What I Went Through Finishing Up a Ph.D. (Jan. 13, 1997)

Tom Loos

Howdy, I'm just about ready to accept a job after finishing up a Ph.D. Since no one ever told me what it was like to finish it up, I figure I'd let you know a little bit about my story.

Preliminaries (up to July, 1996)

I started at in the Indiana University (IU) Computer Science (CS) department in August, 1991 after working for for 4 years. I had a master's from North Carolina State University (NCSU) when I came here -- 'bout all that did for me was make this trip "Ph.D. or Bust". I passed our screaming exam (written quals) in 1993 and had never given a talk before my orals in Nov. 1994, except for class lectures and one 5 minute oral report. So, my orals were probably a much bigger deal to me than most people. It was the first time I was naked and alone, left only with my wits to defend me from a potentially hostile audience. Actually, my orals committee (who were also my research committee) were actually very nice to me and they passed me. Shortly after that, my girlfriend and I broke up, and a couple of months later, my best friend moved to take a job. Needless to say, that semester was a bust other than finishing up classwork. After getting over that, I started to work in semi-earnest in the summer of 1995. In other words, it took me almost 4 years to get thru the preliminaries to get started working which is a about average around here.

My research involved figuring out how to estimate the time a given parallel program (a linear system solver) takes to run. Since my advisor takes an empirical approach, I first was told to write a solver. My first two stabs at it didn't work, mainly because I was trying to adapt someone else's system to my problem instead of just solving my problem. Figuring that out only took a year or so, after I had to present my not-so-hot results at a conference. Once I got back from the conference, I started working on my third version of code from scratch.

Doing the work -- June 8, 1996 thru Sept. 24, 1996

Coding went really smoothly for about a month or so. Then, I started to vacillate between serious procrastination and serious work. After a few days of this, I realized I was terrified about finishing. During this time, each new bug wasn't greeted with "Oh, shit, I'll never figure this out." Instead, it was more like "There's another bug. It should take maybe an hour to fix it. So, I better take 2 days to make sure it only took an hour." I had lots of thoughts about "What are the consequences of this design?" vs. "I should just get the damn thing done." I'd probably still be working on my code if it weren't for my advisor, Randy Bramley. My advisor only really said one thing during this time "Do." This was the perfect advice for me.

Then, about August 1, two things happened: research progress hit a plateau and my mind went down some different alleys -- thinking about the limits of human perception, the nature of the universe, and my relative position in it all. Now, I've never been much of a spiritual person, so I seriously weirded myself out for about a week with these thoughts. I probably also weirded everyone out in my lab, too, but instead of running away in horror like I would have, they just watched me climb the walls, listened to me ramble, and went about their business. I hope everyone has such great friends when they go off the walls. That may sound facetious, but face it, friends don't have to do ANYTHING when we're weird -- they just have to weather the storm by politely ignoring us.

After about 2 weeks of this, I told my advisor I was thinking of setting a defense date for the middle of November, three months away. Being his usual self, he said "Do." So, I asked my committee and by Aug. 23, I had my defense date set for Nov. 15. On this date, I didn't have even a partial result, I only had an outline of a thesis and part of one chapter done, and I didn't have 3 months.

After I realized this, I was in serious panic mode. What the hell had I gotten myself into? I don't usually bluff much at penny-ante poker, none the less with my professional reputation. So, I started working with a frenzy, but only made fitful progress and the clock was ticking. At first I gyrated between writing code and writing the dissertation, but finally decided that if the code worked, I'd have something to show even if the dissertation wasn't done yet. Then, the morning of Sept. 24, 1996, I realized I had one very basic bug, fixed it, and boom, RESULTS.

Let me say that again:

RESULTS!!!!!!

I don't think I was ever that happy before in my life. I'd made the big bet on myself and then just covered it. Different people have different reasons for working on a Ph.D. -- the common ones seem to be to avoid "real life", interest in a subject, working with a particular person, and even to become an academic. I guess that my real reason was to see if I could bet on myself. Now that I knew that I could cover such a big bet, it was time to bring it on home.

Writing it up -- Sept. 25, 1996 thru Oct. 16, 1996

A friend of mine had told me it took him 2 months to write up his thesis and he didn't have to work that hard at it. Well, I had to have it done by October 18, so I didn't quite have a month. Fortunately, I did have the outline and most of the library work done. Most of the rest of the library work was done on the Web. Let's see, my big three tools were UCSTRI to search for computer science techreports, Alta Vista for searches, and the MLA Style Electronic Citations Guide as my citation style guide. The text was done in LaTeX with Emacs, though another friend told me about Lyx , a nearly what-you-see-is-what-you-get editor for LaTeX about a week too late for me to use, but it looked both cool and useful enough that I would have if I would have started writing with it.

The first complete draft was done on Oct. 2. My advisor also gave me his critique: "It sucks. Do it right." He left town for a week and a half -- I did a major rewrite while he was gone. Then, he let me sit on pins and needles for a day until he read it the first GOOD version. While he read and critiqued the first good version, I edited. After he was done, he said "Well, it doesn't suck." If I would have known that was all the better it had to be, up front, well, I wouldn't have gotten into such a tizzy. Let me make this perfectly clear: A DISSERTATION DOESN'T HAVE TO BE GREAT, IT JUST CAN'T SUCK. With that glowing recommendation, I handed it off to the committee on Oct. 16, 1996 and took off for home to (a) see my family and (b) leave the scene of the crime.

