I love that line.

I am using Python right now to experiment with a new idea. My friend Greg proposes that there be a stronger link between the file system and the Internet, the cloud, accessible from anywhere.

I think he has the right idea. No company currently offers a service that syncs files easily, quickly, and correctly. He proposes even more. Any data which can be on both your computer and the Internet can be tied together. For example, even though a text file and a blog post are very different in terms of format, you can write a program to sync them together.

And that’s what I am testing out right now. I am typing this in VIM, on Ubuntu, into a text file. I will run the command python load.py test.txt to upload the blog post. I much prefer to write using VIM.

The hard problem, of course, is concurrency. If I change the blog over the Wordpress interface, how do I update the file on my computer without something breaking? The text file must pull the changes. Should it check every minute? Every day? As I open the file? Also, how does the computer know that this text file corresponds to a specific blog post when I edit, as opposed to a brand new post? Possible solutions are to create a file with metadata associated with this text file (problems with keeping it up to date), or changing the text file to include the id number (ugly). The solution is not difficult.

It’s tragic that scientists exist.

People often think that science is about chemicals in test tubes and electricity flowing through big coils and massive explosions. This is not science. Science is grounded solely in the Scientific Method.

At the root of it, the Scientific Method argues for carefully methodical observation, model creation, testing, and re-testing. Each component is very important. Everybody observes things every day. A key step to doing real science is to create a model or hypothesis. The model doesn’t just explain the observations you noticed, but it also makes predictions about other observations that we could make in the future. This makes our understanding of the world more complete, or at the very least our ability to predict events in the world. The next key step in the Scientific Method is testing. This differentiates science from, say, Literature and other Soft subjects. Writers often do not test their theories and exponentiations. They assert it and believe it, without looking for further evidence to disprove it. Finally, re-testing is what has made science truly useful, truly transformative. Openly publishing the models and test results, and asking others to independently retest and verify results. Science asks others to constantly question the models, all models. Constantly and with wanton disdain toward any orthodoxy.

That is why I think it’s tragic that scientists exist. We shouldn’t have to categorize people who test and verify into an entirely different population, occupation. Everybody should be testing to see if what they assert is actually true. Everybody should constantly question themselves and the world around them. Instead, since people do not, society needs to go through great lengths to train, test, and certify that someone follows these patterns of thought. The process takes years, costs money, and excludes potential scientists who don’t want to go through the approval process. It’s inefficient, but partly necessary since people don’t regularly approach questions scientifically.

No, science is no special magical talent. It simply requires that you question the next thing you say. Generalize it, test it, question it, and continue to question it. That is science.

I think Investment Banks do some good things for the world. I disagree with how they sometimes view the world and interact with people. They can become dishonest and empty.

I received the e-mail below from JP Morgan today. It’s a standard invitation to apply to one of their leadership programs. I’m probably on some big list of theirs. This is all fine, except that this recruitment manager claimed she “wanted to personally tell” me about the program. This is a lie to feign sincerity. There is nothing personal about this message.

There are several telling clues. First and very obviously, I’m not in the To: line. In fact, there is no To: line. That’s because this manager probably BCC’ed a huge list of students. The point of BCC is to not let the recipients know to whom else the message was sent. She wants to hide the massive number of people she personally invited. In a personal e-mail, you would never BCC the recipient.

Second, it doesn’t say Hi Joseph, Hi Mr. Perla, Hi Joseph, or any variant thereof. It’s the generic “Hi” you use when addressing a large group. Adding my name would have been pretty easy, but it seems she didn’t even take the trouble of this modicum of effort.

Finally (fixing this would be really putting forth extra care), the body of the message says nothing specific to me. Nothing about my major, nothing about my interests. I’m glad she looks forward to seeing me on campus, but I’ve never met her. In fact, she wrongly assumes that I could be having a good semester (I’m taking the year off).

Now, maybe I-Bankers have a different definition of “personally”. Maybe, I-Bankers define personally inviting someone as mass-mailing an indefinitely large group a short generic message plus an attachment from their own Outlook. “I used Outlook myself on my computer. I didn’t have my secretary do it: it’s personalized.”

