped me develop beyond normal.
And it's still a member of our family.
Next year, in 2013, it will be 50 years old. I can't just sell it. I will send it to my grand children so that 50 years from now they can show it to their own grandchildren and say, this was a member of your great, great, great grandparents' family. It helped buy our mom's (dad's) shoes when they were kids. It opened doors for scientific discovery by our grandfather, and it helped him confirm some mathematical algorithms he invented.
