Sunday, July 29, 2007

I go to America! AMERICA!!!

I'm just leaving to Cogsci2007 to present the results from Jarba's thesis. Very quick trip but I hope to meet my friends Abhijit, Eric Nichols, and Mr Lara-Dammer soon. I wonder what they'll think about what our small group here is doing. Hopefully, some of their own solutions ideas will make it into our projects. Here's the abstract:

We present a new hypothesis concerning cognitive reflection and the relationship between System 1 and System 2, corresponding roughly to intuition and reason. This hypothesis postulates a tighter integration between systems than is implied by the common framework of separate modules. If systems are tightly coupled, as we propose here, an explanation of cognitive reflection may rest in the premature convergence of an ‘entropy’, or ‘temperature’, parameter.

Friday, July 27, 2007

Three stopping points

(i) The ability process ambiguous, confusing, isolated cues into a meaningful, coherent, situation.

This is what FARG already does now. It's what CAPYBLANCA does. But we want to replicate these, using a single framework, at least in NUMBO and Copycat.

(ii) The ability to do it again, now faster

How can the system self-adjust such that, in the future, things get more and more efficient? One idea here that goes beyond FARG is that of (positive or negative) feedback changing future behavior; the other idea is the emergence of preference, in which the IOWA gambling task, or perhaps the N-armed bandit problem can help in addressing.

Here's Grandmaster Nakamura crushing Svenderlunk in speed chess:

(iii) How to (automatically) capture the abstraction behind an idea?

How can this be done? But hey, what the heck can this mean?

♫ ♪ ♪ ♫ I believe the children are our future ♪ ♫

I believe the children are our future

Teach them well and let them lead the way
Show them all the beauty they possess inside
Give them a sense of pride to make it easier


Yesterday I found out, to my utter surprise, that Fabio Evangelho, a former master's student of mine, had just published, alongside Cristian Huse (a PhD candidate at the London School of Economics), a new journal paper:

  • Huse, C., and Evangelho, F., (2007) Investigating business traveller heterogeneity: Low-cost vs full-service users? Transportation Research E 43, pp. 259-268.

Fabio was my first grad student ever to publish in an international journal. Now he does it again with our previous co-author, his longtime friend Cristian. Isn't it incredibly wonderful? The baby is all grown up now!

Congratulations, Fabio and Cristian.

And for those with some musical aesthetics, you might find it deserving to see some better alternatives to this post's title.


Melissa Higgins, Nightminds

Carla Bruni, Tous le Monde

Portishead, Glory box

Tuesday, July 24, 2007

Principles for a Fluid Concepts class library

What are the design guidelines for the fluid concepts class library?

Here are some:

  1. Closed for modification. The framework should, obviously, be closed for modification, once ready. This means that there must be a core set of underlying mechanisms which should not need to change in any other domain. This leads to the second requirement:
  2. Open for extension. The framework, though having a closed, fixed, core, should be extensible, and applicable to new domains. This is obtained through inheritance, composition, loose coupling, the Hollywood principle, and so forth.
  3. Information efficiency. It should be relatively easy to describe, in a high level, the workings of each part of the system. It should be possible for an astute observer to implement on their own the same ideas without having to look at our actual code.
  4. Elegant implementation. This is a hard to define one. The system should be elegant, in the sense of providing all needed functionality, all the while maintaining a sense of simplicity and neatness.
  5. Clean code. The code should be readable and understandable, without need for heavy (or even not-so-heavy) refactoring or heavy commenting.
  6. A test history kept at each step. So that others may follow a step-by-step approach to implementation and be certain they're on the right track, tests for each major class should be preserved, and maintained even in the first executable version of NUMBO.
  7. No globals. There is good evidence that there are no global variables in either the brain or the mind. So alternative approaches to global variables (such as temperature, the obvious candidate) should be strived for.
At one point in time, these posts might start to disappear from the blog, and go to a report, containing both a step-by-step how-to, analysis of the source code, and ideas on how to proceed in new domains. But that remains far in the horizon. For now, work proceeds gradually, for, today, a chunk has been born unto this world.

Information efficiency and design patterns

In the language of design patterns, a FARG architecture is composed of:

  • Codelets and the coderack, which can be implemented through command patterns;
  • Working memory, which are composite patterns;
  • Slipnet, which can implement spreading activation through the observer pattern;
  • Temperature, which may also benefit from using the observer pattern.
This is the basic design being used in this implementation of NUMBO. Perhaps it can lead to some promising things we've never seen before. With a little luck, we just might get some real innovation soon.