Personally, I have a problem with that definition.

———————————————————
from: [employee]@jpmorgan.com”
date: Sep 24, 2007 11:06 AM
subject: JPMorgan Launching Leaders Scholarship. Apply by November 15th

Hi,
I wanted to personally tell you about JPMorgan’s Launching Leaders scholarship. The details are in the flyer attached below. Please feel free to share this with your friends. For more information and to apply go to jpmorgan.com/launchingleaders. Hope you are having a good semester. Looking forward to seeing you on campus soon.

Best,
[employee]
———————————————————

I was researching intelligence quotient and IQ tests on Wikipedia. I stumbled upon, as one always does on Wikipedia, an interesting kind of IQ test: Raven’s Progressive Matrices.

It had a link at the bottom to an iq test: http://www.iqtest.dk/main.swf. It’s pretty interesting. I recommend you check it out. You know how I feel about IQ tests. So, I decided to figure out how it works. It’s actually pretty simple. I think that anyone smart can follow my simple tricks and figure out how to get a perfect score pretty easily and well under the time limit of 40 minutes.

It took me a little longer than twice as long as the test to figure out and document the general patterns as well as verify all of the answers.

Raven’s Progressive Matrices

The puzzle is very simple, and does not even require much explanation. It simply shows you a 3×3 (or 2×2) matrix of black-and-white symbols. The lower right corner is not filled in, but the rest are. You are supposed to deduce the pattern and figure out what should most logically fill the lower right corner.

For example:

\ | /
{ | }
( | ?


What would go in the “?” spot? Good, a “)”. That’s a pretty simple pattern. They get much more complicated, but they still are all based on just a few basic rules. Please note that I made up all these terms. You don’t know which kind of matrix a given matrix is, but you can figure it out pretty quickly.

Rules

Momentum

Look at problem 2 on the iqtest.dk site. That’s momentum. If the first symbol and the next symbol look the same, except for one little thing moves or changes or adds to itself, and then it moves or changes or adds to itself by the same amount on the next symbol, then that’s momentum. Just follow that.

Example:

( (( (((
_ __ ___
{ {{ ?


The answer: {{{.

Note that this rule can become less obvious if there is what I call “carry”. That is, if the symbol itself is a little 3×3 matrix, and you “move” to the right, then some of the elements will fall out of the little matrix, so then you must “carry” them over to the next row of the litle matrix.

Set Completion

Look at problem 8 on iqtest.dk. That is simple set completion. Think of each symbol having a number of properties: size, color, shape, etc. If you can’t sem to follow a progression like you can in Momentum, but it just looks like a bunch of random, but somewhat related things with similar properties, then the problem can be set completion.

I can best show you this in an example (use your imagination about the shapes):

^ O []
O [] ^
[] ^ ?

Answer: O

You need to complete the set of shapes on the last row. Notice that the last column also needs to complete the set of shapes (the diagonal too in this case, but that’s not always the case).

A common property of set-completion that makes this kind of problem much easier, is to look at the triangles made. Notice:


^ x x
x x ^
x ^ x


and


x x []
x [] x
[] x x


and


x O x
O x x
x x ?

.

Obviously, the ? should be an O. Set-completion is simple, if the first row has a red, white, and blue, and the second has one red, one white, and one blue, make sure the third has one of each.

This can get more complicated because you can have multiple properties, shapes and colors etc, all compounded on each other. But, if you just find the triangle, this problem is simple.

Composition

If one symbol looks like the other two put together, then it is just composition. You just have to figure out in what way it should be put together. Maybe the rule is, always put it on the inside of the first. Maybe it’s, always put it on the outside. Whatever it is, this one is usually pretty easy. I won’t even give an example. Question 30 uses composition.

Subtraction

Subtraction is much like composition, look for one thing looks like the other two put together, but with a twist. The subtraction could be complete, just one shape minus the other. Or it could be XOR (exclusive-or). You take two symbols, and take out the lines which are in one, but not the other.

