Sunday, October 28, 2012

Framed out Hypatia artwork

I finally got around to mounting and framing-out the artwork that my friend Hollyana Melear made of the story of Hypatia that is mounted over the fountain on the back porch.

Monday, October 8, 2012

Installing Shibboleth

My friend Ken has worked for a week getting the authentication system "Shibboleth" to work with appsoma.  It was a nightmare.  He's written a comprehensive guide here for those who will follow.

Tuesday, July 3, 2012

Virus illustration

This is a virus illustration I commissioned from Beverly Garland for a story I've been writing.  Love it!

Saturday, June 2, 2012

Make Rating Agencies into Insurance Agencies

I just watched this video interview between Paul Krugman & Charles Stross.  An interesting issue that Krugman brought up concerned how the change in Intellectual Property (i.e. the increasing difficulty of protecting one's IP) changed the rating agencies' monetization strategies and ended up contributing to the financial crisis. Specifically Krugman argues that because no one was buying their books anymore the rating agencies switched to a monetization strategy where the issuer of the security paid for the rating.  The obvious conflict of interest (and the shopping around by the issuer for a better recommendation) caused junk securities to get rated as triple A.  Krugman bemoans the question of how this is every going to get fixed.  I suggest that the fix is fairly obvious and centuries old: insurance.

Suppose that the ratings agencies were also security insurance agencies.  As a customer buying a security you would buy an insurance policy from the agency who also rated the risk of that security.  Under that scenario, the insurance / rating agency has every incentive in the world to judge the value of security assets accurately (and refuse to insure incomprehensible derivatives for example) as they would stand to lose considerable money if they judge the security incorrectly and make maximal profits by calling it right.

The question is, how might one convert the financial industry -- and by this I don't mean Wall Street, but rather collection of all fund managers around the world -- to buy these insurance policies.

One: The insurance product has to exist.  If the rating agencies see themselves going out of business then it's time that they start converting themselves into insurance companies or insurance companies need to get into the ratings business.  Market opportunity!

Two: Regulation / Information.  Start bottom-up.  If fund holders -- for example, retirement account holders (i.e. everyone) -- got a big fat health warning on their yearly statements that said something like: "WARNING: Your retirement savings are un-insured.  The Finance General of the United States suggests you consider moving your funds to an insured policy.  Talk to your account manager about your options." Then suddenly you'd find a lot of pressure to create these policies.  One might also consider a top-down approach where funds over a certain size or containing more than X% of retirement funds are required to carry insurance.

Monday, May 21, 2012

Cancer: The call of the wild

The other day someone asked me: "What is cancer exactly?"  I responded: "It's the call of the wild."

Cells in your multi-cellular body are in a very odd, almost "un-natural" state -- they aren't reproducing as fast as they could.  In fact, sometimes they even kill themselves.  From an evolutionary perspective, this is very, very odd.

We multi-cellular creatures are both new and rare in the grand scheme of things.  Almost all life on Earth is and was single-celled and, leaving aside the complexities of living in colonies, the typical cell simply makes copies of itself as fast as it possibly can limited only by its ability to acquire food.  Holding back or, even more radically, killing oneself is completely anathema to a cell.  If cells were politicians, "reproduce as fast as possible" would be their only platform.

The evolution of multi-cellularity required that collections of cells work together and this implies that some of them give up the right to reproduce in favor of others.  That is, a random cell in your body, say on your finger, is giving up the right to produce descendants but by doing so it "hopes" that it is increasing the chance that its nearly identical cousin cell (your sperm or egg) will reproduce.  As implied above, that is no small request of a cell.  Everything in a cell's history prepares it to, indeed demands, that it make copies of itself as fast as possible.  Yet, in multicellular creatures such as we, all the cells except a lucky few are being asked to halt that directive.  In a sense they have signed a contract: "I forgo the right to reproduce so that my nearly identical cousin will survive."  The vast majority of the time the contract is upheld.  Until it isn't.  And when cells don't respect the contract anymore, we call that cancer.

In other words, cancer is the call of the wild.  A cancerous cell is reverting back to what its ancestors have always done -- reproduce.  A cancerous cell is, in effect, saying: "To hell with the contract, I'm reverting to what worked for my ancestors.  It worked for them it will work for me." Of course, they're wrong, but they don't know that, they're just "stupid" cells after all.

It's a miracle that we strange multi-cellular creatures are here at all -- multicellularity requires the most incredible yet tenuous compromise in the history of the world.

Tuesday, May 15, 2012

In this week's New Yorker, Joan Acocella has a nice piece summarizing the competing philosophies of language analysis known as the "prescriptivists" versus the "descriptivists".  The otherwise fine article is glaringly lacking an important piece of analysis: Google.

Language is a form of collective memory.  We use a word or a grammatical form because it *communicates* which implies some sort of an agreed upon form, i.e. memory, between transmitter and receiver.  The state of a language is encoded in the distribution of usage throughout the population.

It used to be the case that there was only one of two ways to measure this collective memory.  Either, one, you listened to a sampling of the language around you or, two, you used a reference source which usually claimed to have done this sampling for you.  Both measurements are prejudiced.  To a large degree, the argument between the prescriptivists and the descriptivists falls along these measurement lines.  Either you listen to what's around you which is biased by the circles you inhabit (which you admit as a descriptivist) or you believe in a reference which is biased by the writer's opinions (which you may or may not admit as a prescriptivist).

But now there is a third way.  You Google.  Not sure how to spell something?  You type in your pathetic phonetic attempt and Google politely asks you if you meant to spell it differently.  There's no arguing prescriptivist or descriptivist with Google -- Google is impartially telling you what the collective memory has to say about this.  Google is *both* prescriptivist and descriptivist.  It is both prescribing a usage and describing the mode of the sample.

Google of course also has biases --  toward the young, wealthy, and techy.  But compared to one's everyday selection bias or the bias of a dictionary (especially the dictionary's temporal bias), it is the paragon of impartiality.  I argue that from now on Google and its inevitable successors will define the language much more than the dictionary did and in a way that is de facto fair.

Anyway, to consider the issue of modern language usage as this New Yorker article does without considering how Google shapes the modern landscape seems to me to be exceedingly antiquated.

Monday, March 26, 2012

BBQ finished

A the BBQ is now complete with the addition of a nice green granite counter-top. :-)

Sunday, February 26, 2012

BBQ brickwork done and cleaned.

Now all that's left is a countertop. :-)

Saturday, February 25, 2012

BBQ area - almost complete

Worked for the last few days finishing up the backyard BBQ area. It's only taken my ~4 years! The grill and the doors still have their protective coatings on them. They are actually stainless steel. The firepit area is will be fill with crushed glass and a ring of fire will rise out of it.

Tuesday, January 31, 2012

Probability Theory Lectures by Bill Press

Last year at sat in on my friend Bill Press' computational statistics class. It was great. This year he has recorded it and posted it online. If you're into comp. stat., I *HIGHLY* recommend this series of lectures and this course page.


Thursday, January 26, 2012

Goldsworthy inspired algorithmic art

I watched this documentary about Andy Goldsworthy today and was inspired to create the following in