Waiting, Phase I -- Oct. 27, 1996 thru Nov. 14, 1996

By now, I'd decided I was going to try to get a corporate job, so I had arranged an interview in September with a large computer company. The dream of getting that job was part of what kept me going through September and October, so I have to say thanks to them. Needless to say, I loved the place, the work, and everything but, of course, I didn't get the job. I really don't blame them -- I don't think I was quite in the real world yet when I interviewed with them. The only really bad thing about it was that it took 5 weeks to get a definitive answer. That taught me (a) not to be shy asking my (future) boss how things went and (b) delay usually means no.

While I was waiting to find that out, I got my defense talk ready. I figured if the dissertation's outline was good, why not use that outline for my talk? So, that made things fairly simple. I don't really remember being nervous on a day-to-day basis, but I did have some really severe panic attacks. Then, I realized that my committee was there to help me do the best job possible with my dissertation. Using that approach, the defense was just a way to explain and clarify my ideas. Once that was done, it really wasn't that bad.

Defense -- Nov. 15, 1996

I defended. I was pretty loose for it, but I didn't feel cocky. I knew the answers to most of their questions, and all in all, felt pretty good about it. I was really lucky that one of my committee asked me to describe my work informally while waiting for the last member to show up, so I was both ready to explain and informal while I talked, which was much better than the orals. I don't know about anywhere else, but the IUCS dept. just has the defender leave the room, they discuss things, and then they call you in to tell you the news within a few minutes. Waiting to find out wouldn't have been so bad, except my defense was in a conference room right off of the CS dept. office so I was asked "How did it go?" and felt like I had to make small talk when I really didn't want to. The committee was really in a good mood because they passed me after about a 15-20 minute discussion. My talk still sucked enough that they made me give the talk in front of the dept. to practice it and they asked for some significant revisions.

Finishing Up -- Nov. 16, 1996 to Jan. 2, 1997

I made most of the changes, went home for Thanksgiving, and rewrote my talk for the dept, which was held on Dec. 6. I finished up the changes and deposited my dissertation on Dec. 11. To deposit at IU, you need to turn in one unbound copy on 100% cotton rag paper (at $0.20/page) with the original signature pages of both the dissertation and the abstract. Eventually, for a CS Ph.D. two bound copies are required: one on cotton rag for the Grad School and one with any old kinda paper for the dept. Since I had 160 pages in my finished dissertation, it cost something like $70 for the required copies, and since I wanted a couple of extras, the total copying bill ran to $105 or so. Then, I had to pay the bursar $55 for a dissertation fee and $35 more to copyright my book. Finally, there are two binderies in and around Bloomington -- one is in town and Franklin Bindery, outside of Spencer, about 20 mi. away. I went with Franklin because they charge $25/copy for 2 week service vs. the one in town charging $24.50/copy for 3 week service (which, for me, would have actually been 4 week service, since the bindery in town didn't work over Xmas week). My total "finishing up" costs were around $300. So, after going home for Xmas, I picked up the books from the bindery and turned them in to the grad school on Jan. 2, 1997 -- finished with the Ph.D!

During this time, I had two job interviews: one with Lucent (the successor to AT&T) and the Aerospace Corporation. After my experience with Compaq, I learned to act like a house guest during an interview -- be polite and don't say anything nasty. Actually, don't say anything at all, if you can help it. After all, I figure I'm there to learn if they're cool people doing cool things, so let them show me that they are, and the less I say, the less chance I have to screw up. This interviewing strategy led me to two job offers, so I figure it was OK. My experience with interviews is that if they don't sound like they'll say "Yes, we'll offer you a job" at the end of the day, it's gonna be "No."

I got the interviews by posting my resume to the Web and/or Usenet -- I'm not exactly sure where the interviewers read my resume. Some good places to start are Career Mosaic, the On-Line Career Center, and the Usenet newsgroups misc.jobs.resumes and us.jobs.resumes. For lists of resources, I really liked The On-Line Job Meta-List and Purdue's Job Sites List. Posting my resume directly to employers seemed to lead mainly to postcards that said they'd gotten my resume, not to phone calls. The best way to get a job is to ask people who already know you; i.e. former employers, through friends, etc. In hindsight, I should have asked my friends for jobs, since it's (probably) better to owe a friend a favor than some stranger, but I didn't really ask friends. I will the next time I need a job though, so any of my friends reading should be forewarned... Of course, your mileage may vary. One last thing, no matter what they say, no matter how good it sounds, get it in writing before you decide. Otherwise, you'll be on pins and needles until you get that offer letter...

Well, that's about it. If it goes the way I think it will, I'll be leaving Bloomington within a week. If you decide to go for a Ph.D., keep plugging along (you'll eventually get there if you do!) and good luck.

A Few Words of Wisdom?

For some good advice, try reading How to Be A Good Graduate Student. For my advice, read on...

I often thought that doing research was like was navigating an ocean without a map. Sooner or later, if I kept sailing in some fixed direction, I'd most likely hit (pay) dirt.

A few do's and one don't:

Finally, Deepa Viswanathan and I came up with the three rules of driving:

  1. Be comfortable.
  2. Go with the flow.
  3. Follow the rules.
They work pretty good for research, too.

Thanks!

Thanks to my parents and my siblings -- Jerry, Kelly, and Heidi.

And thanks a bunch to a bunch of people for making IU an interesting place (in rough order of meeting them):

tom loos