My New Years’ resolution: to get a real job
Get a leg up on the competition, and offer me a tenure-track position in computer science right now! Here’s everything you’ll need to decide:
In your offer letter, make sure to specify starting salary, teaching load, and the number of dimensions you’d like spacetime to have.
(Note to Luboš: Unfortunately, I wasn’t planning to apply to the Harvard physics department this year. But if you make a really convincing pitch, I might just be persuaded…)
Comment #1 January 7th, 2007 at 10:24 pm
I thought Motl had to buy your services, not offer you a tenure track position…
Comment #2 January 7th, 2007 at 10:39 pm
The RS is a great read. However, the Dawkins-based conjecture seems to be saying something like “any optimization algorithm can be simulated efficiently by a local search heuristic”. Sounds highly dubious to me…
Oracles lie. Not as often as they tell the truth, but close enough.
Comment #3 January 8th, 2007 at 3:44 am
Cheshire Cat: Yes, I purposefully tried to come up with a conjecture so strong that it might even be false! But can you think of a counterexample? I couldn’t…
(Of course, the conjecture would be stated in such a way that exhaustive search, pruning, branch-and-bound, etc. would not be counterexamples.)
Comment #4 January 8th, 2007 at 4:36 am
I think it would be more of a problem if the conjecture were true for a trivial reason. It’s a reasonable hypothesis that PLS-complete problems take time 2^{\Omega(n)}, and outputting the “evolutionary sequence” could be done in time 2^{O(n)} just by exhaustive search, so the running times are polynomially related. But this is misleading – the optimal algorithm for the problem might not bear any real relation to the algorithm outputting the evolutionary sequence.
I think a good conjecture (even if it were false) would exclude such an eventuality.
Comment #5 January 8th, 2007 at 4:55 am
The key point is that the algorithm should be outputting an evolutionary history as it goes along — i.e., that it shouldn’t take more than ~T*poly(n) steps to output the Tth element of the sequence. Exhaustive search doesn’t satisfy that property. Sorry if I didn’t explain that more clearly.
Comment #6 January 8th, 2007 at 6:56 am
Can’t offer you a job, unfortunately, but a big *amen* to your teaching statement. If enough profs followed this, I would have majored in Comp Sci. Sadly, it seems that these are slender minority views in most places.
Comment #7 January 11th, 2007 at 11:49 pm
Scott, how much longer will you be at Waterloo? Please teach another graduate course before you leave =)
Comment #8 January 12th, 2007 at 2:29 am
Scott, how much longer will you be at Waterloo?
Dunno — at least till the summer! Unfortunately, I’ll be away for too much of the spring semester to teach a course this term.
Comment #9 January 12th, 2007 at 11:08 am
Scott, your honest and courageous trial of applying so unconventionally is really a wonderful fuel for better working ethics and standards. Hope it is properly taken by your prospective employers.
Comment #10 January 12th, 2007 at 2:34 pm
Nagesh, if you keep writing comments like that, people are going to suspect you’re me under a pseudonym. 🙂
Comment #11 January 12th, 2007 at 4:42 pm
Oh:) I feel nice about your response on my comment! I believe you are not serious about the suspicion part:)
Comment #12 January 14th, 2007 at 2:37 am
If you won’t apply to Harvard, you can try Columbia.
Comment #13 January 14th, 2007 at 3:53 am
I would apply to the computer science departments at both Harvard and Columbia, but apparently neither of them are doing a search this year.
Comment #14 January 17th, 2007 at 4:34 pm
Hi,
You may find the following advert interesting:
http://web.comlab.ox.ac.uk/oucl/jobs/ulqcs.html