The Institute for proving Graph Isomorphism is in P

(This post was inspired by Adam Winklers awesome book
Gunfight: The Battle over the Right to Bear Arms in America.
Disclaimers one: Adam Winkler is my cousin and I got a free copy.
Question: Should I give him a free copy of my VDW book when it comes out?
Disclaimer two: Scott did a post on a related matter here.)

If someone started an Institute to prove Graph Isomorphism is in P that would
be very odd since it could be that GI is not in P.
If someone started an Institute to study Graph Isomorphism that would be
much less odd (though still somewhat odd).

Does it make sense to have an openly biased think tank?

1. If a pro-gun-control person writes a book that proves that there weren't that many
   guns in America in the early 1800's would you believe it?
   
2. If an anti-gun-control person writes a book claming that the more guns there are
   the less crime there is, would you believe it?
   
3. The CATO Institute: A Libertarian Think Tank.
   If they did an honest study of gun control and concluded that it does reduce
   crime then would they publish it? I honestly do not know.
   If they did an honest study of gun control and concluded that it increases
   crime then would anyone believe it? Being openly biased might undermine their credibility.
   
4. The Tobacco Institute (they no longer exist). They produced reports
   claiming that smoking was not unhealthy (or perhaps that the evidence is incomplete).
   They were employed by the Tobacco industry. Did they ever have any credibility?
   Did they do any unbiased science, perhaps on non-smoking issues?
   I honestly don't know.
   

It is tempting to say Scientists should not have an opinion before they do a study. But this is clearly not correct in theory or practice. Scientists do indeed have an opinion, even an interest, in what a study will tell. Why is that different from the Tobacco institute?

1. An honest scientist's preconceived notions are hopefully also based on science and not on who is paying him and not on other non-science factors.
   
2. An honest scientist, when faced with evidence that they are wrong, will hopefully pursue that evidence and perhaps change their mind. This might be easier in math than in science since Proof is our accepted criteria. For example, I doubt there are diehards who still think that NL ≠ coNL.
   

The New Patrons

A few centuries ago if you wanted to do science and not independently wealthy you needed help.

Most of the important astronomers and natural philosophers (as well as artists) in the 16th and 17th centuries depended on the patronage of powerful religious or political figures to fund their work. Patronage networks extended all the way from Emperors and Popes to regional nobles to artisans to peasants; even university positions were based to some extent on patronage. Scholarly careers in this period were driven by patronage, often starting in undistinguished universities or local schools or courts, and traveling closer or farther from centers of power as their fortunes rose and fell.

Today most scientists have salaried positions at universities and get funded by the government but with sequestration and budget cuts, scientists have to seek out other sources, such as industrial funds. We've long had various scholarships endowed by private donors: Sloan, Packard, MacArthur. Recently though we've seen some new patrons, the upper 1%, who want to help out where other funds are limited. Some of these work through endowed positions at universities, but we also see some who create foundations dedicated to funding directed at research.

In the past few months I came face-to-face, or at least in the same room, as two of them: Landon Clay in Oxford for the opening of the new Maths Institute partially funded by his foundation and Jim Simons, when I visited Stony Brook and had lunch in the Simons Center for Geometry and Physics. The Clay Mathematics Institute funds several mathematicians and offers the million dollar bounty on P v NP and other open questions. The Simons Foundation supports a few theoretical computer scientists, not to mention the Simons Institute in Berkeley.

Of course the more money coming into our field, the more research we can do. But patronage does have its other side.

Patronage, and the desire for more, also shaped the work and publications of scientists. Effusive dedications to current or potential patrons can be found in almost every scholarly publication, while the interests of a patron in a specific topic was a strong incentive to pursue said topic—or reframe one's work in terms of it. Galileo, for example, first presented the telescope as a naval instrument to military- and commerce-focused Republic of Venice; when he sought the more prestigious patronage of the Medici court in Florence, he instead promoted the astronomical potential of the device (by naming the moons of Jupiter after the Medicis).

 How much do the lessons of the 16-17th centuries still apply today?

