Friday, October 17, 2008

Security in Voting Systems

Security in Voting Systems, Ron Rivest

Opening humor from the talk:

When I die, I want to be buried in Louisiana so I can remain active in politics.
-- attributed to Earl Long

Election official's mantra: let it be a landslide.

There are three kinds of election officials: those who know how to count and those who can't.

This is an interesting topic because of the trade-offs, constraints, and social-dynamics involved. Technically, three goals were outlined--to ensure that votes are
  • cast as intended
  • collected as intended
  • counted as intended
The constraint that makes this challenging is that you can't give the voter a receipt that would allow proof of how he/she voted. This is so votes cannot be sold or voters coerced into voting a certain way.

There was a side discussion on election methods, and in particular the preferential voting that Cambridge uses for city council elections. Some problems with this are that votes from the different wards cannot be tallied separately and then aggregated--you have to collect all the votes and apply the rules across all votes. Not only does this make counting take longer, it also makes acceptance sampling difficult if not impossible.

Ballot stuffing and chain voting are ways to game voting. With ballot stuffing election officials wait until end of day and cast votes as people who don't show up. With chain voting, one obtains a blank ballot, fills it out then finds prospective voters who will cast that ballot and return with a blank ballot in exchange for cash. One idea to counter ballot stuffing would be to publicly publish the names of people who voted. But then you have to weigh that with the practice of paying people not to vote, because with the public list there is a way to prove performance.

Twin and Scantegrity II are some end-to-end solutions, the latter being the more serious one. The basic idea of Twin is to put your paper receipt into a mixer and then randomly draw someone else's receipt and be able to verify that against some public record. The basic idea of Scantegrity is to set up optical scan voting with the modification that instead of filling in bubbles with ink you use invisible ink or scratch-off to reveal a confirmation code. The confirmation codes are randomly linked per ballot to voting choices. The codes can be published and verified, while maintaining the privacy of the vote and preventing vote selling or coercion. If a vote is mis-counted, and a voter protests, then having an unrevealed confirmation code that was actually on the ballot gives credibility to the protest.

Sunday, October 05, 2008

Big Ideas from the Iliad

Just finished reading The Iliad

An outline of the events is here. Some of the big ideas:

  • Timê - acts of excellence in sport, battle, or council that get official acknowledgement through awarding of prizes. Achilles gets Briseis as a prize for battle excellence and loses it when she is arbitrarily taken away. The funeral games also illustrate acts of excellence in sport.
  • Kleos - what others hear. Glory and lasting fame as a result of Timê. Examples abound. Hector decides to part with his wife to win renown for his father and himself. The death of Teucer. Sarpedon in book 12, says let us go win glory for ourselves. etc. etc. Achilles has to choose between kleos aphthiton and nostos--a short life with permanent glory or a homecoming and long domestic family life.

  • Making Amends - (Is there a greek word for this one?) There are many examples of taking an action that angers someone and then doing something to make up for it. Agamemnon and Odysseus in book 4. Antilochus repairs the situation with Menelaus after the chariot race in book 22. Achilles is famously unable to repair the situation with Agamemnon to everyone's detriment.

  • Moirae - The fate of Hector, Achilles, Aeneaus, and Patroclus are well forshadowed and predetermined by the fates. Gods don't interfere with fate even if it conflicts with their personal allegiances. For example Poseidon who sides with the Greeks endorses saving Aenaeaus from the hands of Achilles.

  • Impermanence -- The wall the greeks build is a symbol of this and it's impermanence is highlighted by the conversation between Zeus and Poseidon in book 7. Kleos seen as a way to overcome this. Another angle to the impermanence theme is that blessings come and go. For example Greeks and Trojans go through phases where it would seem gods are helping everything go well for them and then things turn for the worse and other gods are making things go wrong. Priam talks about this in book 24 describing Zeus as having two urns, one with evil gifts and one with good ones.

Thursday, October 02, 2008

Population of Boston

What's the population of Boston? I grabbed some numbers from wikipedia to get a sense of the relative size of Boston compared to some cities with which I'm familiar. The data is from different years so this is just a rough comparision. It's interesting that while Boston is much smaller than other U.S. cities it is more densely populated. My brother-in-law lives in Shanghai so I was curious how that compared because my impression is that Shanghai should be huger than huge. Well, it is, with about 31 times the number of people, but for any given square mile you would find half the number of people you would in Boston--can that be right?