Monday, July 23, 2007

Codelets to handle come what may

I've asked:

Can there ever be "general codelets", to handle come what may?
The same codelets being used in NUMBO, COPYCAT, METACAT, CAPYBLANCA, LETTER SPIRIT, PHAEACO, and whatever else coming up from B-town? That might seem wild at first, but I think it's at least plausible.

What should these codelets do?
  • insert an object into a chunk
  • insert a chunk into another chunk
  • retrieve objects and chunks
  • measure temperature differentials of each potential operation
  • deal with system routines, such as updating temperatures, or spreading activation
  • communicate with active symbols (activating them, or being triggered by them)
They should of course be completely decoupled from objects and chunks and temperature and nodes and links and so forth. They should be able to handle requests without precise any information over what they're actually manipulating at each point in time; and, of course, the probabilities of operations and selected objects must be a function of what's going on in the slipnet.

Is there something missing here? I think this is plausible; and we'll soon know, when we attempt to develop COPYCAT from the library built for NUMBO, with no new codelets.

Wouldn't that be awesome?

Intuition, reason, and trees

Everyone is used to decision trees, such as this one. These trees are key underlying mechanisms in many areas of decision-making; a significant part of operations research, game theory, symbolic artificial intelligence and management science.

They are deeply associated with the rational actor, and the with supposition of rationality. At each step, the tree branches out some possible paths. In due course, all possible courses of action are (theoretically) enumerated in the tree, such that a rational actor would maximize an utility function U, over all possible decision courses of action, in order to select the best one.

Consider, for example, the game of chess. On an average position of the game, there will be 30 possible moves. This means that the tree will expand thirtyfold (on average) at each ply one analyzes. Standing at a crossroads with 30 alternatives, all theories in management science, operations research, game theory, computer science, would assume the following:

The Choice Assumption. If there are N choices at a branchpoint, they will be compared against each other.

In this way, the best will be chosen. Beautiful to model mathematically, easy to implement computationally, this choice assumption is a non-brainer in the decision sciences. Moreover, it brings real results; theoretical and industrial. Perhaps this explains why it took quite a while for psychologists to start questioning it. Gerd Gigerenzer was, to my knowledge, the first one to do it, by presenting what he calls the "fast and frugal tree": a tree with N nodes, and N+1 leaves. At each node, the decision is binary, akin to questioning oneself "is this solution satisfactory"? A yes will bring it to a leaf, a no will bring it to a further decision node: "how about this one"?

This is where intuition enters the question. Intuition is the information processing between any two nodes.

Note that the topology of the "intuition tree" is quite distinct (See, for instance, figure 3 of this paper).

Where is the psychological evidence that people face a "take-it-or-leave-it" decision at each point, instead of a comparison of multiple choices? Besides Gigerenzer, Gary Klein's recognition-primed decision model also brings up this model (in conversation with Gigerenzer, he confided to me that he couldn't understand what Klein had been up to; 'but hey, maybe, that's what he means'). Moreover, Barry Schwartz discusses many examples in which the
choice assumption does not seem to hold.

So, if people do not compare options at each point, how is a good course of action selected? Gary Klein, Barry Schwartz, and Gigerenzer converge on the same solution: it's not the best alternative; it's just one that seems Ok. It's not about optimizing, it's about satisficing.

So here are some questions: first, what do the states of this intuition tree look like? How can they explain heuristics and biases? What is the process of moving from one state to another like? Finally, what's the test at each node?

For the last one, a refutable, scientific, answer is ready at hand. In FARG terms: Temperature is down to an acceptable level.

Bloggin' @ tt30

tt30, the young think tank of The Club of Rome has now, thanks to the good work of Andrew Yang, Jordan Macleod, Michael Dorsey, and "chief" Geier, adopted new policies for our baby (but beautifully growing) blog. So I'll be regularly placing some posts in, concerning globalization, technology, science, limits to growth, and whatever else. In any case, I'll keep this blog regularly posted on the what's going on over there.

