Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Thesis ideas, please.

Author: Rémi Coulom

Date: 03:39:49 10/23/02

Go up one level in this thread


On October 23, 2002 at 04:44:42, Travers Waker wrote:

>Hi everyone.
>
>I'm considering doing a Masters (5th year) degree in Computer Science, but only
>if I can come up with an interesting topic that can be tested as part of a
>chess-playing computer program.  I've contacted the local University,
>(University of Cape Town, South Africa) and found someone who's willing to
>supervise my thesis if I can come up with a suitable topic.  He has expressed
>concerns about the field of computer chess, saying that he feels that "it seems
>to be a very well explored area with little potential novelty."
>
>I'm not really interested in doing a thesis just for the sake of getting the
>degree.  I  want to deeply explore a topic that I'm very interested in, and
>computer chess is the only one I can think of that has the potential to be the
>basis of a thesis.
>
>Despite the discouraging opinion of my potential supervisor, I believe that
>there's still plenty of scope for new and novel research in the area of computer
>chess.  Much like Mikhail Tal could conjour up tactical fireworks from the most
>barren looking positions, I think it's possible to find an interesting topic in
>computer chess that hasn't been properly explored yet.
>
>So, the reason for this post is that I'd like to request the opinions of this
>board's members (epecially those of you who have been involved in academic
>research) on what areas of computer chess I could look at to find an
>interesting, novel topic for my thesis.
>
>Thanks for any ideas.
>
>Travers

Hi Travers,

I also started my own thesis because of computer chess. When I participated in
the Paris World Microcomputer Chess Championship in 1997, I said to myself "hey,
these researchers are doing for a job what I am doing for fun. I should try to
do that job". But, although I had a reasonable starting point with the work I
had done on my program, I did not choose computer chess as the subject of my
thesis, because it seemed a too dangerous research area. As your advisor said,
contributing significant advances to that field is _extremely_ difficult. If you
have not already written a reasonably strong chess-playing program, and you do
not have a precise idea of what contribution you could bring to that field, then
I would strongly discourage you to choose computer chess as your research topic,
unless (maybe) your thesis advisor is a very strong specialist in this area.

I do not agree with your advisor when he says that computer chess has very
little potential for novelty. I believe that computer chess has a lot of
potential for novelty. The two big problems are that those novelties are really
difficult to find, and that, even if you find some, you will have a lot of
difficulty to obtain a lot of consideration from the research community, because
very few researchers would be able to really appreciate them. Strong
chess-playing programs that beat humans easily are so common now, that you would
not be able to impress anybody with another one.

I do not want to sound like computer chess is not an exciting research subject.
It is. But, when choosing a thesis subject, choosing the topic that you find the
most exciting is not necessarily the best choice. You have to take political
aspects into consideration. Doing some good scientific work is neither necessary
nor sufficient to succeed in research. Being a good politician is both necessary
and sufficient. I learned this the hard way during my thesis.

If you cannot think of any other interesting topic besides computer chess, then
I would encourage you to think a little more. There are plenty of subjects that
have a lot in common with computer chess, because they are related to finding
methods to decide on an optimal action. For instance, artificial intelligence in
robotics is also very interesting and much more "fashionable" than computer
chess, and contributions are much more easy to make in that field.

I am sorry this message does not really answer your question, but that would be
my advice. Maybe some more experienced researchers in this board would have
other (better) opinions.

Rémi



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.