Year Rank City          Population  Metro     Density
2006 23   Boston           590,763  5,977,504  12,327
2000 113  Grand Rapids     197,800  1,323,095   4,434
2007 4    Houston        2,208,190  5,628,101   3,828
2007 11   Detroit          916,952  4,467,592   6,856
2000 214  Ann Arbor        114,024    341,847   4,221
2007 3    Chicago        2,836,658  9,785,747  12,649
2006 2    Los Angeles    3,849,378 17,755,322   8,205
2007 1    New York       8,274,527 19,750,000  27,147
2007 14   San Francisco    764,976  7,264,887  16,380
2007      Shanghai      18,450,000              6,790

normalized to Boston as 1:

Year    Rank    City           Pop.    Metro Density
2000    113     Grand Rapids    0.33    0.22    0.36
2007    4       Houston         3.74    0.94    0.31
2007    11      Detroit         1.55    0.75    0.56
2000    214     Ann Arbor       0.19    0.06    0.34
2007    3       Chicago         4.80    1.64    1.03
2006    2       Los Angeles     6.52    2.97    0.67
2007    1       New York       14.01    3.30    2.20
2007    14      San Francisco   1.29    1.22    1.33
2007            Shanghai       31.23    3.09    0.55

Tuesday, September 30, 2008

Clojure

Rich Hickey gave a talk to about 60 Boston Lisp Users.

The overview of his new programming language Clojure:

  • it's a Lisp dialect
  • it's built on the Java Virtual Machine
  • not only are the java libraries available, his claim is (and a small example showed) that clojure turns java programs into lisp, not lisp programs into Java
  • it's a "functional" language with a software transactional memory system that uses multiversion concurrency control
  • instead of getting faster cpu's the trend is that we're getting more cores
  • garbage collection and gc for jvm's are state of the art, ephemeral collection on the jvm is very fast, and we now live in a world where virtual machines are the platforms that a lot of business expects software to run on.
  • nice sequences interface
  • it was clear that Hickey had given a lot of thought to the language design choices. Not only an experienced lisp programmer, he talked about trade offs made by other languages such as Scala and Haskell. He addressed several problems we've had to live with as users of common lisp and scheme. Separating symbols from vars is one welcome example of that.

Sunday, September 28, 2008

Building Robust Systems

Building Robust Systems, Gerry Sussman's talk at the Broad Institute

The talk was based on his paper Building Robust Systems. Sussman on requiring proofs of program correctness:

I am not arguing against proofs. They are wonderful when available. Indeed, they are essential for critical system components, such as garbage collectors (or ribosomes!). However, even for safety-critical systems, such as autopilots, the restriction of applicability to situations for which the system is provably correct as specified may actually contribute to unnecessary failure. Indeed, we want an autopilot to make a good-faith attempt to safely fly an airplane that is damaged in a way not anticipated by the designer!
If you cannot prove that the program is correct why would you put any trust into the safety of something like an autopilot program? Sussman's emphatic answer was that you test it and that there's not an aircraft in use that has been proven safe--we know they are safe because of test pilots. The added point was that if you know anything about floating point numbers you won't trust anything with a floating point number in it, and that anything designed on a computer had better have a lot of real world testing.

Wednesday, August 06, 2008

Book 13 of the Iliad

This book is very tedious. It contains many battle details that range from Teucer felling Imbrius like an ash tree to Meriones hitting Harpalion in the right buttock. The blow for blow description of battle in this book in itself isn't that enjoyable to me. But having gone through the effort to read it all, I've become more invested in the story and I've gain an appreciation of of the battle when looked at from a distance. And that is what we get to do at the start of book 14 with this contrasted setting:

NESTOR was sitting over his wine, but the cry of battle did not
escape him, and he said to the son of Aesculapius, "What, noble
Machaon, is the meaning of all this? The shouts of men fighting
by our ships grow stronger and stronger; stay here, therefore,
and sit over your wine, while fair Hecamede heats you a bath and
washes the clotted blood from off you. I will go at once to the
look-out station and see what it is all about."

