Update (with a plunge into the goop)
Looks like this thesis saga should wrap up today (Monday). It's basically just down to some paperwork and giving the school more money. Next up I have a massive backlog of grading to do. Then I have a report to do for the "shitty" class. At some point I also want to put together my new computer. After that, I have the old contract work I've been putting off that needs to be done by July. Then there's the ol' job hunt.
At some point I'd like to play around with a couple of artificial intelligence projects that could possibly-maybe form the basis of a Ph.D., should I decide to go that direction.
One project deals with first-order logic (and is half the reason I took the "shitty" class this semester). The basic idea is to get a computer to figure out the optimal strategy for a board game without ever playing or witnessing a single match. To my knowledge, no one has ever attempted (or rather, succeeded at doing) such a thing. I have some encouraging ideas, but still feel my chances of success are very slim. There are two obstacles to overcome: 1) shoehorning what I want to do into the theorem-proving paradigm, or figuring out a workable alternative to that paradigm; and 2) keeping the size of the strategy manageable. The rewards are outstanding, though. Board games are just a convenient and fun testing ground--the idea could be applied to more general problems believed to be of great difficulty.
The other project deals with genetic algorithms (and possibly neural networks). This is more of a survey type of project, although I make a few minor contributions. As such, I'm not sure it would even qualify as a dissertation. I figure I could probably keep adding stuff until it qualified, though. However, I don't think I'd be very proud of the result*; the work would be solid and most likely fun, but the results would be neither groundbreaking nor surprising. It wouldn't make much of a name for me. On the other hand, the odds of success are very good.
So, you could say idea 1 is a grueling journey to an awe-inspiring destination. Idea 2 is an enjoyable journey to a disappointing destination. Another way of looking at it is as investments: Idea 1 is the classic high-risk, high-reward investment; idea 2 is the classic low-risk, low-reward investment. I don't really like the investment metaphor, though, because this isn't an either-or decision--I can try both.
* ...much as I'm not very proud of my thesis. Don't get me wrong; I'm proud that I finished it, and I'm proud of the elegance of my solution and that I was able to come up with a proof for it. However, my contribution was to a very particular domain that maybe 10 people in the world have ever heard of, and even within that domain, my contribution borders on incidental. Most importantly, it's not something I, myself, can use; I basically did something for someone else.
At some point I'd like to play around with a couple of artificial intelligence projects that could possibly-maybe form the basis of a Ph.D., should I decide to go that direction.
One project deals with first-order logic (and is half the reason I took the "shitty" class this semester). The basic idea is to get a computer to figure out the optimal strategy for a board game without ever playing or witnessing a single match. To my knowledge, no one has ever attempted (or rather, succeeded at doing) such a thing. I have some encouraging ideas, but still feel my chances of success are very slim. There are two obstacles to overcome: 1) shoehorning what I want to do into the theorem-proving paradigm, or figuring out a workable alternative to that paradigm; and 2) keeping the size of the strategy manageable. The rewards are outstanding, though. Board games are just a convenient and fun testing ground--the idea could be applied to more general problems believed to be of great difficulty.
The other project deals with genetic algorithms (and possibly neural networks). This is more of a survey type of project, although I make a few minor contributions. As such, I'm not sure it would even qualify as a dissertation. I figure I could probably keep adding stuff until it qualified, though. However, I don't think I'd be very proud of the result*; the work would be solid and most likely fun, but the results would be neither groundbreaking nor surprising. It wouldn't make much of a name for me. On the other hand, the odds of success are very good.
So, you could say idea 1 is a grueling journey to an awe-inspiring destination. Idea 2 is an enjoyable journey to a disappointing destination. Another way of looking at it is as investments: Idea 1 is the classic high-risk, high-reward investment; idea 2 is the classic low-risk, low-reward investment. I don't really like the investment metaphor, though, because this isn't an either-or decision--I can try both.
* ...much as I'm not very proud of my thesis. Don't get me wrong; I'm proud that I finished it, and I'm proud of the elegance of my solution and that I was able to come up with a proof for it. However, my contribution was to a very particular domain that maybe 10 people in the world have ever heard of, and even within that domain, my contribution borders on incidental. Most importantly, it's not something I, myself, can use; I basically did something for someone else.