Debian Goodies, and the Mysterious Case of Disappearing Human Knowledge
Since I've added the debian/experimental apt source, I've been upgrading in bits and pieces to get especially nice packages.
Today's upgrades are:
apt 0.6.x with cryptographic signing and verification
OpenSSH 3.9 that can reuse the same connection with -M
Sadly, I can't install anything else because the standard debs aren't signed and I haven't learned enough about the debsign-verify policies to figure out how to use unsigned debs. I probably should have read the directions first.
I mentioned the Cool Tool website a few days. While browsing around there today I found El Wire. I was instantly inspired to mount powerful magnets on the brake mounts of my unicycle, and then attach the electroluminescent wire to my spokes and coils of wire to the rim of my unicycle. That would create a simple electrical generator such that each spoke would light up as it passes the magnets. My significant other has been encouraging me to increase my visibility when unicycling at night, this would be an amazingly tacky way to do it.
I randomly ran across this marvelous writing Who Can Name the Bigger Number? by Scott Aaronson. (makes me wonder if Kolmogorov complexity and the Busy Beaver are opposites, for some value of opposite)
This article is lovely, fascinating, and makes me want to jump right into the writings of the mathematicians Ackermann and Kleene.
But I have no idea how to get copies of papers published before 1990 or so. This seems to be part of a larger problem. Older papers are not available electronically, and papers formerly available online are slowly disappearing.
For example, John Meacham was asking on #haskell for a paper on efficient graph reduction strategies for his JHC compiler. The paper had been on a user page on chalmers.se, but the website was gone. Happily John found the papers via archive.org's Wayback Machine.
It seems to me that our discovered knowledge is disappearing offline. We the human society need a distributed, decentralized, and most importantly well-indexed way to preserve and find the knowledge we gather.
Right now the best option I see is something based on peer to peer networks. I would suggest using a simple, easily implemented and emulated system. That way, no matter what the paradigm shifts in the future this system can be upgraded, and the data replicated to the new systems transparently.
Recent bizarre type system thought... I wonder if the equivalent of QuickCheck tests can be encoded into Epigram's type system?
Goron recently asked how I have time to keep up with so much stuff and still be self-employed. Part of that is reading the right RSS feeds, so, I've included my RSS feeds in opml format.
If I'm missing any RSS feeds on the subject of Haskell, functional programming, type theory, etc, please tell me!