Nestor's tent is a nice comfortable place to be, and at one point the shouts of battle are at a safe distance. But they grow louder. And from the tedious spearing and beheading that went on in book 13 it's possible to really appreciate the kind of raging storm that is approaching. This scene echos the theme of impermanence. It is another way of saying don't get too comfortable with the present state of things, it won't last.

Wednesday, July 30, 2008

Some HDR images using software available on Linux. This is the Cambridge Water Works building next to the fresh pond.



I used the auto bracket feature on my camera. It takes three shots: one with normal exposure, one over exposed, and one under exposed. The images can then be merged into an HDR format like openexr. Then the HDR image is tone mapped and saved as a jpeg.

Here are two of the fresh pond,



Tuesday, July 29, 2008

Notes on Book 12 of the Iliad here.

Favorite excerpt,

He broke both hinges, and the stone fell inside by reason of its great weight. The portals re-echoed with the sound, the bars held no longer, and the doors flew open, one one way, and the other the other, through the force of the blow. Then brave Hector leaped inside with a face as dark as that of flying night. The gleaming bronze flashed fiercely about his body and he had two spears in his hand. None but a god could have withstood him as he flung himself into the gateway, and his eyes glared like fire. Then he turned round towards the Trojans and called on them to scale the wall, and they did as he bade them--some of them at once climbing over the wall, while others passed through the gates. The Danaans then fled panic-stricken towards their ships, and all was uproar and confusion.

This is such a great image. There's got to be a painting somewhere with this as the subject. If you know of one, please let me know! Sadly, searching google images for "hector two spears" comes back with several images of someone else.

Thursday, July 24, 2008

Book 11 of the Iliad has a lot of trash talk.

For example Paris hits Diomed in the foot with an arrow,

Thereon Paris with a hearty laugh sprang forward from his hiding-place, and taunted him saying, "You are wounded--my arrow has not been shot in vain; would that it had hit you in the belly and killed you, for thus the Trojans, who fear you as goats fear a lion, would have had a truce from evil."

Diomed all undaunted answered, "Archer, you who without your bow are nothing, slanderer and seducer, if you were to be tried in single combat fighting in full armour, your bow and your arrows would serve you in little stead. Vain is your boast in that you have scratched the sole of my foot. I care no more than if a girl or some silly boy had hit me. A worthless coward can inflict but a light wound; when I wound a man though I but graze his skin it is another matter, for my weapon will lay him low. His wife will tear her cheeks for grief and his children will be fatherless: there will he rot, reddening the earth with his blood, and vultures, not women, will gather round him."

Wednesday, July 23, 2008

Favorite sentence from Book 10 of the Iliad

The Trojans aren't far away from the wall that protects the Achaean ships and Agamemnon is tearing his hair out,

When he looked upon the plain of Troy he marvelled at the many watchfires burning in front of Ilius, and at the sound of pipes and flutes and of the hum of men, but when presently he turned towards the ships and hosts of the Achaeans, he tore his hair by handfuls before Jove on high, and groaned aloud for the very disquietness of his soul.

And this one isn't a favorite, but stands out as another dose of Iliad brutality,
On this Dolon would have caught him by the beard to beseech him
further, but Diomed struck him in the middle of his neck with his
sword and cut through both sinews so that his head fell rolling
in the dust while he was yet speaking.

Friday, July 18, 2008

Notes on Book 9 of The Iliad here

This book is the dramatic embassy to Achilles. Favorite sentence,

He may offer me ten or even twenty times what he has now done, nay--not though it be all that he has in the world, both now or ever shall have; he may promise me the wealth of Orchomenus or of Egyptian Thebes, which is the richest city in the whole world, for it has a hundred gates through each of which two hundred men may drive at once with their chariots and horses; he may offer me gifts as the sands of the sea or the dust of the plain in multitude, but even so he shall not move me till I have been revenged in full for the bitter wrong he has done me.

because it shows just how bitter Achilles is.

Over on livingepic the embassy to Achilles is mentioned in The mysterious dual: the smoking gun of epic interactivity. I posted this question,

This is somewhat of a tangent, but what would you say the sense was for the original audiences when considering what Achilles should do in response to Agamemnon's embassy? Was Achilles viewed at all favorably for his actions? Or was he seen as petty?

