Social Life of Information

This post is the story of my thesis. I’m recording it for the benefit of any Wolverines interested in the Masters Thesis Option Program (MTOP) at SI. The materials and advice that the school provides about getting started is good information and will get you on the way. But after the MTOC (Masters Thesis Option Committee) gives you the go-ahead to work with an advisor on the topic that you propose, you’re pretty much on your own to make everything happen. Your advisor is there to help you along the way, but the work is all up to you. There’s not a lot of precedent (at the time I’m writing this, there have been 5 MTOP theses completed) and that can be intimidating and frustrating to new researchers. So here’s a summary of what the experience was like for me.

About a year ago, I started working with Mick McQuaid on my thesis. Actually, I started working on my thesis much earlier than that, in December of 2005, when I started collecting the data as an independent research project (outside of classes) and then turned it into a course project for Lada Adamic’s SI 614 course, Networks: Theory & Applications, in the following Winter 2006 semester. I asked Lada if she would serve on my committee at the very end of the semester, so when I put in the MTOP application, I already had the required additional committee member besides my adviser. With over 7 months of work on the project, it still wasn’t officially thesis research until the MTOC accepted my application. I turned my application in early, sometime in July, and had their approval back in under 2 weeks, so I started work in earnest at that point.

I spent the larger part of the Fall 2006 semester on lit review. For my thesis, I had to collect literature from across several disciplines, which can prove challenging to synthesize. I had only done one lit review prior to my thesis, so I was pretty clueless as to how it was going to be useful to me. After collecting a massive load of articles and stuffing their citations into bibliographic software, I read them all. I read about 40 articles on the drive to and from Syracuse to visit the iSchool here, as a matter of fact. And then I summarized them all into about a paragraph each, sometimes including extensive notes on how I thought it might fit in with my research. All in all, I read around 74 articles that made it into my lit review (and more that did not!) and generated a 26-page single-spaced document of notes on the literature.

Then I summarized all of my notes, so that each article’s main contribution to my investigation was discussed in just one or two sentences, and assembled that into coherent paragraphs, subsections, and sections. My final lit review notes condensed all of that material down to 6 single-spaced pages, which turned into about 8 double-spaced pages in the completed thesis. And that was the whole fall semester. Well, besides the part where I presented the early research in Zürich, Switzerland, in October - but that’s another story.

I also took SI 840 in Fall 2006, the doctoral level research methods course. It was absolutely invaluable. By the time I finished writing my literature review, I felt fully prepared to write the methods section of my thesis as well. It also helped me during the course of the semester as I was reviewing the literature, because I was able to keep an eye out for articles that supported the methods I planned to use.

By the start of the Winter 2007 semester, I felt like I had some idea of what I was doing, but not nearly enough time to do it. I wrote up my hypotheses section and despaired of ever being able to answer such a big question. One very useful thing I’ve learned is that hypotheses can be edited! By this time, the focus of my thesis was something fairly different from what I had originally proposed to the MTOC. The methods were the same because the data set that I wanted to work with dictated that much, but the analysis and the hypotheses had shifted quite a bit. Fortunately, I was able to complete my data collection in two (grueling) weeks, because I gathered artifactual data from the Internet.

Once I had my full data set in hand, Mick helped me create some scripts to parse it into a network data set upon which I could perform network analysis. Mick wrote a starter Perl script that I then spent quite some time to customize to my precise purposes. I also wrote my methods section at this time, since my methods were already set. This was without a doubt the easiest part of the thesis for me, and I credit that entirely to Mark Ackerman’s SI 840 class.

My Perl scripts eventually generated the right data sets from my raw data, and then I spent some time on analysis, which is my favorite part of the research. That was about the time that my mother died, which really threw a monkey wrench into everything for me. I had to take some time out just for the ordinary matters of attending to an unprepared estate, and then a little more time to get back up to speed. Grief really slows you down, in a very literal sense.

Despite the setback, I got back to work as quick as I could. At mid-March, the end of the Winter 2006 semester felt very near; I struggled for what seemed like ages to learn enough R to get my statistical analysis complete, and then spent many hours writing up the results. This was much more challenging than I would have expected; I also learned LaTeX in order to typeset my thesis according to the Rackham dissertation style and that was no small effort, but Word never would have been able to handle this document. Eventually, my tables broke across pages as desired and all the figures appeared, but I actually sent it to Lada and Mick for some feedback before I even got that far. Lada’s early feedback on my very rough draft made me return to my regression analysis to verify the results by using a step function, and this had a big impact on the outcome, so I rewrote my regression results accordingly.