Friday, July 20, 2007

Don't fight the laws of economics

Paulo Brum hopes that this blog is not turning into a political outlet. He is absolutely right. It shouldn't. Yet, some things need to be said, before moving on:

First, sometimes neutrality means omission. Tobias Lengsfeld, a tt30 & Club of Rome member for whom I have enormous, utmost respect, once explained to me how Switzerland, his nation, was criticized for being neutral during WWII. By being neutral, the Swiss had been helping Adolf Hitler's moves. Sometimes neutrality means omission. As someone who has written, previously to this mess, about our air-transport in a policy book & in the journal of air transport management, I want to state my position. As the finger-pointing circus goes on; there is only one place to point the fingers: president Lula. He is not guilty; but he is responsible for these deaths. He has looked the other way when a government report warned, from ambassador José Viegas, that serious investment was urgently needed; and that the security of the air system would be compromised. The report was ignored. He has backed up his political team through all their pathetic adventures in this crisis. He is not guilty, he is responsible.

Finally, the decision-making biases involved here are worthy of attention. The wrong questions are being asked. People are talking about the danger of Congonhas, the lack of grooving in the track, the rain, the possible problems with the aircraft, and on and on and on. These are the wrong questions, because they do not explain countless other quasi-catastrophes we have been having. It does not explain why another plane has slipped in that same track the day before. It does not explain why, some months ago, two planes crashed their wings, while still on the ground. By sheer luck they did not explode with everyone inside. It does not explain why the Santos Dummont airport in Rio was under fire the same day the plane would crash in São Paulo. It does not explain the huge delays, the traffic controller strikes, the numerous alarms sounding before this second crash.

None of these questions can explain the impossible accident in the Amazon airspace. The territory of the Amazon is larger than Britain, Portugal, Spain, France, Italy, Germany, Switzerland, Belgium, and Greece, all combined! Even if planes fly through established "tracks", a mid-air crash in the middle of nowhere cries out for explanation. The enormous amount of minor incidents previous to this second crash just points out that this was an tragedy waiting to happen. It was no accident. It is the fruit of acts of negligence. This administration created a governance and operating context in which everyone is in control; this, of course, means that nobody is in control. Fingers are being pointed the wrong ways. The man on the 3rd floor of the executive palace deserves all credit.

Unfortunately, realpolitik teaches that this president is staying, despite the few people who call for impeachment. So how could we get out of this mess?

Today the government announced minor tweaks and minor changes of policy. A new airport somewhere in São Paulo, a minor change here, a minor tweak there. They're killing the hub-and-spoke system in Congonhas, which does not alter anything of consequence besides the economic repercussions. Is this supposed to mean that if, instead of going for a connection, that plane would return to Porto Alegre, then it would not have crashed? A plane can crash regardless of its numbers of connections. These questions do not shed light in anything that would ever change the facts that the first plane crashed in the Amazon, the huge flight delays, or that air traffic controllers strikes.

So what lesson can we take?

"Don't fight the laws of economics".

Why is an airport overcrowded and another empty? Because the laws of supply and demand do not hold. Because the airports are not private, and are government-run, their prices remain either the same or closely aligned. So what is the incentive for a passenger (or an airline) to go to a worse-positioned airport? Better positioned airports should be able to charge much more. A ticket from empty airport Galeão in Rio to empty airport Guarulhos in São Paulo costs the same price of a flight from Santos Dummont in Rio to overcrowded Congonhas. This may not seem as central as it is, but it shapes the demand, and that's why one airport is overcrowded while others remain empty. Without the price signal, there is no incentive to move elsewhere.

Since Southwest Airlines's outstanding success, the low-cost low-fare model uses secondary, lower-cost airports. In Brazil, that does not happen. Ask yourself why. Then look at Lula's ideological team.

Instead of killing the hub-and-spoke system, they should let prices signal demand. Basic economic laws should dictate policy. Airports should be privatized, and criminal responsibility should be placed where it rests. We need accountability, incentive systems, justice, governance, and market forces. This government is unable to provide those, and without them we're at our own luck.

Don't fight the laws of economics. We have tried in the past to kill inflation by prohibiting price increases. We failed.

Don't fight the laws of economics. Those people are dead because ideology and political convenience dictate policy. With this, I hope to end writing about this whole catastrophe. At least until the next plane falls.

