O'Hare McDonald's Monday Morning Norms

Before business school, I stopped by the McDonald's in O'hare airport many monday mornings.  It was generally a "he that shouts loudest" gets his order taken and I often likened it to standing in a commodities trading pit.  There were lines, but they were only for guidance.  Interestingly, this came to mind while reading chapter 12 of Dixit & Skeath.

Specifically, the "norms" that society imposes.  The norms at this McDonald's were definitely different than most.  At most McDonald's, it's stand in line to be served.  However, the O'Hare McDonald's was quite different.  The norm for business travelers (and for the workers) were to step up to an open register or shout to an order taker when they appeared free.  They'd happily take your order, give you a number and one would shuffle over to the side to wait.  

However, those travelers unfamiliar with this process and norm often caused confusion and delay in the system.  Those unaware would apply the norms of most McDonald's to this situation.  Often, that would mean others "cutting" in line in front of them and/or general angst amongst the others in the line.  Their perception of the norm was inaccurate and when the "cutting" did occur, they at times would take offense.  While this reaction would be the "norm" in other locations, it was not for this McDonald's.  Thus, in this one small location, the norm differed signficantly from the norm of society as a whole.  I'm reminded of how different norms across different cultures can influence behavior, but it's interesting that such a particular situation can have such a different norm.

New York WMT Trampling - can game theory explain?

book page 417 highlight

On "Black Friday", the day after Thanksgiving, a Wal-Mart temporary employee was trampled to death after opening the store doors at 5:00am in New York.  Can game theory help explain why shoppers just continued rushing into the store, some even stepping over the man?

Psychologists have developed the term "diffusion of responsibility", which may help to explain this "insanity".  Specifically, diffusion of responsibility basically states that most coming in that morning knew the man needed help, but they figured someone else would aid the man.  Surely, if there had been just one or two shoppers, they would have stopped and helped the man up.  However, given the vast crowd, people assumed someone else would help the man.  This would allow the individuals in the crowd that didnt stop to help the chance to get great prices on DVD players and toys, and also get the benefit of someone else helping the man to his feet and thereby saving his life.  However, no one stopped to help, which is in line with the diffusion of responsibility idea -- the larger the crowd the less likely any one individual is to take action.

big 12 football title fairness

Texas football fans are upset.  They're upset for a good reason.  Texas beat Oklahoma in head-to-head play, yet it now looks like Oklahoma will play in the Big-12 championship game.  Where is the fairness?  People say "style points" matter.  Oklahoma has scored 50+ in several games and even 60+.  However, what does that really mean?  Just because you play teams with no defense, you are somehow better?  If point totals matter, you're effectively encouraging "running up the score", a practice that is frowned on by much of college football.  

The process, which due to the outcomes of numerous games, is down to the final "tie-breaker", which is the team that has the highest BCS ranking goes to the championship game.  

UN shirking responsibility?

Should hostile dictators be afraid of the UN?  Does the UN's past failures to address autocracies enable future dictators?  Are UN sanctions truly "sanctions" or are they easily bypassed by a determined ruler?

These questions must be addressed when discussing detecting and stopping cheating on a world scale.  For example, the UN attempted numerous times to send in "inspectors" before the Iraq war.  Sometimes they were allowed with restrictions, sometimes they were forbidden.  Was this effort by Saddam an attempt to prove that he, as a dictator, could overpower the world's security council?  

Although there are many points that are beyond the scope of this quick blog entry, I would argue that the UN's failure to take decisive action, and carry through with it, may have forced the US's hand in the Iraq war.  The UN effectively shirked their responsibility to hold governments accountable to the norms established by the people of the world.  The UN's unwillingness to effectively punish "cheating" created a state where the "norms" of society were at risk.  Had the US not intervened, those norms may have been lost to society.

It is somewhat troubling that the US plays the policeman of the world, but it is somewhat comforting to know that we will stand for the norms of our society and not let them be destroyed by a few that are after their own prosperity at the cost of all.

Monte Carlo - Buffon Needle

The "Buffon Needle problem" seeks to determine the probability of a needle landing on one of a set of parallel lines at a set distance from each other.  Deep detail can be found at:  http://mathworld.wolfram.com/BuffonsNeedleProblem.html .  For this problem, I wanted to setup a Monte Carlo simulation to calculate the value of PI.  Interestingly, I quickly became much more involved in this problem that I had planned...

Base on the MathWorld example above, I had the equations to work the geometry, however, attempting to simply proved difficult.  The geometry in this problem is obviously quite involved.  I spent some time reviewing the SINE and COSINE functions as well as the basic laws of triangles.  In addition, Excel enjoys working in Radians instead of degrees as I was to learn... eventually.

  This problem assumes the lines are spaced one inch apart and the needle is one inch long.