Example:

_|_| | __|
|__| |_| |_|
__| |__ ?


Answer: | |

Functions

If the first symbol in a row looks like the last symbol, but the middle symbol looks weird or especially if it’s a line or arrows or something simple, then that middle symbol might be a “function.” By function I mean something that geometrically transforms the first symbol into the third symbol. The function is not necessarily intuitive, but usually makes sense in terms of what the function symbol looks like.

In the same example I used above, the vertical bar “|” is a function that reflects the first symbol horizontally over itself, like a mirror.

\ | /
{ | }
( | ?


What’s interesting is that you can apply one function to another function. So, you might apply a rotation function with a flipping function, flipping the rotation function, creating a function that both rotates and flips. Pretty cool.

Replacement

Replacement is where they trick you. The rule might be very simple, but it becomes very hard to figure out quickly, because the elements inside the symbol change for arbitrary reasons simultaneously. Question 25 is an example of movement with replacement together.

Commonality

Finally, if all the symbols look randomly chosen with a bunch of properties and possible configurations, then start to look for commonalities. Don’t look for a 1.2.3. pattern like movement, just look for rules that each symbol has in common. For example, say that each black element should be on top of a white element in exactly one symbol in each row.

This can be difficult, but is easier if you know that it’s none of the other rules, and you are looking for a commonality, not a progression of patterns. Once you have some rules, start ruling out answers until you find a final answer. Question 26 is an example.

Putting it all together

Skeptical that the answers are so easily based on the rules above? They really are. What makes the difference between an easy and a hard problem is that a hard problem will use multiple rules together. Fortunately, using multiple rules together usually doesn’t make it much harder to figure out as long as you systematically think through these possibilities. If you are having trouble with a problem, you should stop, take a deep breath, look back at the matrix as a whole, and then think through each of these rules and rule them out or use them as appropriate. All matrices follow one or more of these rules.

On the iqtest.dk site, the lowest score you can get is “below 79″. The highest you can get is “above 145.” The answers to all of the questions I put below, along with an explanation referencing the rules used to get the answers:

D Come on (momentum?)
F Momentum
B Momentum
G Set completion (angle of line)
A Momentum (size and column)
H Momentum
B Momentum (notice that the squares hide each other)
E Set completion
H Set completion / Momentum ?
A Subtraction
C Application of function (enlargement along axis)
F Set-completion (angle and number lines)
B Momentum
D Subtraction
H Subtraction
E Composition and set-completion with replacement
F Momentum (one example has carry)
C Subtraction
E Momentum with carry
D Set-completion (angle and number black/white)
G Momentum
A Oppositing? (a bit like subtraction but from sets of attributes of platonic ideal)
B Set-completion (1. small ball color, 2. big outside shape, 3. inside v. outside)
H Set-completion (1. flat bottom, 2. widening, 3. partially-closed top)
B Non-repetition? Set-completion? (Movement and replacement?)
A Commonality (180 deg rotational symmetry and middle pegs always covered)
H Subtraction
G Set-completion
E Function-application (and function-application on other functions!)
A Composition
D Subtraction yields line which is a function you apply which is reflection and delete line
E Movement
G Set-completion?
G Valuation (attach negative integer for ball inside circle, positive outside) then add
C Function application (with a bit of spatial reasoning)
F Set-completion (big-stack color, two-stack color, bar-chart position)
H Movement and replacement based on progression (replace as hidden by dark square)
F Function application (functions on functions)
B Movement with carry and replacement

So

I think I will write a program that creates Raven Progressive Matrices dynamically. I think that would be really cool, and probably useful for some psychologists.

I just discovered that Dr. Raven has a website with links to papers describing the inner workings of the Matrices: http://www.johnraven.co.uk/pubs/pubs.html . I guess that might have made my job a little easier.

I was eating in the Center for Jewish Life at Princeton the other day last Spring. I was eating with Chris, who was introducing me to his friend Yue. He’s a brilliant guy who had taken Graph Theory the year before. We were asking him questions about a problem that troubled us.

