eldan.co.uk
a sign that reads: PLEASE DRIVE SAFELY - UNMARKED NUCLEAR WARHEADS TRAVEL THESE ROADS - KEEP YOUR CHILDREN RADIATION FREE

Tuesday, April 27

black is officially white
posted @ 12:32 PM -

there is a lot of weather today

Last week it was so hot that I decided to go and buy a couple of fans this week. This morning I went out and it was snowing.
posted @ 9:58 AM -

Monday, April 26

the scouring of the shire

I had a couple of irritating encounters with the police over the weekend. One officer ordered me off the road, when in fact he has no legal right to do so, and then one almost ran me over because he was talking on his mobile phone while driving around in his cop car. So I was intending to write something about how in the supposed land of the free there is an absurd excess of police exercising arbitrary powers, and how I was coming to appreciate how much freer Britain really is than America.

Then I read this. And suddenly I was glad that round here we 'only' have Ashcroft, not Big Blunkett.

Why can't New Zealand be a major centre of biologically-inspired AI research?
posted @ 8:29 AM -

Sunday, April 25

this phrase may be familiar to a few people: security theatre

posted @ 4:45 PM -

Wednesday, April 7

and they call this the first world?

Number of power cuts I have experienced since moving to America last August (after the big blackout): 5

posted @ 9:05 PM -

1% inspiration

I'm going to put this blog to sleep for a while—probably a little over a month—while I deal with the other 99%
posted @ 2:07 PM -

Monday, April 5

Quote of the day

You know you are at Case when the only thing that goes down on you is the network

This moment of genius courtesy of Cindy.
posted @ 5:40 PM -

hag sameach

Passover starts at sundown today. Come to think of it, it must have already started for all of the Old World. I had intended to write something engaging about it here, and have a nice semi-traditional meal with some friends, at which I told the story my way. However, it's not just because I'm a tool that I spent Saturday night and most of Sunday chained to a computer, so I just haven't had time.

In a vague nod to tradition I might buy some matza and lamb on the way home, but it feels a bit hollow to be doing the outward signs without anything else. In lieu of having the time to write anything sensible here, I will link to the Head Heeb who, like me, is a contrary son and proud.
posted @ 1:57 PM -

Saturday, April 3

A deeply sad observation

At this hour on a Saturday night, I am close to finishing a section of the report for one of my courses. I'm looking forward to getting it finished. All well and good, except that I'm not looking forward to it because I get to stop working, but because I get to switch to the course I like better.
posted @ 8:13 PM -

Friday, April 2

guilty until proven innocent
posted @ 10:46 AM -

Thursday, April 1

The Head Heeb has some interesting musings about Attatürk's legacy around the world.

posted @ 7:17 PM -

showing vs proving

One of the things I find most difficult, out of all the courses I've taken so far, is working out what constitutes a proof in computer science. I flip-flop between writing things that demonstrate a trend but fail to prove that this trend always holds, and just freezing because I demand an unmeetable level of rigour of myself. Really I'm much more comfortable with (and better at) experimental science, where I can go and gather data and say this relationship held up in a large enough sample that it's safe to assume it is general.

Right now I'm working on a group project for Analysis of Algorithms, in which we've been asked to prove a conjecture in a 1986 paper, which no-one has yet been able to prove. We can't either, so unless our group of three are hit by a bolt of inspiration in the next few days we're going to be writing a report summarising the existing work and saying we haven't been able to add to it. Having realised that I favour data over theory so strongly, I've landed myself the job of writing about empirical work, in which people who also couldn't prove this conjecture ran millions of tests to see if it holds up in practice. It does, and it's the strongest case anyone's been able to make for the hypothesis being true, but somehow the theoretical computer science indoctrination I've been receiving has been getting to me: I'm not satisfied with this.

Demonstrating a relationship falls far short of proving that it will always hold. Mike sums up the problem succinctly.

posted @ 6:42 PM -
recent
photos
www.flickr.com
eldan's photos More of eldan's photos
random
photos
www.flickr.com
eldan's photos More of eldan's photos
sponsored links:

Your browser does not comply with current web standards. If you upgrade to a newer browser this page (and much of the Web) will look far better