My thinking has always been that if you sacrifice everything to your job, then you should not be surprised if you end up at 55 or 65, alone in a house with only a cat and nobody calling you.

The first thing I did when I finished my Ph.D. thesis is to go hunt for a wife. I put everything else aside: research, almost all job hunting efforts, teaching gigs… I often forget about this part of my life, but for almost 6 months, I was doing nothing else, but trying to start a relationship with a woman I’d love for the rest of my life. I found a great, beautiful, extremely smart lady (intelligence was the criteria for me) and we’ve been together almost 7 years now, and we have a great son (14 months old now). Maybe I take it for granted sometimes, but I really shouldn’t. The important point is that it didn’t happen by accident. I worked really hard to find the right woman, and it was my full time job for a little while.

The person I met worked as a freelancer over the Internet (and still does), so moving was not a big no-no: the truth is that you should expect to have to relocate when you hold a Ph.D. and want Ph.D.-level jobs. Sometimes you can have to relocate quite far as well (which I avoided). I also don’t think it is fair to ask someone else to drop everything because you have to move. So, if you just got a Ph.D., the hunt for love is actually more complicated.

I don’t know how many of my colleagues are single. I would estimate that among Ph.D. holders my age, around 30%, maybe only 20% of all of us are single…

Household Opera comments on The “single woman in a rural college town” blues and she cites an interesting account:

When I was not in the classroom, the silence became deafening and I became clinically depressed. I love to read and I love solitude, but like everyone, I need some social interaction.

My colleagues, on the other hand, often worked at home, and when they came into the department, they shut their doors and hibernated. Having spouses and families at home, they had no need to create social relationships at work. I found myself drifting with my only interaction being with my students or a clerk in the grocery store.

I don’t know how real this problem is for single Ph.D. holders. It seems like it can be a real problem, and apparently, more so for a women.

I also think that women may tend to forget too easily that their window of fertility closes around 40. Most woman are infertile at 45. And even if you have kids after 40, they stand a much higher chance of having all sorts of medical conditions. While a 40 years old man can start a family, a woman should think twice about it, assuming she even can anymore.

I think we should rethink the entire academia roadmap. People used to become university professors after getting a degree, maybe a master, then the Ph.D. became a requirement, and now, in many fields, you need at least 4 years of experience after the Ph.D. if you are to get close to a professorship. All the while, the expectations at every step become tougher and tougher. The trend is clear: as competition increases, we will hire older and older professors… and these people have sacrificed more and more for their work…

Note: I’m still very much sick. I talk with a funny, deep voice this morning.

I’ve been suffering for the last two days from a terrible cold. My brain power is running at 20%.

Not only can’t we cure the common cold, but we also cannot efficiently take away the most severe. Yet, we can presumably clone human beings relatively cheaply. It is really troubling.

It goes to show that science is often unable to address common and very important problems.

Whenever I present my research work to my wife, she asks why I’m not working on something really important, like curing cancer or stopping world hunger… I guess the answer is that scientific research is not about solving problems, it is about trying to make some patterns unfold, and these patterns lead you where they do.

Cygwin is a marvelous idea: run a Linux-like shell under Windows. It allows me to run Python, CVS, Perl… almost everything I use under Linux, under Windows. Well, it doesn’t quite work as well, but for small things, it does the job.

One thing that has annoyed me is their implementation of vim. The keyboard support is bad. In my .inputrc, I have these lines

# enable 8-bits characters ...
set meta-flag on
set convert-meta off
set output-meta on

They seem to clash with vim in a bad way. Ah! But you can also install a version of vim running directly on top of windows. If you do this, then you can use this other version instead of the one that comes with cygwin.

All I had to do was to create this little script:

 "/cygdrive/c/Program Files/Vim/vim63/vim.exe" `cygpath -w $1`

The trick here is that you need to convert Linux-like paths (like /tmp) into Windows path (C:…). My little script is bad in many ways, but it will work if you call vim with only a file name as an argument.

My solution does fail in some nasty ways. For example, when I do a CVS commit, I can’t enter my comment. Bad.

See also my post Grep is just not for matching lines anymore.

Subscribe to this blog
in a reader
or by Email.

Yuhong worries about CS students. She points to two recent articles on the CS job market:

• Endangered species: US programmers (David R. Francis ,The Christian Science Monitor ,October 14, 2004)
• Study finds dramatic loss of tech jobs (Ed Frauenheim, CNET News, October 18, 2004)

Here is what she has to say based on the people she talked to:

I recently talked to some master CS graduates. (…) They both said programming jobs are no more and many new hires are master graduates.

Here are two quotes from the second article she cites:

There are certain areas in the technology sector that are thriving. Demand is high for those who specialize in network and IT security.

Technology services companies like IBM and BearingPoint are hiring in the United States, though they are increasingly looking for employees who can combine technology chops with business savvy.

The message is quite clear, I believe. If you want to train yourself or students to produce software (programming or software engineering), you better be damn good because the job market is not there anymore. Will jobs come back? Automobile workers in North America are still waiting for the jobs sent to Mexico or elsewhere to come back. Now, programming or software engineering are not useless skills, far from it, but it might be a better strategy to aim for a business jobs where your programming or computer networking skills can be put to good use, for example. It seems that the job market is moving toward information technology (security, networking, using the right technology at the right time, understanding the implication of a given technology for business).

Yesterday, I gave a talk at the Centre de recherche en géomatique. I was invited by Yvan Bédard to visit his lab, meet his students and discuss his projects. To give you an idea, Yvan published 400 papers, he has 4 full-time contract employees, two post-doctoral fellows and something like 15 graduate students in his lab. He has for $750k in software alone in his laboratory. He must be one of the top researcher in “spatial OLAP” (which he calls SOLAP) in the world. He works at the geomatics department in Laval University. It is one of 3 geomatics departments in Canada. (While I don’t have a definition of geomatics… it is all about mapping and measuring distances, and so on.)

Some of his projects include decision-support tools for trainers who work with top athletes: through fine-resolution GPS, you can track the an athlete, collect all this data, pour it into data cubes and make various comparison, see where, on the track, the athlete is improving, where he is not improving, and so on. He also built a tool to mine road conditions and help set repair priorities.

We discussed various research issues such as the fact that rebuilding the cubes is expensive in a spatial-temporal context, and how view maintenance appears to be lacking from current commercial tools like SQL Server (I’ll need to investigate this particular issue). He has committed to achieving real-time data cubes in 5 years (which implies continuous updates) for his current portfolio of applications.

The use some already available classical multidimensional indexing techniques such as R-trees and R+-trees. However, they are fairly dependent on currently available commercial (and open source) tools as they focus mostly on the application layer and specifically on the geometry and topology of the data.

I’m very impressed.

Myself, I presented some of the work I do with Owen Kaser as part of the Lemur project, including Attribute-Value Reordering and Relative Prefix Sum Methods. I got people laughing, so either they thought I was ridiculous or else, they enjoyed my style. There were many questions and people seemed to be interested, so maybe they thought it wasn’t silly work after all.