I recently accepted an offer from Ksplice, inc. to join them starting this summer. This means I will be leaving the M.Eng program at MIT and not continuing work towards my Master's degree.

This was a tough decision for me, but in the end I'm really happy with it. Ksplice was founded last year by four SIPB members that I've known since I was a freshman, and worked on a number of projects with as a student. I know them all well and am really excited to be working on their project with them. They have an awesome technology, and I think they have a great chance of taking over the world with it.

They're located in Cambridge, so I'll still be around here next year. I feel like I should have more to say, but I guess that's it for now.


I finished The Audacity of Hope yesterday. I was on the whole impressed and inspired, and will probably vote for him, but I'm still not really happy about it.

Not because I don't believe in anything in particular he said, but because I don't actually have any evidence (from the book alone; I don't follow current events basically at all) about whether or not he's actually capable of pushing through any of the kinds of things he wants to do.

And because while I was impressed, I have nothing to compare him to -- I know ~nothing about any of the other candidates, so he gets compared against some weird politician stereotype and/or Bush, neither of which is particularly instructive.

I don't care enough to put in much effort into finding out more, though, because it really doesn't matter what I vote for.

Also, I really feel like all this wanking about the details of the issues and the specifics of anyone's voting records are missing the point. I have _no interest_ in micromanaging my candidates' choices and national policy ... I want to elect someone I basically trust and who broadly agrees with me on things that are important to me, and then trust them to do something reasonable. That's the whole reason we have a representative democracy instead of a direct democracy.

You know you're a geek when...

Somehow the topic of two-column geometry proofs came up in conversation last night. I remember when we did two-column proofs for a short while in high school. Everyone else found them either hopelessly tedious or utterly inscrutable, but I found them comfortingly familiar. You see, I'd spent the previous year learning z80 assembly for my Ti-83+, and found two-column proofs to operate at a similar level of abstraction -- and be similarly formatted to boot.

Lessons of the week

Nelson's lessons of the week:
  1. Don't drop laptops. They don't like it.

    [On boot]
    /etc/init.d/rcS: line 50: 3096 Segmentation Fault $i start grep: /proc/filesystems: No such file or directory * udev requires tmpfs support, not started.
    [Later, after booting from a LiveCD...]
    $ mount /dev/hda3 /mnt/hda3 Floating point exception $ cat /mnt/hda1/etc/fstab cat: Segmentation Fault
  2. Make backups. Luckily, I have full backups from <24h before the disk ate itself, so I lose practically no data. And I'm still under extended warranty.
More posts probably coming soon.


If I do use livejournal, it would make me very much happier to post from emacs. ljupdate.el seems significantly less broken than some of the older packages I've tried for blogger and WordPress, so this just might work.