I finished writing my thesis (the first time) about a week before I was set to defend it. The defense is another thing that’s entirely up to you; Mick and I went down to the administrative office and booked the Ehrlicher room for a convenient time during the last week of classes, a Tuesday afternoon when he and Lada and I were all available. Then I had to write an abstract that Mick emailed out to SI-all; that too was intimidating because I have almost no experience with writing abstracts. How do you condense a year and a half of research down into about 200 words? After all the time I’d spent writing the thesis, it wasn’t as hard as I had expected.

The week before my defense was spent in honing the visualizations from my thesis, which is extremely time-consuming, and putting together a coherent PowerPoint presentation. The thesis defense format was a 45-minute presentation with some time for public questions and then private discussion with my committee. How do you condense a year and a half of research down into 45 minutes? That effort was simplified by the time limit; I didn’t have to go too much in depth into some of the involved details because there just wasn’t time enough to do so. 45 minutes was really just about enough time to cover the basics and provide the highlights of the results of analysis.

As I was preparing my thesis defense presentation, Mick did yet another thing that really helped me out - he provided a list of some questions that he might consider asking during the public question part of the presentation. Some of the questions were a little scary, and it turned out that the very first question posed to me was the one that had worried me most. Because I’d had a little time to consider it, however, I had a confident answer ready. A few people asked a few good questions, and I answered as well as I could; then the room was cleared so my committee could confer. I spent 5 minutes hanging out in the student lounge, feeling generally relieved, and then returned for private comments from my committee.

They were generally complimentary, but brought up a few things that I had to follow up on. This is apparently the usual for a thesis defense: you defend your thesis and you pass, but your committee also gives you more edits to be completed. Lada specifically suggested that I pursue a couple of entropy measures, and so sure enough, once I got around to trying that out, it completely changed my results all over again.

By this time, I was in the middle of relocating to Syracuse; the home-buying process required most of my attention for awhile, so I had to put thesis edits on hold until we moved. Relocating takes up a lot more time, effort and attention than I had expected, so it was late July before I continued my thesis edits. It had become much easier to write by then, since it was the third time I rewrote my regression results and the discussion thereof! One last draft went out to Mick, who suggested just a few more edits. So I made those final edits, and then I was done.

Well, mostly. Each thesis adviser and committee member can set different requirements for the final product; Mick required a bound, printed copy, while Lada required a pdf. I went through Lulu.com for on-demand printing of my thesis, and like everything else, producing an acceptable document for printing required some additional unexpected effort. Because of the way my figures were generated, a bunch of non-embedded fonts were used in the document; I solved the issue by completely rasterizing the document. Maybe not the best approach, as it took a 3 MB document and turned it into a 28 MB file, but it solved the problem that was preventing my printing. At the same time (this past weekend) I also posted a PDF copy of my thesis on my web site with new visualizations of the full iSchool and Computer Science department hiring networks.

And that takes me up through today. This morning, I ordered two hard-bound copies of my thesis, one for me, and one for Mick. It cost about $80 for full color printing, binding, and shipping; when I receive them in a couple of weeks, a few more dollars will send one copy off to Ann Arbor. While I’m aware of Deep Blue, Michigan’s institutional archive, there are currently no guidelines that I’m aware of regarding depositing MTOP theses for archival purposes. (ironic, no?) I’ve emailed Deep Blue to find out what is required and will post a follow-up if I have any success with getting my thesis deposited in the institutional archive.*

I could say that’s the end, that after a year and a half of effort, I’m finished. But that’s not really true; the work to turn the thesis into a journal article remains, and that will take some time. Still, having ordered the hard copies, I have completed the largest single project I’ve yet undertaken. I expect the experience will make it a little easier next time.

*follow-up: The good folks at Deep Blue replied in under 24 hours. Theses and dissertations from SI students are sent through Sue Schuon, and there are some nice instructions on how to prepare an archival PDF for inclusion in the institutional archive. Good show, Deep Blue!