Roger Travis's response:

My own view is that the bard who first improvised the withdrawal of Achilles from battle must have done it in order to ask the very same questions--that is, is it ever OK to desert? is it OK to desert over a slavegirl? The luxury of stories and of games (or maybe of the PPP) is that the bard and his audience, and the developer and his gamers, can ask the question without having to find an answer right away, or ever.

You'll find Homerists all over the map on this question, though, from some who support Achilles completely to others who think that even though Agamemnon's a jerk the epic is trying to persuade you that Achilles could never be justified.

I actually don't find it hard to imagine a game getting to this level of meaning, perhaps along the lines of Star Wars: Knights of the Old Republic and Mass Effect: the rudimentary light/dark systems in those games seem to me to begin to give the player the ability to ask what feels justified and what feels unjustified in certain kinds of situations.

Tuesday, July 15, 2008

Notes on Book 8 of the Iliad here

Favorite sentence,

They then offered unblemished hecatombs to the immortals, and the wind carried the sweet savour of sacrifice to heaven--but the blessed gods partook not thereof, for they bitterly hated Ilius with Priam and Priam's people.

because it shows the tension of having Jove completely on the side of the Trojans with the consequence of pissing off the other gods. What's a better situation to be in? To have the favor of the strongest god and the resentment of the others, or to have the favor of several lesser gods with strongest being, for the most part, neutral?

Sunday, July 13, 2008

Notes on Book 7 of the Iliad here

Favorite sentence, Jove to Neptune:

Surely when the Achaeans have gone home with their ships, you can shatter their wall and fling it into the sea; you can cover the beach with sand again, and the great wall of the Achaeans will then be utterly effaced.

because it makes the interesting point that although humans strive for kleos, they can never accomplish anything permanent.

Saturday, July 12, 2008

Notes on Book 6 of the Iliad here

Favorite Book 6 quote:

And Hector answered, "Wife, I too have thought upon all this, but
with what face should I look upon the Trojans, men or women, if I
shirked battle like a coward? I cannot do so: I know nothing save
to fight bravely in the forefront of the Trojan host and win
renown alike for my father and myself.
because it is a clear reference to the idea of kleos, and references how the son is responsible for building on the kleos of his father.

Thursday, July 10, 2008

I've been calling it honor ...

... but the idea is Kleos,

Kleos (Greek: κλέος) is the Greek word often translated to "renown", or "glory". It is related to the word "to hear" and carries the implied meaning of "what others hear about you". A Greek hero earns Kleos through accomplishing great deeds, often through his own death.

Wednesday, July 09, 2008

thwarthwimple on The Iliad and honor.

I see the Iliad as being primarily about power. The honor theme is a part of the power theme. To put it in terms of French and Raven in 1959, honor is the path to referent power, or the ability to attract others and build loyalty.

I think the Iliad is painting a picture of a set of attributes that can lead to glory, honor, and ultimately power. It also shows how things can go bad when the power is handled poorly--usually due to not possessing the complete set of attributes.

These attributes are

  1. strength and courage as a warrior
  2. ability to communicate in counsel
  3. ability to see beyond one's self-interest and manage conflicting interests

Mars has merit as a warrior, fights on both sides (for the fun of it?), fails to be sensitive to the conflicting interests, and gets wounded by Diomed via Minerva, and censured by Jove.

Venus is pitiful as a warrior, causes all the conflict by spurring on the Helen and Alexandrus affair, gets stabbed by Diomed and laughed at by Minerva and Juno.

Juno and Minerva are formidable warriors as we see in book 5. They are also very skilled in their ability to manage their conflict of interest with Jove who could squash them at will if they annoy him too much.

Alexandrus stole another man's wife, was chickened out of a fight, and despised by all as a result.