My excel spreadsheet is setup as the following:

Column A - Random number to generate an acute angle (<=90).  This will be used to determine whether or not the needle crosses a line.  The random number generated is multiplied by 90 to get the angle value.  Distance is also a critical part of this equation.

Column B - Distance from the Center of the Needle.  This allows for simple math no matter which way the needle is pointing.  The excel equations would get quite complex if this was distance from anywhere else other than the center.  This is a random number divided by 2 because it can be no further than 0.50 from a line (as they are only 1 inch total apart).

Column C - Hit Line.  This is where the magic happens.  This column determines whether the needle is "hitting" (or crossing) a line.  It compares the distance from the center of the needle to the line, to the length of the needle at the defined angle (using the sine function and dividing by 2 to account, again, for the middle of the needle measurement point).  If the sine of the needle angle (converted to radians for excel's pleasure) is greater than the distance to the line, the needle will cross the line.  This is counted as a hit and marked as a 1.  Misses are counted as a 0.

**Monte Carlo Action**

Now that the geometry is correct, the Monte Carlo functionality comes into play.  Initially, I ran this series of columns out to 65,000 rows.  In his equation, Buffon reasoned that the odds should come down to 2/PI or about a 63% chance.  Thus, I set out with the Monte Carlo simulation to demonstrate this and see how accurate PI could be calculated.  Through taking 2 over the hit rate, I was able to get "close" to PI.  

At 65,000 rows, taking a sample of 200 entries, the results were an average of 3.14127631 with a standard deviation of 0.00877293.  Of course, PI itself is 3.14159 and change.  Thus the Monte Carlo simulation proved that Buffon was correct in his predictions.  It also proved that PI could be calculated from a large data set.  Overall, this took several hours (a significant more than planned), but it was a very interesting project.  I learned quite a bit about Excel 2007, Monte Carlo simulations, and had a nice refresher on geometry.

bidding on whole numbers

After reviewing the coin bid data from class, I found it interesting that some 40% of the class bid on a "whole number", or an integer.  In addition, 7% of the class thought they'd be "smarter than the rest" and big the additional $0.01 over whatever whole number they had chosen.  However, only 1% decided to be "doubly smart" and bid $0.02 over the whole number... The same was the case for $0.03.  No one bid $0.04 over the whole number.  

Also of interest, 9% of the bidders submitted a bid at the 50 cent mark.  Additionally 3% thought they'd be "smarter than the rest" and bid 51 cents.  Also visible on a graph are clear spikes of bids at the remaining two quartile levels of 25 and 75 cents.  

This is interesting in that it shows a human tendency to focus on whole number or quarter numbers.  This can also be seen often in the stock market.  For example, here recently the market bounced strongly at 10,000 and then at 8,000.  The "big-round numbers" are easy targets for people to say "wow, 8,000, I should buy b/c we haven't been this low in forever!".  Of course, 8,000 should be no different than 7,995, but there is a strong psychological aspect of the round number.

Further, during a test in 1975 by Rosch, it was found that subjects are more likely to agree with the statement that "996 is almost 1000" than with the statement that "1023 is almost 1027", although both sets differ by 4.

All of this is fascinating and opens the door to a potential advantage for those that recognize this human tendency.

The author of this blog plans to explore this more through the "Round Numbers and Security Returns" article published by Harvard's Devin Shanthikumar.  This paper can be found at:

http://www.people.hbs.edu/dshanthikumar/RoundNumbers.pdf

goin' phishing - the signaling involved

"Phishing" occurs when a criminal attempts to gather personal information about a potential victim by using some type of familiar "lure".  Pervasive on the web, phishing often involves seemingly legitimate emails from legitimate retailers.  However, in a phishing scenario, this innocuous email does not link back to the real retailers site, but to a criminal's site, thus allowing for the collection of data.  Additional information around phishing can be found online, but this entry is focused on the signaling that must occur for phishing to be successful.

Very similar to the "fishing" many enjoy today, "phishing" requires a criminal to create a false perception of reality.  If the "bait" (in this case an email from an online retailer requesting username/password information) is too unexpected, people will recognize it as suspicious and often ignore the bait.  Thus the target victim is not convinced by the signal created by the phisher.  

However, the successful phishers are painstaking in their work.  They often create a sense of urgency (by requesting immediate action or "else"), which can cause unsuspecting individuals to gloss over the mental decision points and simply react.  This attempt to compel the victim may even come in the form of a mild threat ("If you don't reply now, you're access to the site will be denied", etc).  Given this scenario, one should immediately question why a retailer would threaten to cutoff a potential paying customer.  Granted the threat (by it's definition) must be costly for both parties, but is there not a way for the retailer to find a "win-win" solution?  Thus, when faced with this tactic, one must immediately assume the communication is fraudulent and an attempting phishing scheme.

Unfortunately, phishing will be with us for the foreseeable future.  As individuals, our ability to recognize the attempted signalling involved with phishing and to question any potential threats will keep us safe from these attacks.