Sunday, January 28, 2007


Last Worker

is here.

Updated link: Last Worker - Holonation

Thursday, January 25, 2007


Beale Street Blues

updated link

Thursday, January 11, 2007


Firefox 3 and OpenID

I first found out from Andy Payne and it's also being rejoiced elsewhere around the net. It seems Web 2.0 will arrive with Firefox 3.

Wednesday, January 10, 2007


Identifiers as a Data Compression Problem

Ashley and Madame X's comments about identity persistence got me to thinking about Bob Wyman's recent post which proposes an enhancement to Zooko's Triangle for naming. The elegance of Zooko's idea that you can only achieve 2 out of the three design goals (uniqueness, memorable, decentralized) has been praised by many in the identity conversation.

A triangle that expresses the tradeoff that one of three must always be sacrificed has been around for a long time in the data compression community:

1. uniqueness ==> quality
2. memorable ==>low bitrate
3. decentralized ==> computational cost (this relation requires more imagination than the first two: scalability)

I remember Jerry Gibson sharing with us the data compression tradeoff model shortly before the Internet boom.

Tuesday, January 09, 2007


Open Source Deep Blue

Make Magazine is calling for everyone to collectively request access from IBM, an Open Source advocate, to the Deep Blue source code:
About 9 years ago the humans lost the chess battle to Deep Blue "Game Over: Kasparov and the Machine" is definitely worth seeing if you're interested in that sort of epic...

IBM is really active in the open source community, perhaps we could collectively request access to the Deep Blue source to not only see how it beat our best human chess player at the time, but to run our own versions of Deep Blue (it could run on a modern computer for sure by now). It might also clear up a lot of questions on how exactly IBM beat Kasparov too.
I reviewed the documentary here.

Saturday, January 06, 2007


Austin OpenID Mashpit

In Episode #47 of the Story of Digital Identity (seek to the 30 minute mark), Chris Messina says
What I would love to see at these events is a semi-coordinated effort to actually build and implement demo-applications that start to actually tease out the faults in the system.
In Episode #47, the question was posed more than once "Is OpenID ready for E-Commerce?" Maybe the answer is "Who cares?"

In preparation for the mashpit at Cafe Caffeine, I've spent the past week meeting up with security developers around town who have been asking really good questions. Although Chris is especially interested in user experience issues, these conversations have been more related to web services built on top of the OpenID libraries. Those conversations and the sharing of some cool projects will continue at the mashpit.


Wednesday, January 03, 2007


...the i-names thread

If you're a character in Before Sunrise, you've just spent a magically romantic day in Vienna with someone you just met on a train. Instead of exchanging phone numbers, you agree to meet at a rendezvous spot six months from this day. This isn't like meeting someone in town for coffee. This is a test of commitment. This is also a prisoner's dillema. If the other person doesn't show, besides the loss of what could have been, you get to eat an expensive plane ticket.

What does this have to do with i-names?

Phone numbers and email addresses and URLs are unique but they're not as easy to remember as Jesse and Celine. I can imagine a dystopian future where our characters would introduce themselves to each other as =texasjessek and =greencelinefr. Even if they never consciously exchanged information, they could easily hook up again. The penalty outcome has been substantially reduced: If either Jesse or Celine doesn't show up, he or she can be hunted down by the loser of prisoner's dillema game.

, Kaliya presented more value propositions for i-names. Her post was thought provoking. For instance, it has allowed me to consider how many people might prefer an i-name. However, for the time being, none of the cases or reasons made me any more enthusiastic about i-names as a whole.

Kaliya included a snippet from Phil Windley's blog. The main point was that if the different types of addresses (blog, voice, rss, etc) his i-name pointed to changed, people could always find his new addresses through his i-name ( which stays relatively static.

The problem with this is that XRI, as it is similar to DNS, seems to introduce another single point of failure through a limited set of root nodes. That doesn't seem like a good thing.

This page is powered by Blogger. Isn't yours?