Achilles is the ultimate warrior, but the impression I get is that other characters do not really like or respect him beyond this one dimension. He doesn't see beyond his own immediate self-interest and allows many an Achaean to die by not entering the battle even after Agamemnon tries to make amends. livingepic makes an interesting point on this matter:

In Book 9 as we have it, bad old Agamemnon sends to the tent of Achilles three “ambassadors” in what’s been called forever after the “Embassy to Achilles”: Ajax [strongest of the Achaeans], Odysseus [smartest of the Acheans], and Phoenix [old friend of Achilles]. When they get to Achilles’ tent, each of those ambassadors gives a speech about why Achilles should accept Agamemnon’s offer and come back to the fighting. It makes sense, and it builds to a very nice crescendo in the speech of Phoenix, after which, from an ethical point of view, the audience is in a lot of suspense about what Achilles should do, despite being in no suspense at all over what he will do.

One could argue that Achilles is at fault for the death of his own dear Patroclus due to his choice not to accept Agamemnon's offer.

Agamemnon seems to communicate well, manages relationships well--he seems skilled at making amends when he offends--and he's not too shabby a warrior. He's pretty balanced and above average but not perfect in all the attributes that can earn him honor.

The Thersites/Ulysses incident in book two is an example of how not to speak well in counsel and manage conflicts.

And finally Hector is the ideal. He manages the lameness of his brothers, and his army is composed mostly of allies from outside of Ilius. He's respectful of the gods. He sees beyond his short-term self-interest when he explains to his wife why he will return to battle. He's a courageous, powerful warrior.


Tuesday, July 08, 2008

Notes on Book 5 of the Iliad here

Favorite sentence:

Mars roared as loudly as nine or ten thousand men in the thick of a
fight, and the Achaeans and Trojans were struck with panic, so
terrible was the cry he raised.
A thematic sentence to pay attention to:
They that shun dishonour more often live than get killed, but they that fly save neither life nor name.
Another favorite--Diomed wounding Venus:
The point tore through the ambrosial robe which the Graces had woven for her, and pierced the skin between her wrist and the palm of her hand, so that the immortal blood, or ichor, that flows in the veins of the blessed gods, came pouring from the wound; for the gods do not eat bread nor drink wine, hence they have no blood such as ours, and are immortal.

Wednesday, July 02, 2008

Notes on Book 4 of the Iliad here

Favorite sentence, Agamemnon to Ulysses after calling him slack:

"Enough; I will make you amends for what I have said, and if any ill has now been spoken may the gods bring it to nothing."

Tuesday, July 01, 2008

Notes on Book 3 of the Iliad here.

Favorite line,

When she marked the beautiful neck of the goddess, her lovely bosom, and sparkling eyes, she marvelled at her and said, "Goddess, why do you thus beguile me?"
Guess who the goddess is. And this line,
If Alexandrus kills Menelaus, let him keep Helen and all her wealth, while we sail home with our ships; but if Menelaus kills Alexandrus, let the Trojans give back Helen.
makes me think of this line from Love and Death,
Sergeant: If they kill more Russians, they win. If we kill
more Frenchmen, we win.
Boris Grushenko: What do we win?

Friday, June 20, 2008

A wonderful screen adaptation of the Iliad.

It is a must-see. The score is exquisite. The directing brings out the multiple layers of motives and conflict. The performances of Hector and Ajax are unmatched. And the special effects--well, you have to see them to get the full experience: here.

Thursday, June 19, 2008

More notes on the Iliad.

Book 2 notes here. Favorite sentence from book 2,

There are no longer divided counsels among the gods; Juno has brought them over to her own mind, and woe betides the Trojans at the hands of Jove.
Another good one,
The chiefs disposed their men this way and that before the fight began, drafting them out as easily as goatherds draft their flocks when they have got mixed while feeding; and among them went King Agamemnon, with a head and face like Jove the lord of thunder, a waist like Mars, and a chest like that of Neptune.

Wednesday, June 18, 2008

notes and favorite exerpts from Book 1 of the Iliad

Favorite sentence from Book 1,

As he spoke the son of Saturn bowed his dark brows, and the ambrosial locks swayed on his immortal head, till vast Olympus reeled.

Thursday, May 08, 2008

Running a red light on a bike can get you a ticket in Cambridge. A couple of relevant links if you commute by bike here:

Sunday, April 27, 2008

The Gorilla King on PBS.

