Just some notes on the past few days, which have mostly consisted of finishing a paper and leaving Minneapolis.
First things first: I'm as done with my Senior Thesis in Physics as I really want to be. My advisor even signed the sheet that says it's done!
Last March, I went to talk with my advisor and told her I thought there was so much left to do that I didn't see how I could finish in time. She told me I should just write down what I'd done. So I did. I wrote a beautiful little technical manual on the satellite subsystems and the IDL code I'd produced to work with the data. When I sent this to her she wrote back that this was the stuff of appendicies, not theses… and there was so much more to be done! We both knew she was right, at least in theory, but I'd thought, when she said write it down, that she knew what was being wrote. A big mistake, that.
So, as covered in earlier posts, I moved out to a farm in southeast Minnesota and, with the help of my good friend Beth, finished up my philosophy thesis.
Coming back to Minneapolis, I tried, I really did, to work on the physics thesis. But IDL is a proprietary, for-profit language with a dirth of examples and suitable, I think, primarily to analysis of discrete, limited data sets. What I needed to do with the data was to analyze (quickly, please!) thousands of power spectra. And the method I had for doing it—using ~40 individual commands to plot a power spectra, waiting for it to transfer via SSH-X to my screen, and then exploring the graph with more non-intuitive, messy IDL commands—was not working.
Granted, I could have gone into the space lab to use the computers there, but the ones I had access to were lumber Sun UNIX systems with huge, ancient CRT monitors. The kind that flicker when you look at them. The kind that seem designed to give you headaches. So that was right out. As time went on, the sysadmin of the lab began making it more and more difficult to transfer files into and out of the network, saying that there were security risks. For me, it turned into an insurrmountable access problem and I tried moved everything off of the network. This created several copies of many of the files, making it increasingly difficult to manage the project's data.
So I went to Seattle to visit Vivle. That trip, which, as of the moment, I'm still writing about, was probably the best possible choice I could have made. No small part of the awesomeness was that Vivle's fieldwork was out on the Olympic Peninsula in places where there was no cellphone service, much less internet. In places where they still used dial-up.
We'd muck about in river beds all day long. She would plant hundred of little flags on sand bars and around trees, and I would map them with a total station (aka: awesome surveying equipment). In the evenings, we cooked fabulous food and talked late into the night.
During this time I was still feeling guilty about the physics project so, I decided to do something revolutionary: to rewrite all my code, from scratch, in C++ and Python. To Ditch IDL. And you know what? It worked. I accomplished in a week what I'd been trying to do for more than a year. It was a week in which Vivle and I did less friend-things than I'd have liked.
When I moved back from Seattle, somewhat unexpectedly, I decided to finish the bloody paper by December. And I made a real-time graph to show this progress.
I stayed at my uncle's house and typed furiously. Flat lines in the progress graphs or keystroke graphs (and that's a lot of keystrokes, ain't it?) represented things unconducive to completion, such as sleep, eating, and friends.
By later in the month, I'd finished a draft and my advisor said maybe one or two more iterations would bring us to completion. So I spent good chunks of Christmas vacation working on it. After that, I needed to think about jobs. I was offered one in a place I loved—Juneau—doing work with glaciers, which I liked, but the work was unexciting. I was also given a shot at exciting work in Minneapolis, which I wanted to leave at that point. Having difficulty deciding, I took a trip to Georgia and decided on the job in St. Paul, came back, and started working.
But that couldn't last! By late April, I knew that I'd have to make a big push to finish things, once and for all. So I did it. I know from prior experience that there are times when, if I put myself to it, I can get a lot done. There have been nights, in the past, where I've stayed up all night coding for some class and, in the morning, I don't remember much of what happened, but beautiful, functional, sometimes brilliant code is sitting in front of me. There was this one time when I spent 70 hours trying to get this one physics equation right. There was that semester where I'd repeatedly spend large chunks of my Saturdays and Sundays churning away in the physics lab trying to figure out the mass of electrons, hydrogen spectra, or radioactive decay rates, while my girlfriend looked on in patient exasperation.
Clearly, it could be done again. And it was: 130.4 hours, my time-keeper tells me.
- April 10–16th: 1.4 hours. “I should start thinking about this…”
- April 17–23rd: 26.1 hours. “Hey, I'm making progress!”
- April 24–30th: 40.3 hours. “Uh oh, there's a lot left to do!”
- May 1–7th: 49.0 hours, with two all nighters. “I WILL BLOODY FINISH!”
- May 8–14th: 13.6 hours, so far. “Argh!”
Captain Tony, who spends long days relaxing on the porch of our coop (his relaxation probably cancels out my stress), didn't understand how it was possible for me to stay awake all night for multiple nights with no caffeine. I've always figured that our bodys are capable of nearly anything. You just have to treat them as best you can and train them.
The last week I pulled two all-nighters. I found a flow state both nights, and listened to Israel Kamakawiwo'ole sing “Somewhere Over the Rainbow” for more than twelve hours at a time.
When I was six, my dad and I were driving out to the woods at five in the morning and a song came on about a ship wreck. The ship struck a rock and dropped beneath the waves, but the faithful crew—at least, those who lived—didn't forget their ship and vowed to bring it back to the surface. The ship's name was the “Mary Anna Ellen”. This was before there was an internet, so for years I just remembered the song. Then the internet became a thing I would search for it once or twice a year, but never with any luck. Somehow, though, deep into one of these all-nighters, inspiration struck and I searched something I'd never searched before and learned that it was the “Mary Ellen Carter”. For years, I'd been thinking the wrong name and had been thrown off by it!
Powered by the song, I got the paper turned in and my advisor only wanted a few things added, including the locations of the lower hybrid frequencies. This is something that she may have thought was easy; it was something I thought would be easy, if annoying.
It turns out you have to synchronise two different instruments, for which you have to write data decoders, find a common time-base for the instruments, and then combine the results to get the frequency, which turns out to be discontinuous over time, non-smooth, and difficult to interpret.
If you're looking at the above, thinking “WTWTF?”, so was I dear reader, so was I.
Well, I've sent all that to my advisor now, and we'll see where things go from here. I'd like to publish something from this, but I need to get back to work on my real job (which is awesome and interesting). We'll see how that goes.
Overall, a lot of learning took place as a result of this project. Life, writing, research, research methods, everything was mixed in with it. I'm proud of the process, I may even be proud of the resulting paper. People got hurt as a result of this, I got hurt as a result of this. Sometimes hurt is justified, sometimes it's not.
Would I do it again? Probably. Sometimes I'm stupid like that. There were some things I'd change, though. You should get paid for your work. It holds both you and the people employing you accountable for each other's time and sanity. It represents personal cost in some medium of exchange. It puts a number on something that is otherwise without a number. It also enables you to prioritize work appropriately. Throughout this whole period I had to support myself with other work and because this other work was paid there were expectations, deadlines, and accountabilities. Not so the physics work. It is no wonder that the timeline got extended as a result.
Deadlines are important for many of the same reasons as above. Projects that are important, projects that are done at cost, should be projects that have a timeline and deadlines. Without this, progress is difficult to judge and deliverables don't tend to happen.
Being self-driven and self-directed is important, but it's also important that the people you work with have expectations and state them.