(Some people are speaking it out)

Don't fight the internet

One of Google's key philosophical ideas is "don't fight the internet". This explains, for instance, their instance on some of these issues, among many others:

  • Offering Google apps for your domain, for free, for as many accounts as you'd want.
  • Buying services such as analytics or writely or blogger or Google earth or feedburner, and offering them for free.
Now, contrast this stance with those companies that try to fight the internet:
  • Myspace, which blocked photobucket users (probably as a Schelling threat maneuver to obtain a better bargain in the sale);
  • Websites such as the Nobel academy, or MIT opencourseware, which make it difficult, or impossible, for one to embed their videos in other sites (I have embedded from both in posts here, but it's hard to do so). Contrast that to the easiness which is embedding videos from youtube or metacafe.
  • The Belgian news outlets, which have actually sued Google for sending them traffic through Google news!
The basic point is: some people and companies simply want to control how their information gets distributed after it is on the network; but that is counter-productive and reflects the thinking from a previous long gone age. If your newspaper is not on Google, other newspapers will get more traffic (and more profits). If your videos are hard to find, people will flock to youtube or elsewhere. And Myspace is seeing users flocking to Facebook, perhaps as a consequence of its stupid decision, among others.

Now here's some crazy news. European governments want to make an European "Google". Not that they want to breed geniuses to make the next killer application. What they want is a top-down, the-central-goverment-controls-everything approach. A company full of politically appointed people; not world-class geeks imported from Bangalore.

First it was French-based Quaero, now German-based Theseus. Of course, as with any government organization, the only real objective turns out, in the long run, to go deeper into the taxpayer's pocket. They stand no chance against Google. It is, indeed, quite third-worldish for the EU to pose as if their pride has been hurt by the power of a 10-year-old company. And, of course, the European projects are already fighting internally:
"THESEUS developed out of the Quaero initiative suggested by a German-French industry study group in April 2005. In the wake of working out the details of the individual research projects, both sides realized that different focuses had evolved. Therefore, at the request of France, the decision was made at the turn of the year 2006/2007, to continue with the two programs separately for the time being."
I'm not yet an EU taxpayer, but I'm a Brazilian one, and what we do here is just as bad: we fight the laws of economics.

Thursday, July 19, 2007

The president responds to the crisis

48 hours after the disaster, president Lula (pictured left) boldly responds to the crisis, quieting those who charge him with omission.

Yesterday, he canceled all items on the agenda; and went though minor plastic surgery.

(Almost) sorry for Mr Gates

I am beta testing Buzzword... and it's incredible! I hope to be developing all my students's masters and PhD theses and papers with this by next year (assuming math formulas as graphics).

These folks will make a well-deserved fortune. Paraphrasing Richard Dawkins, I'm almost felling a little sorry for Microsoft; but only almost. And only a little.

Just check this out.

Wednesday, July 18, 2007

Преступление и наказание

Преступление и наказание, things they've never heard of in Brasilia.

In 2005, I published a chapter called "Por uma nova logística nacional" in this book. Easy money... I wrote the plain obvious: the federal government is broke. Everything needs to be privatized, starting by the airports, in order to keep things going.

Also in 2005, a study I conducted with Fabio, a former masters student, and Cristian, a London Business School Economist, came out in the Journal of Air Transport Management, downloadable here. The whole planet, except Brazil and perhaps China, has seen deregulation, decentralization, privatization, market forces take over the soaring demand for air travel.

But over here in our jungles, our weak, populist, demagogue, president, will always respond with ideology. Never decide by looking at the sheer plain naked data. His omission and lack of leadership have created an incredible circus in which authorities from ANAC, INFRAERO, the ministry of defense, the airlines, the air traffic controllers, among many many others, can't speak the same language. We've had, 10 months ago, an impossible accident, a Hollywood-style mid-air crash in the middle of the Amazon; something of unbelievable, epic, proportions. Now, after months of an intensifying crisis and countless small incidents, the obvious happened. Around 200 people are dead, in the heart of São Paulo.

The news today was pathetic. All authorities have denied problems. The laws of physics were claimed not to hold, as an official, with a straight face, explained to reporters that it was "impossible for a plane to slip over the landing track"--an event that had happened the day before the tragedy.

In the end, the dead will be at fault. The pilots will be claimed to be responsible. The pilots can't defend themselves now, can they? "So sad, it was human error", they'll say. Over here, nobody is ever held responsible for anything.

Brazil stands before a decision. Either the president falls, or more planes will.

So here's some hoping for Преступление и наказание.

Time for impeachment proceedings

"Relaxa e goza?"

The time for impeachment proceedings has come. Brazil cannot stand any further without a president. It is time for this drunken sailor to go.