This documentary is about a gorilla named Titus, who researchers have studied some 30+ years. Titus is a silverback, age 30+, and it is rare for a gorilla to be this old and still be in charge of things. What typically happens is that at some point another silverback will wage a violent take-over. The new silverback in charge will then kill all of the babies born by the previous leader. When another silverback, Kuryama, started to become a threat, Titus's move was to take the whole group of gorillas on a long journey to the top of a mountain, where it was cold, uncomfortable, and the air was thin. The gorilla's eventually had to huddle together to stay warm as it became colder and colder. Finally Kuryama stood up and began walking back down the mountain. Half of the group followed him. Those loyal to Titus remained. Then Titus went down the other side of the mountain into the Congo.

Sunday, April 20, 2008

Recent librivox recordings listened to:

From The Mice in Council:
Once upon a time all the Mice met together in Council, and discussed the best means of securing themselves against the attacks of the cat. After several suggestions had been debated, a Mouse of some standing and experience got up and said, "I think I have hit upon a plan which will ensure our safety in the future, provided you approve and carry it out. It is that we should fasten a bell round the neck of our enemy the cat, which will by its tinkling warn us of her approach." This proposal was warmly applauded, and it had been already decided to adopt it, when an old Mouse got upon his feet and said, "I agree with you all that the plan before us is an admirable one: but may I ask who is going to bell the cat?"

Tuesday, March 25, 2008

Tthe selection of video podcasts is blooming.
Recently watched:

Thursday, February 21, 2008


Here's a picture of the moon prior to the eclipse. This was taken from my back porch at about 9:30pm EST last night.

Saturday, February 02, 2008

At work last week David Karger gave a talk about how end users shouldn't need to be database administrators to manage their own personal data. He touched on the good and bad of the semantic web, outlined some incremental goals, and described the follow-on to Haystack, Exhibit.

The nice thing about Exhibit is that it can spread in a way that "home" pages spread in the 90's. You copied someone's html file, replaced their name with yours, and you had a home page. Exhibit is a way to make rich interfaces to data that is nearly as easy. To try it out I looked up some tax data for New Hampshire municipalities and packaged it up in an exhibit of New Hampshire Property Tax Rates showing which one's have high rates and which low.

Monday, January 07, 2008

I've been thinking more about an assertion from Database in Depth: Relational Theory for Practitioners, "null should be prohibited."

Date argues that the three valued logic arising from null values allows incorrect queries. He shows this with an example where the 3VL breaks down. Consider a query restriction of (A <> B) or (B <> C) with A=London, B=Unknown and C=Paris. It doesn't result in true; it has a value of Unknown. However, if B=Paris then it has a value of true, and if B<>Paris it also has a value of true.

But what would it be like to work with a data model that cannot contain null? Take a contact database as an example. Let's suppose we want to store the following values (primary key in brackets):

CONTACT: [ID], FIRST, LAST,EMAIL,PHONE,BDAY

But we don't know everyone's email, so if we want real relations we have to do

CONTACT: [ID], FIRST, LAST,EMAIL,PHONE,BDAY
CEMAIL:[ID],EMAIL

And we also can't guarantee we will know everyone's phone number or birthday.

CONTACT: [ID], FIRST, LAST
CEMAIL:[ID],EMAIL
CPHONE:[ID],PHONE
CBDAY:[ID],BDAY

And come to think of it I wrote down a phone number in such a hurry that I only wrote down the first name and now I've forgotten the person's last name.

CFIRST:[ID],FIRST
CLAST:[ID],LAST
CEMAIL:[ID],EMAIL
CPHONE:[ID],PHONE
CBDAY:[ID],BDAY

Now show me a list of contacts with a column for each of the original attributes and don't print anything if it is unknown, and outer joins aren't allowed. I think it's interesting that this exercise in uncertainty just devolved into something like RDF triples.

Thursday, January 03, 2008

How to explain economic history?

The introduction to A History of Money and Banking in the United States (by Joseph T Solerno) places the book squarely into a methodological dispute over how one goes about interpreting and explaining economic historical events.  In delineating two approaches Joseph T Solerno poses questions such as

  •     What motivated the mass granting of right-away for railroad expansion?
  •     How did the expansion of railroads affect the net growth of the GDP?

  •     What motivates the steel industry to lobby for tariffs and quotas?
  •     What is the effect of steel tariffs and quotas on domestic steel prices?

  •     What political maneuverings, intellectual disagreements, economic interests were at the root of the formation of the Federal Reserve?
  •     How has the stock of money in the United States historically affected various economic variables like real prices, and growth?  

Each pairing of questions reflect contrasting approaches to understanding economic history.  The first question in each pairing is concerned with the motivations and interests of people and institutions involved in the outcomes.  The second question in each pairing is solely focused on quantitative information and results derived from statistical analysis.

Solerno is setting the context for Rothbard's method of "specific understanding" which is used to interpret the history of money and banking in the United States.  This method is a product of the theory of praxeology, or the study of the aspects of human action that can be grasped a priori. This is an approach that holds, for example, that a specific historical price is a unique event that requires analysis in the same manner due an explanation of the origins of the Spanish-American war.  Rothbard employs an extreme a priorism that starts from certain universal truths about human action that hold across all contexts.  These are abstract, general truths, such as: humans employ means to reach their ends, humans are constantly trying to exchange current circumstances for more favorable circumstances based on their value preferences, etc.  In short they are truths derived by Ludwig von Mises' theory of human action.  A simplistic account of Rothbard's approach is that he explains historical events by inquiring into what motivated the participants.  Using this approach, how would one explain the rise in German commodity prices during hyperinflation of the German mark in the 1920s?  One of the causal factors involved the heavy speculation against the German mark in foreign exchange markets.  As Solerno writes, this aspect would be given little weight by an Austrian-oriented economic historian,

because he approaches this event armed with the supply-and-demand theory of money and the purchasing-power-parity theory of exchange rate.  These "presuppositions" derived from praxeology lead him to avoid any attribution of causal significance to the actions of foreign exchange speculators in accounting for the precipitous decline of the domestic purchasing power of the mark.  Instead they direct his attention to the motives of the German Reichsbank in expanding the money supply.

In short, then, the theory guides the deductive inquiry and weighting of factors in the explanation.  Solerno lays out another approach for contrast and criticism: cliometrics.  It is an approach with a focus on quantitative information and statistical methods.   Causal relationships are demonstrated using the counterfactual method.  For example, why didn't the panic of 1902 lead to widespread bank failures or even temporary closings?  Solerno offers Milton Friedman's approach in the Monetary History of the United States, 1867-1960 as an example.  Freidman looks at the fact that banks promptly limited withdrawals on deposits and hypothesizes that this limit on withdrawals is a causal factor.  He then "tests" the hypothesis by seeing that it holds in other historical situations.  During the financial crisis that started in 1929 banks didn't curtail withdrawals until 1933 and in that time there were widespread bank failures.  Would swift limits on withdrawals have changed this outcome?  The counterfactual method tests a hypothesis analytically by looking for historical events that involve it, conjecturing that things might have happened differently, and examining the consequences.

Social sciences suffer from the problem that controlled experiments are difficult if not impossible to carry out, and the two approaches delineated by Solerno are in disagreement over the amount of explanatory power they credit one another.  Solerno's central objections to cliometrics is that human motivation is given so little consideration that the explanations omit essential human goals, interests, and deceptions* that are more fundamental to historical explanation.  Essentially, the light of explanation is prone to focus on the wrong aspects of an event.  Solerno also claims the counterfactual method is circular because the theory it develops is constantly open to revision by new evidence and it's explanatory power is diminished.

Despite Solerno's enthusiasm, the method of specific understanding is not without its own drawbacks.  Solerno notes that there is an impression that Rothbard's explanations are grounded in little more than idiosyncratic conspiracy theories.  He labels this impression as mistaken.  However, the heavy use a priori reasoning does have a larger margin of error.  In considering a person's motivation, weights have to be assigned to various factors.  How much of a person's motive is ideological?  How much of a person's motivation is economic?  In Super Crunchers, Ian Ayres describes how a simple computer algorithm employing statistical methods was better able to predict decisions of Supreme Court justices than a panel of court experts.  Ayres attributes the algorithm's better performance to the fact that it is hard for humans to be objective when assigning weights to causal factors.  For example everyone knows that a justice's decision, in part, depends what state the case originates.  But the experts are prone to give that factor the weight it deserves rather than the weight it has shown.  [Econ talk interview here.]

