Sunday, March 26, 2006

Behavior predictability and free will

Predictive models of one's own or another's behavior can be developed easily. One mechanism is via life capture video-taping. The degree of repetitiveness of activity, response and language is quickly seen. Biological micro-level differences in blood sugar, hormones, etc. day-to-day or hour-to-hour do not change the bigger picture of being able to predict the next behavior by having seen many past similar situations.

Another mechanism for predicting behavior is knowing the values, beliefs and goals of a person. All three can be elicited and known and are the hierarchy by which we behave. Actions and behaviors may seem irrational (people appear to act irrationally all the time) but are generally fully rational in the context of the person's values, beliefs and goals. Transparency not rationality is of larger import in predicting behavior.

If behavior is fully predictable as argued here, does that mean that there is no free will? In reality, free will is a convenient and helpful construct for humans and it is irrelevant whether it exists or not but useful for humans to believe that it does (and that they have it). There is some flexibility in a narrow band that we label free will where actions can be chosen but also predicted.

Saturday, March 18, 2006

The New Bank: spread lending emerges at Prosper

There is almost more MLMing and loan spread trading at the eBay of loans, Prosper, than regular lending. This is an example of emergence at its best, financial entrepreneurs stepping forward to create lending groups to serve as banks. Here are two examples: Example One, Example Two

Prosper competes with and potentially replaces banks and credit card companies for loans (currently) $25,000 or less and provides a credit marketplace for those below the tier of bank attractiveness and for uncollateralized loans.

Entrepreneurial lenders with high credit ratings are banding together to borrow low on Prosper and lend higher to those in lower credit tiers with poorer ratings. Just as banks have long been spread traders in the money business using their AAA-rated balance sheets to borrow at 4-5% and lend at 8-12%, entrepreneurial banking groups at Prosper are doing the same thing, squeezing the spread to a more efficient purer risk-based premium (yet theoretical risk-based premium). Prosper "banks" are currently borrowing at about 6% and lending at 8-14%, having essentially no costs except fees to Prosper and MLM fees to the bank group leader.

Are the sophisticated speculators playing both sides scaring away genuine erstwhile borrowers?

What about Risk Assessment?
It is amazing that there is so much liquidity available at Prosper. Why are these credit sob stories getting funded? Just because 120 lenders each put up $50 for a $6,000 loan does not make the loan a better idea or improve the chance of default.

Lending to a member of a group is supposed to reduce risk but this is a fallacy, group members are not in close touch and do not co-sign loans as in the micro-lending poster child Grameen Bank model, and group membership seems to be available to anyone who wants it, especially given the group leader incentive for MLM fees on group activity.

It will be very interesting to know default statistics. The risks of lending on Prosper may be substantially understated. There is no recourse. There is no collateral. There is no direct connection to the borrower. However, as junk bonds showed, lower credit tiers are often better risks, sometimes because the lending mechanism they use is their only option.

The biggest question is whether it is safer to loan to a real borrower or a new bank entrepreneur on Prosper!

Sunday, March 05, 2006

Micro-lending marketplace Prosper has $750,000 available to borrow!

Prosper is the eBay of micro-lending, a person-to-person lending marketplace launched in January 2006. Not since junk bonds has there been such a brilliant and potentially impactful new credit product. Deployed in the US (and by UK competitor Zopa), possibly to the chagrin of the $750,000 of lender funds apparently deposited with the site, it is not the romantic notion of the Grameen Bank, sponsoring female entrepreneurs in developing nations, but rather reflects the need to pay off or consolidate credit card debt as the biggest loan category.

Mechanics of the Prosper lending marketplace:
Loans are unsecured and have a maximum amount of $25,000 (typically much lower) and a term of three years. Borrowers post their loans, confidentially specifying their maximum interest rate. If the loan is funded, Prosper takes 1% of the proceeds from the borrower, it is free to list loans. Borrower credit ratings are pulled from Experian and posted on the site (an interesting victory for transparency, self-approved public posting of credit scores). If the borrower defaults, it is reported to credit agencies and Prosper hires a collections agency on behalf of the lender group, but that is all the recourse to the lenders. Lenders must also agree to a credit check and pay a 0.5% annual servicing fee to Prosper on funded loans. Monthly loan payments are electronically withdrawn from the borrowers' bank accounts and deposited into the lenders' bank accounts.

Community is another interesting and important aspect of the person-to-person micro-lending marketplace, affinity groups can borrow and lend amongst members (Porsche drivers, libertarians, etc.). Trust relationships improve repayment and willingness to lend and lenders can fund causes they would like to support. Various flavors of directed and socially-responsible lending movements can easily arise.

Potential criticisms are obvious, the risk of naive lenders extending credit, fraud, and most importantly default per the unfavorable macroeconomics of individual fiscal management. Savings rates continue to be at an all-time low, interests rates and credit card debt averages continue to rise, together with the pressure of cyclic mortgage defaults (particularly on interest-onlys) potentially commencing and the pressure of the November 2005 bankruptcy law revisions.

Still, the person-to-person micro-credit marketplace is an important means of increasing the efficiency of lending (i.e.; decrease bank spreads), allowing a new tier of investors to provide credit, providing credit to traditionally-disenfranchised yet typically low-defaulting borrowers and sponsoring cause-supported lending.

Thursday, March 02, 2006

Universe to hit computation limit in the 22nd c

Ray Kurzweil's "The Singularity is Near" gives play and synthesis to a variety of powerful and provocative ideas at multiple levels.

One of the broadest threads is that the implicit goal of intelligence (either human integrated and uploaded to a digital platform or AI) is to expand itself, meaning to (in the usual sci fi theme) increase its computational capacity by turning as much, and finally all, of the matter and energy in the universe into computing power.

Given Moore's Law and the other familiar exponentiating curves of technology development, Kurzweil estimates that intelligence will reach the full compute capability limit of the matter and energy in the universe sometime in the 22nd century.

The total compute capacity / matter and energy of the human brain is 10^16, of all humans 10^26, of the solar system 10^80 and of the universe 10^94 (although Seth Lloyd has estimated that the maximum capacity of a universe-sized computer would be 10^90 which takes into account the compute capability of all of the particles in the universe). Considering that visible (to humans) matter is only 4% of the universe and that dark energy (73%) and dark matter (23%) could also be used for computronium only raises us to 10^95.

In Seth Lloyd's new book, "Programming the Universe," he suggests that the compute capability of the universe grows over time, however, probably not fast enough to compete with exponentiating curves.

This analysis also assumes that our flavor of intelligence is the only one using the compute resources of the universe.

The solutions are of course obvious, first, as Kurzweil suggests, if possible, to modify the laws of physics (of which we do not yet have a deep understanding), particularly the speed of light and gravity to travel to or create new universes into which intelligence can expand or second, to use black holes for as a more efficient computation medium.

To survive, future intelligence must either modify its demand for computronium or accede to one or more other venues.