As people burned to their deaths this very night, Brazil stands at a crossroads; are we going to keep on faking democracy? The president's incompetent inaction is responsible for this awaiting tragedy; yet, impeachment should be applied not for his incompetence, but for the high crimes committed, and admitted. The law must apply.

It is very sad, but realpolitik dictates that the law will only apply after a disgrace like this one.

After the unprecedented corruption scandals broke out in 2005, all around the president's inner circle, I talked to a high-ranking official from the opposition: "The law must apply here". The response I obtained, to my astonishment, was: "ok.. hmmm... but... but this vice-president... he's horrible". That, of course, was 2005, and the opposition was underestimating the president's ability to deceit, to lie, to pose as the savior, the messiah.

It is time for impeachment proceedings. Of course it won't happen. But it should. The vice-president is obviously horrible; yet, the law must stand above opinion and convenience. Brazilian democracy is not working. The law must apply.

Brazilian democracy is not working. This is of serious, sobering, consequence.

Sunday, July 15, 2007

The Club of Rome & the coming global bargain

I've been invited to speak about how my work relates to negotiation. After a back-and-forth stride between a more mathematical, game-theoretical, presentation, and a look at practical issues that challenge us all, I've decided to introduce The Club of Rome, Limits to Growth, the coming global bargain, and what we've been doing these last years.

Imagine you are the owner of a large resort, with a beautiful lake. However, danger lurks in the horizon. Imagine you have a lillipad problem, with those growing at an exponential rate, doubling as each day passes.

When would you find out that it was a serious problem?

Consider three cases. If you found out whenever the lilipads reached 50% of the lake, you would have a single day to respond before they covered the entire lake. If you found at 25%, you would have only two days. If you found at 12.5%, three days. And if you found at 6.25%--c'mon, this is hypothetical, and you don't need to lie. Nobody would find it at 6,25%; but even if they did, that's only four days to respond!

This is the immense problem with exponential growth. We've been having exponential growth in numerous areas, such as population, economic activity, pollution, and so forth. How long do we have until a serious limit is reached? China's economy, for instance, has grown at over 12% this year alone. This means that they are, in practice, doubling their economic activity each 6 to 7 years. Thus, you might want to call the exponential explosion in industrial output circa 2000 as "China" (see slide 11 below). I mentioned also that China is moving cities around like FGV would make minor redecorations. Below is Ed Burtynsky's TED prize video, which I wish I had had time to show in my presentation. Sobering stuff.

Here are the slides.

Friday, July 13, 2007

The shape of things to come

How to avoid the handcoding problem? When does semantic memory touch episodic memory? How do semantic networks evolve, reorganize, grow, and learn over time? Can a node compute its own correlation to others as it receives their signals, without global association mechanisms? How do new nodes "pop"out? How do abstract chunks come to life to enable analogy-making? Can there ever be composite command codelets? Can there ever be "general codelets", to handle come what may? How do feedback systems and the credit assigment problem fit into this picture? Finally, what does the N-armed bandit problem or the IOWA gambling task have to do with all of this?

Oh, but I digress...

Thursday, July 12, 2007

When is this thing going to end?

From the pages of The Economist:

STRATEGY lies on the high ground of business management, the zone where long-term plans are made and far-sighted visionaries set their companies on course to untold riches. Books on the subject establish their authors' intellectual reputation, for strategy is full of paradoxes and dilemmas that challenge the brain to find seven simple steps to their resolution.
"seven simple steps to their resolution"; is such hilariously beautiful mockery of the whole ordeal warranted?

I mean, isn't it ridiculous that these books are still being churned out by the dozen? The Economist was reviewing books by publishers such as Currency and Harvard Business School Press. It is absurd that people in charge--not only in America but everywhere--keep on reading this type of nonrefutable, nonscientific nonsense, mixing causation with correlation, backed only by in hindsight anecdotal evidence (and under a most convenient interpretation). Authors of these things get notoriety, some get some money or tenure, university research funds go to these frauds, as these self-help cooking recipes get spread around the globe. It gets so impregnated in Businesspeople that they can't see there's serious stuff to absorb out there. Some time ago I had a discussion with a student or two concerning what I see as relevant in the field of decision making. And since it did not include these seven magic steps to "untold riches", I guess that student wasn't too pleased--but here's a student who's supposed to write a master thesis. All I could say was the obvious: "I think the Swedish academy is on to something relevant--though, of course, you may not agree"; or "Science is much too important to be left to scientists alone".