Another problem with placing a heavy emphasis on deductive reasoning from a particular theory is that theories don't always capture exceptions well.  In March of 2007 Peter Norvig gave a talk at MIT with a theme captured by a quote from the 1891 book The Adventures of Sherlock Holmes, "It is a capital mistake to theorize before one has data."  Norvig described the application Google makes of the vast amount of data it collects from the internet.  In particular, Google has built software that performs automatic translation of text from English to other languages using statistical methods.  In constructing the translation model Norvig's team brought in a linguist to build in information about noun phrases and verb phrases.  The problem was that while linguists have theories about what are noun phrases and verb phrases, there are exceptions.  The statistical models were better at dealing with the exceptions and dealt with the general cases at an acceptable level.  Overall Norvig's team achieved better translations using just a statistical model. 

These examples argue that theory and the deductive reasoning of experts isn't without problem.  Praxeological theory certainly can have explanatory power and inform lines of inquiry for economic events.  Its drive for unifying consistency is important.  However there is a large margin of uncertainty over the weighting of factors and the accounting for exceptions to theory. Perhaps there is a balance to be found between the two approaches.

* since professed intentions are not always consistent with actual intentions

Friday, December 28, 2007

Here are some of my favorite gifts I received.

Dynamics of Complex Systems (Studies in Nonlinearity)
The author, Yaneer Bar-Yam, came to ITA last June and gave a talk about factors that lead to failures of large engineering projects. The prominent example was the failed attempts of the FAA to upgrade the air traffic control system. It is an example of a system where functional testing is difficult and the system can't just be shut off while a new one is installed because it is too complex. Bar-Yam argued that changes, to be effective, have to be evolved. The book develops models and modeling techniques that can be applied to understand complex systems like this. I've wanted the book because I've read the chapter on neural networks from the version that is available for download. The account of how one can measure the storage capacity of neural networks is fascinating and Bar-Yam develops an interesting hypothesis about why sleep is a necessary function. The back cover describes the book as "the ultimate in interdisciplinary fields" and I hope by reading it I'll gain deeper insight into a variety of topics.

A History of Money and Banking in the United States: The Colonial Era to World War II
I'm surprised that others in my family said that they'd like to read this book too. I haven't read any other books by Murray Rothbard and I get the sense that this one will involve much controversy. More on that later. With the dollar having charted new lows and the prominence of the federal reserve in economic news and outlooks I'm interested to read about Rothbard's explanation of how and why this institution came to be. I'm also interested in the regulation, preservation, and prediction of the value of money--especialy mine.

Dragon NaturallySpeaking 9 Preferred
This works surprisingly well. It also integrates well with my sony voice recorder. What I find it useful for is dictating notes while I'm reading a book. Then I can have dragon transcribe the notes on the computer later. The only drawback is that it requires a quiet setting.

Monday, December 24, 2007

I portedSICP's query language to common lisp [sicp-query.lisp - I really need to set up some svn access somewhere]. I'm curious how the relational model connects with logic programming and deductive databases. Here are some queries using the relational model and here are some queries using the sicp query language.

A few weeks ago I went to a talk at MIT given by Maria-Cristina Marinescu titled Reactors: A Data-Oriented Synchronous/Asynchronous Programming Model for Distributed Applications. She contrasted the implementation of a simple zipcode look-up application in php+html+etc+etc with an implementation using the the reactor model she is helping develop at the Watson Research Lab. I understood the basic idea to be that one composes reactors which consist of relations and rules and an output value. The rules are based on datalog. Each reactor can receive an update bundle that is processed by the rules to update its relations. If a rule refers to another reactor then that reactor is pulled into the transaction if the relationship is synchronous otherwise an update bundle is messaged to the related reactor if it is asynchronous.

Thursday, December 20, 2007

I’m reading Database in Depth: Relational Theory for Practitioners.

I've sketched a relational model in common lisp. It is dynamically typed and allows for a null (unknown) value. This goes against the grain of the book, which calls for static typing. It is interesting that the relational model doesn't have to be strongly typed though if a generic equality test dispatches on type.

What I'll do next is sketch a version that uses pattern matching so I can see the similarities and differences between the relational model and the query model of prolog or SICP’s query system.