Anyway, we continued with lunch and Yue started talking to Chris and the others about their SAT and GPA for some reason. I think they were comparing each others’ test scores in a macho battle for superiority. Now, as a rule I don’t talk about scores and tests for a variety of reasons, so I was quite disengaged from the conversation. I was disengaged, that is, until Chris asked Yue about his IQ. Instead of talking about what his intelligence quotient was reported to him when he was in elementary school, he would not tell Chris.

I was interested, so I asked him why not. Yue said that he thought his intelligence quotient was a measure of his true intelligence, so he did not want people to know it. I thought that was very odd. First, I thought it interesting that he places so much value in reliability of the WISC (Wechsler Intelligence Scale for Children) to measure a “true” intelligence of himself. I also thought it was odd that he would freely talk about his SAT and GPA, but not his IQ.

Intelligence

The WISC doesn’t measure intelligence any better than a fair history exam on a period you should have studied can. The originator of intelligence tests was Alfred Binet. He created and administered them to identify children who were behind and may need remedial teaching in order to catch up. It’s only other people who corrupted the idea of an “intelligence quotient” and turned it into what people now think is a strict measure of “intelligence.” Binet even denounced using an “intelligence quotient,” saying that

The scale, properly speaking, does not permit the measure of intelligence, because intellectual qualities are not superposable, and therefore cannot be measured as linear surfaces are measured.

It makes no sense that the broad array of human mental feats can be summarized into a single dimension, one number, which you can use to line people up like you can line up people by height. Your mind is not a piece of string that can be measured. It is unfathomably complex. You, a person, are smart because you can think about the world: guess about it, change it, poke it, run away from it, and create it as you please. You are not smart because you can recite five letters backward within the allotted time.

Still, a lot of smart scientists, who have high IQ’s and are proud of them, try to justify their importance. Some put forth an idea of g, a general intelligence factor which they claim IQ tests, among others, approximate, even if they do not measure it perfectly. They show through studies that g displays a strong correlation with grades later in life. Of course, they disregard cultural bias in testing, and correlations with the affluence of parents which itself is strongly correlated with high grades, or if they don’t they try to “factor in” the effect in statistically corrupt ways. Stephen Jay Gould, in The Mismeasure of Man, delivers probably the most painful beating of IQ proponents’ ideas.

SAT & GPA

At best, IQ can be shown to be a predictor of grades. But that is the purpose of the SAT: to predict grades for college. And, in fact, the two are more than 80% correlated: http://en.wikipedia.org/wiki/IQ#Other_tests .

Which is why I found it strange that Yue wouldn’t talk about his IQ, but would his SAT. Even more strangely, since his IQ is supposed to predict his grades, his actual GPA reveals much more than his IQ. In fact, in a broader sense, even if you believe the rubbish about IQ, the GPA is much more revealing. The GPA tells you more about how you can complete assignments on time, and how much extra (or not) you will do to do well. It is a much better predictor in how hard you work in life, which is really what matters. Life, not a square in a circle with a dot in it.

Of course, your GPA is just another number that doesn’t describe how hard you worked to prove that last problem on your Mathematics take-home final, or how many hours you spent looking for information in dusty books that nobody every summarized quickly on Wikipedia. But that’s another discussion.

So

I feel bad for the people with relatively high IQ’s who still don’t understand the right amount of value to place in an IQ. It’s all the people in MENSA who keep the idea alive. I think Yue had one thing right when he said, “Wow, the people in MENSA all join because they are insecure about their intelligence.”*

* I apologize to all the people in MENSA. It’s just a joke.

More reading:
http://www.rso.cornell.edu/scitech/archive/95sum/bell.html

I moved into the Bridgewater-provided housing yesterday on Fairfield Beach. Mouse over or click for captions.

