Sparklines, Very Cool we are lucky to have Dr. Tufte. I did something similar once with histories of build failures. Good things happen when you take Tufte’s principle of minimal ink combined with large amounts of data. I really love the way he inlines these charts right into the formated text; encouraging faster left/right brain switch!

Two things got me interested in power-law distributions. The first was back in the early 1980s. I was involved in a big software engineering project and the system we had built was horribly slow. Collected lots of statistics about it’s behavior; calling patterns, data structure topology, paging behavior, etc. etc. These all came out power-law distributed and I became very disconcerted that I didn’t know anything about this distribution.

So some time later when people started pointing with alarm that the distribution of wealth was growing less and less equitable. Again the power-law distribution! Hm. The system designer in me found that fascinating. Could it be that my software systems shared something fundamental with the economic system?

I haven’t written about the distribution of wealth much recently. The Fortune 500 survey of wealthy dudes is out for 2004. We make scatter plot the 400 wealthly dudes. That’s the first chart. (There are gaps in the data shown because my scrapping script isn’t perfect and there is only so much time I’m willing to spend working on it.) That chart doesn’t do justice to the long tail, since it doesn’t show the other 300 million dudes in the country.

We can plot that data on a log-log graph to show that it’s power-law. More fun is to plot all the data for the last few years and look for trends. The second chart does that; along with fitted lines for each of the years. Presuming that sampling the top of the curve is a good estimate for the rest of the wealth distribution you can calculate your rank by plugging you net worth as Y into the 2004 formula and solving for X. How far out on the tail are you?

The third chart show the trend. The number I label wealth is a measure of the size of the economic pie. You can see the pie swell up during the bubble, pop and then try vainly to recover. The number I labeled equality is a measure of the severity of the distribution of wealth.

I really love being unemployed.

Yesterday was great.

I spent the morning with a geek friend. We talked about distributed build systems (like CPAN, or Freebsd’s ports – Where is the survey article on these?), the build recipes, the contracts for the ingredients, the naming of components. “Mix in one library of PNG kind conforming to API spec PNG std Foo minus these routines (which by the way nobody ever implements right) but with this variation (which by the way everybody does because, well it’s useless without that).” We talked about algorithms for finding clusters in graphs; marvelous computational intense algorithms. We talked about diffuse peer to peer networks for distributed assertions injected by anonymous untrusted agents around the perimeter. We talked about how Php makes it so easy to create your first web page and how Python makes it so hard. What fun!

Then it was off to a room at Harvard with expensive chairs, wood paneling, matures trees russled outside the windows to a “conversation” about (take a deep breath): “From Personal to Impersonal Trusted Exchange in Physical and Digital Domains An Evolutionary Perspective.”

There, Kevin McCabe, told three fun stories. Here’s the first one.

Experiment. Subjects are paired up, call them Alice and Bob to participate in a game, but they remain totally anonymous. Alice and Bob get 10$ for showing up for the experiment. Alice is offered an option. She can take portion of her 10$ and stuff it in an envelope to send to Bob. But wait! The amount she puts in the envelope will be tripled before it get’s to Bob. Alice is told one more thing. Bob will have the chance to send a portion of the money he receives back to her.

Later Bob gets an envelope from Alice, maybe with some money it it. He too is presented with an option; the option of what to put in the envelope that goes back to Alice. He keeps the rest.

One way to look at this experiment – what’s it cost the experimenter? The minimum cost to the experimenter is 20$ the maximum is 40$. After the experiment Alice and Bob have split the 20$ to 40$ dollars. Lots of words got thrown about: Alice is optimistic/pessimistic or trusts/distrusts Bob. Bob is rational. Bob reciprocates. Apparently if Bob is a graduate student in economics Alice is much more likely to get an empty envelope back. The rest of us are apparently more optimistic about our fellow man more likely reciprocate acts of trust.

There as a lot of discussion about eBay.

The sprit of the stairwell would like me to post these questions:

• What is the comparative value of these various kinds of reputation: eBay – 1000 feedbacks 97.8% positive; Apache Software Foundation membership; tenure at Harvard.
• Is reputation fungible?
• Who owns reputation; the individual or the institution?
• Can the institution sell it, i.e. can eBay sell my reputation data?
• If economists know that trade barriers are bad should we strive to make reputation more fungible
• Reputation exchange rates? Reputation Euro?

It’s fun being unemployed, but I doubt this is a sustainable business model.

Ha! Recipe:

Ingredients

• An old powerbook with a mostly broken screen and a very broken hinge (priceless).
• A long wire to the stereo (10$)
• A short wire the printer.
• Ethernet cable to the household hub.
• VNC.(free)
• Web Remote, lets you control iTunes over the web. (10$)

Instructions

• Plug in power adaptor
• Install OS10 from scratch, wiping disk clean.
• Plug into Ethernet
• Run short cable to printer
• Run long cable to stereo, considering selling CD player.
• Install VNC. So we can work on machine without leaving our chair.
• Enable print sharing, so we have a print server.
• Enable file sharing
• Copy all the music (mp3 etc.) into the household onto this machine.
• Load mp3 into iTunes
• Via iTunes preferences enable iTunes sharing, so we have a tune servers.
• Install Web Remote; have it start up at reboot.

When you done you can play your iTune’s library on the stereo; just like you’d bought an Airport Express. If the old powerbook has a airport card you can also let is act as a wireless base station.

I gather some people can get their powerbooks to keep running even when the lid is closed. That didn’t work, dependably, for me. (Did you know there is a magnet on the lid of most of the powerbooks; aligned with the mousepad?)

You can print from OS9 to an OS10 printer if you create on the OS9 machine a desktop printer that uses the right lpr server and queue.

I couldn’t get Darwin to run on an olde iMac; maybe later. I was doing something analogous on a FreeBSD box; but this was really easy.