We barely understand
how humans can play chess. To get explanations with seven magic steps about systems as complex as corporations in a mere 208 pages is a fraud as large as intelligent design. Intelligent design, however, is always on the spotlight. These business book frauds, however, are passed on and on as serious works, and are often conducted on serious universities. And they'll eventually influence people in charge. What a sad, sad, ordeal.

Isn't it refreshing to see The Economist bringing it on to them?

Wednesday, July 11, 2007

Here's to some overcoming bias

Mahalanobis reports:

The WSJ reports for eight years John Mackey, CEO of Whole Foods, used a pseudonym to talk smack about how great his company was:

Rahodeb began posting messages about Whole Foods shares on in the late 1990s. He quickly gained a reputation as being one of the stock's biggest cheerleaders, and gamely defended himself when other posters chastised him for being too rosy. "I've never pretended to be anything but enthusiastic about WFMI," he wrote in 2000, using Whole Foods' stock symbol. "I admit to my bias -- I love the company and I'm in for the long haul. I shop at Whole Foods. I own a great deal of its stock. I'm aligned with the mission and values of the company... Is there something wrong with this?"

Rahodeb often sparred with other users, deploying a rigorous analysis of financial statements. "Your quarterly cash flow variance isn't statistically meaningful because the time period is too short," he complained to another user who had criticized Whole Foods in March 2006. He then pasted a summary of the previous six years of Whole Foods' operating cash flow. "Over the past 5 years operating cash flow has increased 330%," Rahodeb noted.

Mahalanobis is Ok with it, and so am I. In fact, I think this could be a very promising way for someone to avoid overconfidence, groupthink, appeals to authority and confirmation bias, status quo bias, among other serious biases. If a CEO has regularly to step down and be a nobody and argue for his stock prices using publicly available data, all these biases vanish. The guy obtains crucial, incisive criticism. If only the folks at the White House (or, for that matter, Brasília), were to try it.

Sunday, July 8, 2007

The Club of Rome: looking back

It's now official. I've just received a letter from the Club of Rome's interim president, Ashok Khosla, and Secretary-General Uwe Möller, formalizing my associate membership in the organization. I join the six current Brazilian members of the organization: Prof. Heitor Gurgulinode Souza, Prof. Paulo Alcantara Gomes, Dr. José Aristodemo Pinotti, Former President Fernando Henrique Cardoso, Prof. Eda Barbosa, and Ms. Edna Roland.

Perhaps it's time to take a little look back, and thank those for whom I am so grateful, and a look forward, and think about how I can contribute. As Thomas Schelling might put it, this strategic commitment will most probably make it more likely that I will, in the face of strong adversity (i.e., procrastination), be able to comply. But before some plans for the future, some words of thanks. And, as a student of Hofstadter, I couldn't make it in other way than by throwing in some memories.

I guess the whole thing could be traced back to some months after the tragic events in New York and Washington, under the Japanese Government's 2002 Global Youth Exchange program in Tokyo, discussing international security and the threat of terrorism. I remember getting immersed in the literature of international relations, the history of Israel, the nature of the Middle East conflict, a great deal of Thomas Schelling, and also, of course, the enormously difficult question for a legal definition of terrorism (was the attempt to kill Adolf Hitler by bombing that office a terrorist attack? If not, why not?). Two years afterwards, I'd be joining tt30, the young think tank of the Club of Rome, for a meeting in Jordan, followed, a month later, to an invitation to Helsinki, for the 2004 Assembly General. Some guys over there--more specifically, Tobias, Joerg, and Ildiko, got me on a corner and asked: next year we're going to Brazil; can you pull it off? To which I must have sounded quite like Forrest Gump: "Ahmm... Ok!"

As I got back to Brazil, former president Fernando Henrique Cardoso agreed to meet me for a discussion on the potential event. He'd back it and join us, but it was still up to us to organize and, of course, obtain sponsors. Prof. Gurgulino, which had been planning to launch a Brazilian Chapter of the Club, put the pedal to the metal in the whole effort.

I'd like to thank our sponsors and supporters: Embratel, FGV, Fiesp, Computer Associates, Souza Cruz, GOL Airlines, UNIBANCO, Radisson, Dannemann, Siemsen, Bigler, & Ipanema Moreira, IESB, Conjuntura Econômica, CEBRI, the government of the state of Rio de Janeiro and the secretary of culture, and UNESCO.