For my Physics 210 class, my team and I built a self-balancing robot. It stands on two wheels, and constantly adjusts itself so that it does not fall over, like a Segway. Building something that moves by itself and reacts to its environment is fun and very cool. The link to our final project report is below:

http://docs.google.com/View?docid=dgbtdz37_23pmw7j

Below is my schedule for the Fall 2006 semester. ORF 309 is Probability and Stochastic Systems, and a requirement for the Engineering and Management Systems and Finance certificates. Art 210 covers Italian Painting and Sculpture. PHY 209 is a Pass/Fail seminar on Computational Physics, so it should be a very relaxed, fun class. My last Computer Science prerequisite is COS 217, covering the fundamentals of low-level computing. A more fun COS class will probably be COS 402, Artificial Intelligence. Finally, I’m taking PHI 203 for broadening my philosophical knowledge. The course is entitled Epistemology and Metacognition.

In other news, the campus looks beautiful on this spring day. Pretty pink flower petals float around Frist. This is very unusual weather for New Jersey.

My tentative classe schedule for my next four years here is now up. I am signing up to major in Computer Science in the B.S.E. program here, while keeping the option to switch to Operations Research and Financial Engineering by taking those required courses simultaneously. I’m also looking at a mathematics, computing, robotics, and Woodrow Wilson school certificates.

The William Lowell Putnam Math Competition is an incredibly difficult 6-hour mathematics examination for college students. Last December, over 3500 undergraduates from across the country, most of them majoring in math, sat for the test. Of them, most receive 0 points of a possible 120. Nobody has ever achieved a perfect score.
I also took the exam last year, with the goal of scoring at least 1 point in order to best the average. Instead, I scored a whopping 19 points, placing me around the top 500. A few Princeton undergrads did significantly better, acing most of the test and scoring among the top 10 in the nation. I admire their hard work and logical genius.

Princeton’s new Student Course Guide was released recently. I made that in some spare time, updating often to make it easier to use and more functional. It sports over 2000 courses and even more sections, with over 5000 potential users all interacting to help each other. I’m guessing that the new SCG will accumulate many thousands of reviews. For now, everyone seems to appreciate it.

The Princeton Pre-Business Society took a trip to New York City to visit Lehman Brothers.
A number of employees set up a Market Trading Simulation, where we all ran around buying and selling oil contracts, S&P futures, and 10-Year T-bills. Some people acted as salespeople, who interacted with the “customers.” Lehman employees played the role of the customers: T Rowe Price, SAC Capital, etc. They gave orders on what to buy and what to sell, and in what amounts, to the salespeople. Salespeople, after creating a connection with the customer, relayed this order information to the trader in his or her respective “pit.” The S&P pit roared with energy as everyone bought and sold thousands of shares. On the other hand, the other two pits hardly moved from their starting prices. Afterwards, we had an opportunity to talk with some of the employees who told us about their jobs. Many were in fixed income, which relies heavily on the quantitative, mathematical side of the investment banking business.
The experience ended up very chaotic. When I played the role of a trader, my balance sheet made no sense, with missing blocks of trades. I think a simple computer interface could make this really easy, effective, and avoid the hassle of recording every transaction.

I saw Avenue Q tonight on Broadway. Princeton subsidized the trip, so I only paid $25 to go and see it. The show has its highs and lows, but mostly funny.

I may have over-extended myself this week. Details coming soon.

It snowed yesterday!

This first week has been very, very hard. I have to go work now, too. Hopefully, I can get more sleep and write more next week.

Woohoo! Spring classes have started!

The Princeton Registration website just updated where my classes will be. Here are my classes again:




My classes are here:




Click on the picture to get a larger view of the image. H is my dorm, and each number stands for a class, indicated on my schedule. As you can see, my classes are pretty close together, with my only long, quick journey being between Linear Algebra and Computer Science, 2 and 3. But that probably only takes about 7 minutes, a workable walk.

I took my last final exam of my first semester here, MAT 201 Multivariable Calculus. It went pretty well. I’m glad to be doen with exams, although it means I have only three and a half more years here.

My classes for Spring 2006 next semester look exciting: