Archive for February, 2009

Are we a nation of cowards?

February 25, 2009

Eric Holder said “In things racial we have always been . . . essentially a nation of cowards” because “average Americans simply do not talk enough with each other about race.”

Eric Holder chose an unnecessarily provocative term, and the media (and anyone who doesn’t like Obama, including Jeff Jacoby) went after Holder as an easy target. But I think Holder was making a simple and true point, which I will paraphrase as follows: “An average American feels uncomfortable about having honest discussions about race with an American of a different race, and therefore avoids these discussions.” This paraphrased comment reflects a tendency of many to avoid uncomfortable conversations, and not a universal truth or a judgment.

The storm of protest over his comments is further evidence that too many Americans spend too much of their mental energy reacting to very mild insults that they can feel aggrieved about.  By the way, there are many Americans who would be very insulted by this remark about how easily they are offended by very mild insults.  But they probably wouldn’t share this with me because they are cowards   🙂

What should OR professionals know, and how will they learn it?

February 22, 2009

No one understands all of the talks at an INFORMS conference, which has 60 parallel sessions  covering a wide range of topics.  One would need about 75 years of education.   (Note: I made up that number.)  We can reliably conclude that this is not the measure of what an OR Professional should know.  But what is the right measure?

There seem countless valuable things for an OR Professional to know  — especially a consultant —  and there is very limited time to learn them.   We traditionally emphasize probability, statistics, optimization, and mathematical modeling.  But one can also make strong cases for operations management, transportation and logistics, artificial intelligence, economics, computer science, knowledge of computer software such as MatLab and Excel, systems analysis, soft management skills, and possibly lots more.  For Masters programs we try to fit the required knowledge into anywhere from 4 to 6 subjects.

It’s even worse.   Even in the best of graduate programs, students probably don’t learn enough about a subject to employ it in an expert manner, assuming that OR in the real world is harder than OR on problem sets.   So, what can be done?  Here are some high level suggestions, with all of the key details left for open discussion.

  1. Use the subject requirements very carefully.   We should make sure that our students have knowledge that will be useful to them as professionals.
  2. Expect that our students will learn a lot on the job.
  3. Admit extremely smart and knowledgeable students, and claim full credit for their accomplishments.
  4. Provide continuing life-long education.  (Note added 2/22/09:  when I refer to continuing education, I am thinking in the broadest sense of the word.  It includes learning from books and websites, and is not restricted to courses offered at Universities or by firms in the private sector.)

The fourth point is a really interesting one for INFORMS.  How can we as a society foster life long learning in OR?  We currently provide some continuing education in our annual Practice Meeting.  But only a small fraction of OR Practitioners attend in a given year.  (Maybe 10%.  I don’t know.)  And for those who attend, an OR conference accounts for about 1.5% of a professional’s work time during the year.  Put another way, if a practitioner attends 50 to 75 years of conferences, he or she is putting in a comparable effort to a Masters program in OR, except that there is no studying and no testing.  This is not enough continuing education.

Can we draw upon the talents and energies of our professionals – especially academics –  in order to help our practitioners in continuing education?  And can we do quality on the cheap?  Before you laugh at this last question, remember that the quality of Wikipedia is quite high.   Sometimes there are innovative solutions that are both effective and cheap.

So, we need to figure out how to support continuing education, and do it in a way that doesn’t require many resources.  But, what do we need to support, and how can we support it?  I encourage you to post any suggestions you have as comments to the post.  I will alert those in charge of strategic planning at INFORMS to be on the lookout for good suggestions.

Update on Logic Puzzle

February 21, 2009

Five readers sent in solutions to the Logic Puzzle from the other day.  They are (in order in which they sent the e-mail)

  1. Isaac Moses  (Isaac’s solution  — spreadsheet).
  2. Larry D’Agostino  ( Larry’s solution — spreadsheet).
  3. Hakan Kjellerstrand (Hakan’s solution:  constraint program)
  4. Pierre Schaus (Pierre’s solution : constraint program)
  5. Joao Natali  (Joao’s solution:  constraint program).

For those interested in more logic puzzles as well as solution techniques, see the following websites provided by my daughter Jenn.  The fourth one has a  word puzzle like that of my previous post.

http://brownbuffalo.sourceforge.net/
http://www.hakank.org/minizinc/four_islands.mzn
http://brownbuffalo.sourceforge.net/DessertBarClues.html
http://brownbuffalo.sourceforge.net/BuildingBlocksClues.html  (another block/dice puzzle)

Thanks everyone.  

Incidentally, the ease for solving it using Constraint Programming makes me think that Constraint Programming should be considered a fundamental tool in the OR toolkit.

Colored letters, labeled dice: a logic puzzle

February 17, 2009

My daughter Jenn bough a puzzle book, and showed me a cute puzzle.  There are 13 words as follows:  BUOY, CAVE, CELT, FLUB, FORK, HEMP, JUDY, JUNK, LIMN, QUIP, SWAG, VISA, WISH.

There are 24 different letters that appear in the 13 words.  The question is:  can one assign the 24 letters to 4 different cubes so that the four letters of each word appears on different cubes.  (There is one letter from each word on each cube.)  It might be fun for you to try it.  I’ll give a small hint at the end of this post. The puzzle was created by Humphrey Dudley.

This problem is also a graph coloring problem.  There are 24 vertices, one for each letter.  Two vertices are adjacent if there is a word containing the two letters.  The puzzle is equivalent to assigning one of four colors to each letter so that adjacent vertices have different colors.

It can also be expressed as an integer program with variables x(i, j) for i = 1 to 24 and j = 1 to 4.  If anyone wants to write an Excel spreadsheet and solve it via integer programming, please let me know.  I’d be happy to post the Excel spreadsheet if you send it to me, or link to it if you post it and send me the URL.  My e-mail address is jorlin at mit dot edu.