I'd like to thank Sebastian, all my tt30 friends--whom I'll not even attempt to list them all here, & the people at FIESP: Fabio Cervone, Directors Carlos Cavalcanti and Thomas Zanotto, Amanda Pinto, and President Paulo Skaf.

In an even inner, more involved circle, I have always had enormous support from Profs. Bianor and Deborah at FGV. Others that were there from day one, for the good and for the bad days, were Tobias Lengsfeld, Carla Winter Afonso, and CoR Secretary-General Uwe Möller.

Yet, in the end, three incredible people made it happen. I'll be forever indebted to these three people. Three people that made all the difference in the world during the days of storm. The first one is Claudia Santiago de Abreu, who pulled it all off. She knows all secrets, and holds the keys to all doors, and that's not by god-given powers; it's all by her unbounded merit. Yet, she operates on stealth mode, and I don't think I have a single picture next to her.

Joerg Geier was my European counterpart in arranging all the details, and had to handle organization of the tt30 internal meeting schedule almost all by himself. At one point, I remember sorting my emails by recipient, then by sender, to count hundreds and hundreds of messages between us over the course of some few months. In a typical day, we'd probably exchange some half a dozen messages. In a bad day, 25 messages and skype and phone calls could top it off.

Prof. Gurgulino was, and is, the driving force behind Brazilian presence in the Club. Embedded here is the image that to me, marks the year 2005: with Prof. Gurgulino, somewhere, planning, counting costs, studying possibilities, deciding on the large and small, over long term strategies, and all kinds of minutiae.

Before I lay down some thoughts on what I'd like to offer the Club of Rome; all I can say is that, up to this day; what an incredible privilege it has been.

Thursday, July 5, 2007

Ultimatum games: The Testosterone Monologues

This week's The Economist is reporting on Burnham's study relating the ultimatum game and testosterone levels:

Dr Burnham's research budget ran to a bunch of $40 games. When there are many rounds in the ultimatum game, players learn to split the money more or less equally. But Dr Burnham was interested in a game of only one round. In this game, which the players knew in advance was final and could thus not affect future outcomes, proposers could choose only between offering the other player $25 (ie, more than half the total) or $5. Responders could accept or reject the offer as usual. Those results recorded, Dr Burnham took saliva samples from all the students and compared the testosterone levels assessed from those samples with decisions made in the one-round game.

As he describes in the Proceedings of the Royal Society, the responders who rejected a low final offer had an average testosterone level more than 50% higher than the average of those who accepted. Five of the seven men with the highest testosterone levels in the study rejected a $5 ultimate offer but only one of the 19 others made the same decision.

What Dr Burnham's result supports is a much deeper rejection of the tenets of classical economics than one based on a slight mis-evolution of negotiating skills. It backs the idea that what people really strive for is relative rather than absolute prosperity. They would rather accept less themselves than see a rival get ahead. That is likely to be particularly true in individuals with high testosterone levels, since that hormone is correlated with social dominance in many species.

Here lies a major point that connects FARG architectures (active symbols), economists's misperceptions (throwing the baby with the bathwater--or, more technically, failing to perceive that other people's perception of a model may go way beyond the model), linguists (connotations), and artificial intelliegence researchers (in the Eliza effect). Here's hoping that Anne's work will be a significant step in this arena.

On the economics of adultery

Free exchange brings forth this economic puzzle:

The Adultery Puzzle

Why hasn't the lemons problem killed adultery? To be more specific, why would any women want to steal a man who lies to, cheats on, and then dumps his wife? This is particularly clear in Woody Allen's Match Point - the mistress angrily insists that her boyfriend leave his wife, even though he's shown her in a hundred ways that he's a lying, cheating parasite.

In the actual market for used cars, of course, the markets has largely solved the lemons problems using reputation, inspection, and warrantees. You don't want to sell low-quality products if it will ruin your firm's reputation, if they have to pass inspection first, or if a dissatisfied customer can return the product and get his money back. But it's hard to see that mistresses can rely on any of these mechanisms. Few adulterers build up a reputation for standing by their mistresses. Most adulterers wouldn't pass inspection. And I've never heard of an adulterer giving a credible money-back guarantee ("If I don't leave my wife within a year, you get a full year of your life back!"). So what's the point of stealing another woman's man, if you can only steal the bad ones?

The adultery puzzle doesn't seem like a case of a lemon market to me; perhaps it's a false analogy. There's asymmetry of information between buyer and seller in a lemons market, but how much asymmetry is there between the adulterous male and the family breaker? Not too much. First, the adulterous guy has already committed to some extent to her, since, in most cases, she could have acquired enough information to break up the marriage. Moreover, does he really know what he is up to? In many of these cases people may find themselves in the Jekyll and Hyde situation, like, as Schelling put it, the smoker that destroys the last pack of cigarettes in the middle of the night, then goes insanely craving for more, then throws the car keys on the dark garden so he can't buy some more at least during the night, only to get his flashlights the next minute.

Perhaps many adulterous folk out there are constantly going back and forth between their luscious dreams and thoughts of the kids and the house and lifelong commitments. When this is the case, the adulterous guy may not even know what is going to happen; how things will turn out in the end; for his own control of the situation is in doubt, just like the addicted drug user can't really tell if he'll recover. For cases like this, there should not be asymmetry of information between "buyer and seller", thus it's not really a lemon case.

When it is indeed a lemon case (i.e., the guy clearly has no intention of leaving, and is under perfect control of his feelings and actions), my guess would be that there would also be large asymmetry of resources, and that some naughty girl out there is getting little gifts every single week. The guy is a lemon, but turns out to "tell her" that fact, by subtly and constantly bribing her way into silence (these presents have, in fact, been documented in studies--though I've never seen a study relating asymmetry of resources and lemon-ness--however that might be measurable). In this case, again, there's no asymmetry of information. Even the real lemons do not apply to the "lemon problem", as asymmetry of information is not a key driving factor behind the transaction.

Finally, being "a lying, cheating parasite" is not an absolute all-or-none proposition, but really a function of one's current state of affairs. They guy that lies and cheats to girl X might actually turn out to be not a monster in another relationship--and here's a true motive for the "family breaker". Carl Sagan and Ann Druyann, according to his biographers, have lived a Cinderella story after both left their original marriages. Perhaps these stories fuel some girls to go after committed man.

I wonder why Free Exchange didn't mention the fact that women will also philander.

A bug's life

Eric Nichols says:

One of the things I've been thinking about is determining some way to tell if everything is going to decay to 0, or whether some loop of activation can form, or whether things can explode and get super-activated and excited, like in a runaway nuclear reaction. I think you can get any of these types of behaviors, depending on the configuration of nodes and the parameters for your spreading

I think the people in neural net research must have some sort of theory that describes these conditions for us. I wonder if there is a way to set reasonable parameter values (or bounds on parameters) to, for example, prevent any runaway reaction from ever occurring? Alex -- how did you set parameters for your activation decay model?

Tricky, tricky issue. I haven't been able to find any theory or solid model from neural network people. Melanie Mitchell, back in '93, was writing about careful parameter tuning in Copycat to prevent this:

The current version of the program always uses the intrinsic link length rather than the shrunk link length for this [spreading activation] calculation, even when the label node for this link is active. [...] When shrunk link lengths were used for spreading activation, the network tended to become too active. It is possible that a different mechanism (e.g., some kind of inhibition technique) should be used to control activation in the network. This is a topic for future work on Copycat. (p. 254 of "Analogy-making as perception")

In my current implementation of NUMBO, I'm using two inhibition techniques, and I'm having, essentially, what is a closed system. The first "closed-system" technique is that a node does not spread its current_level of activation, but really the received increase from the last round. So if A sends x activation to B, B will receive x times a drag (inversely proportional to distance), and might as well propagate it back to A. But instead of propagating the whole activation of B, only the increase will be leaving for the road. A stone falls in a calm pool. The waters propagate at large intensity, waves reach the border, then re-propagate with smaller intensity, and so on.

The second closed-system idea is that a node spreads its activation to neighbors, but divided by the number of neighbors it is spreading activation to. So, neighbors really receive very small amounts of activation, unless it's coming from a boring node with few neighbors. I do, however, have the counterpart function: an amplification function. If a node is receiving activation from more than one node, than this amplification function will multiply that activation. At this point it will become sum(activation_received) * (1.1^sum(nodes_sending)). The rationale behind this? We can call it the nonlinearity of curiosity.

Yet, despite all these inhibition mechanisms, the slipnet is still prone to become crazily excited. Reminds me of Jenny Curran while in Berkeley, and of course others, these runaway nuclear reactions.

Worst lesbian movie of the decade

Notes on a scandal, most definitely. Yet, still worthwhile.