Here is the small hint.  five of the words contain a “U”.  The way I solved the puzzle was to focus first on the cube containing the U.  Once that cube is determined, the second cube is much easier.  And once the second cube is determined, the last two cubes are very straightforward.  Determining the last two cubes reduces to recognizing the two parts of a bipartite graph.

By the way, we didn’t time it, but I think Jenn solved it faster than I did, and she used a different approach.

Mike Trick and INFORMS Resources

February 16, 2009

One of the greatest successes of INFORMS over the past 15 years has been Mike Trick’s OR web page, followed by INFORMS on Line and including INFORMS OR/MS Resources.  These have all been labors of love, and incredibly valuable to the OR Community.  Mike deserves at least three rounds of applause and perhaps many more.

Mike has now developed a way of using Google customizable search engine to develop a search engine for OR Resources.  Check it out.

Here is his post describing it.  

I was the one who pointed out Google’s capability to customize web search.  I was happy to play a role in keeping INFORMS OR/MS resources future successes.

An apology from a born-again blogger

February 16, 2009

I apologize for not blogging recently.

My daughter Jenn gave me “The Huffington Post Complete Guide to Blogging”.  I thank her for it.  I found it inspirational.  Surprisingly, one of my main conclusions is that I should spend less time trying to sound anything like the Huffington Post. 

What you can expect from me in future posts:

  1. Shorter posts (with exceptions when needed)
  2. More frequent posting
  3. More Operations Research, including posts about OR education
  4. Less politics (except when there is an OR perspective)

If you disagree with any of this (or even if you agree) let me know.  Also, if you see interesting OR items or if you see an OR way of doing something where OR can help out, let me know.

An Open Letter to Mitch McConnell

February 8, 2009

Dear Senator McConnell,

You’re in a tough situation.  Even though you are the minority leader of the Senate, you have far less power than the moderate Republicans to influence policy.  Moreover, the Republican track record on economics and foreign policy over the past eight years was dismal.  Your party caters to its some of its most intolerant members, notably Rush Limbaugh.   And your party somehow believes that being hyper partisan is a plan for the future.

Truth in advertising:  I don’t want to solve your problems.   But I do want to see an honest debate in Congress, and I don’t mind if Republicans have some limited influence to scale back overly enthusiastic Democrats.  So, I’ll share one idea with you.

My suggestion is for you to create two levels of Republican “support” for legislation.   The upper level of support is when the Democrats and the Republicans genuinely compromise on a bill that meets both sides half way (or close enough to half way.)  In this case, many Republican Senators can vote with the Democrats.

The second level is a partial compromise.  Perhaps the Democrats meet you ¼ of the way rather than close to half way.  It’s better than nothing, but not enough for Republicans to support the bill.  (Think “stimulus package.”)  In this case, you and other Republicans can guarantee enough votes to permit cloture (and no filibuster), but not guarantee any votes from Republicans for the bill.

By permitting this two-level approach, you will regain some level of power since you can negotiate for the low level support.   Moreover, the legislation will have a broader input from Republicans, and the debate will be more robust.

By the way, you better do this fast.  In two years, you may no longer have any options at all.

Regards, Jim Orlin

 

Saving jobs

February 8, 2009

Premise.  It’s generally easier, cheaper, and more effective to save an existing job than it is to create a new job.  

Conclusion:  A larger percentage of the stimulus package should be directed at saving existing jobs.  

More support for state governments would help a lot.  I suspect that the talented pool of economists in the Obama administration can come up with many more ideas.

Thanks, but we don’t need an arts czar: Jeff Jacoby’s column of 2/8/09

February 8, 2009

Background.  Jacoby argues that art is fundamentally important to society, but that government support of the arts is not a good idea.

 

Dear Jeff,

I’ll keep this short since I don’t have much to add.  I agree with you.  When government decides how the arts get funded, it opens it up to politicization in ways that ultimately hurts the arts.   The government already has enough problems keeping things such as the Justice Department (and justice itself) from being politicized.  There is little reason we should trust it with the arts.

Regards, Jim

Changing the tone in Washington: a prisoner’s dilemma.

February 7, 2009

Americans hate Congress.  So, why doesn’t Congress improve its behavior?  The answer can be summed up in two words.  Prisoner’s dilemma. (http://en.wikipedia.org/wiki/Prisoner’s_dilemma).  Moreover, it will be harder for Congress to improve its tone than for prisoners to be honorable.

Here is a plausible payoff matrix, with the payoff representing approval ratings.  (Don’t get too hung up in the numbers, which are rough guesses.  If you prefer, you can add a 5% do Democrats in every entry since they are more popular.)

When I say nonpartisan, it doesn’t mean caving in to the other side.   It means being honest about your positions, and not tearing down the other side in insincere and hypocritical ways (also known as the current way of doing business.)

 

                                                          Democrats
                                       nonpartisan              Partisan
Republicans
   nonnpartisan         D:50%  R:50%        D:35%    R:15%  
   partisan                 D:15%   R:35%        D:20%    R:20%           

It is best for the Congress and the country if both parties adopt a nonpartisan tone.  Unfortunately, with respect to political speech, the primary focus of each party is on its performance relative to the other party.  There is only a secondary focus on “the good of the country.”

If each party’s focus is on how well they do relative to the other party, there is a strong incentive to be partisan.  In fact, with respect to advantage over the other party, there is no difference between both parties being partisan or both parties being nonpartisan.

So, how do we change the tone in Washington?  The key is to make Congress realize that they are optimizing the wrong objective, one that does harm to the country.  (And we also need to make it clear that we will vote for those who act less partisan.) Obama gets it.  I hope that he can help Congress to get it.