Local Reductions

With the STOC deadline passing on Monday, now is a good time to look at the arXiv to see what has been posted since then. Hamid Jahanjou, Eric Miles and Emanuele Viola have a new paper, Local Reductions, that gives a new reduction from NTIME(t) to 3-SAT formulas of size t polylog(t). The twist to their new reduction: there is an NC⁰ circuit C that maps the number i to the ith clause. NC⁰ means every output bit depends on only a constant number of input bits. The proof uses old-fashioned parallel routing.

Should have some interesting applications. It does save a step in Williams' proof that ACC⁰ ≠ NEXP but the combined proofs are longer.

In other news, I've been getting several email from other CS chairs looking for students to hire as faculty in their departments. The latest CRA News is 59 pages, 50 of them are faculty job ads. It's a good year to be on the job market.

Four answers to the Recip problem

In my last post I asked you to solve the following question which
was from the Maryland Math Competition:

The inequalities 1/2 + 1/3 + 1/6 = 1 and 1/2 + 1/3 + 1/7 + 1/42 = 1
express 1 as a sum of three (resp. four) reciprocals.

Find five positive integers a,b,c,d,e such that
1/a + 1/b + 1/c + 1/d + 1/e = 1.

Prove that for any positive integer k GE 3 there exists positive intgers numbers d1,d2,...,dk
such that 1/d1 + ... + 1/dk.

The HS students had the following solutions.
I list the answers to part b first.  I sketch the proofs. They are all by induction.

1) Use 1/n = 1/(n+1) + 1/n(n+1).  This was the most common solution.  This leads to (2,3,7,43,1806) for part a.

2) Since the question itself gives the solution for m=2 and 3 we only need P(k) --> P(k+2)
Use 1/n =  1/2n + 1/3n + 1/6n.  This leads to (2,3,12,18,36).
One of the students later told me that knew the solution (i) but did it this way to
avoid having to multiply 42 by 43 which is needed to get part a using that solution.

3) Inductively that the largest denom n is even Use 1/n = 3/3n = 1/3n + 2/3n = 1/3n + 1/(3n/2)
Less than five students did 2b this way.  This leads to (2,3,7,63,126) for 2a.

4) If (d1,...,dn) is a solution then so is (2,2xd1,...,2xdn).
Only two student did it this way.  It leads to (2,4,6,14,84), which they both used.

NOBODY did in the non-inductive way mentioned in the last post.

There were THIRTY TWO solutions to 2b.  Several people had their part 2a and 2b not
related to each other at all.  This was far more solutions than I anticipated.
While grading I got good at adding reciprocals.
I list them in lex order along with how many people did that answer.
(This is likely approx- I may have miscounted a bit, but its basically right)

(2,3,7,43,1806) - 91 (linked to solution 1 above)

(2,3,7,48,336)  - 3

(2,3,7,56,168)  - 1

(2,3,7,63,126)  - 6 (linked to solution 3 above)

(2,3,7,70,105)  - 1

(2,3,8,25,600)  - 1

(2,3,8,30,120)  - 1

(2,3,8,32,96)   - 6

(2,3,8,36,72)   - 5

(2,3,8,42,56)   - 11

(2,3,9,21,126)  - 2

(2,3,9,24,72)   - 4

(2,3,9,27,54)   - 3

(2,3,10,20,60)  - 5

(2,3,11,22,33)  - 1

(2,3,12,15,60)  - 1

(2,3,12,16,48)  - 1

(2,3,12,14,84)  - 2 (linked to solution 4 above)

(2,3,12,18,36)  - 12 (linked to solution 2 above)

(2,4,5,25,100)  - 3

(2,4,5,30,60)   - 1

(2,4,6,14,84)   - 3

(2,4,6,16,48)   - 1

(2,4,6,18,36)   - 2

(2,4,6,20,30)   - 1

(2,4,7,12,42)   - 4

(2,4,7,14,28)   - 2

(2,4,8,12,24)   - 6

(2,4,8,10,40)   - 2

(2,5,6,10,30)   - 1

(2,5,6,12,20)   - 2

(3,4,5,6,20)    - 3

A problem on Reciprocals

(I thought I had posted this a while back but I can't find it in past blogs
so I think I did not. I DID post a diff problem on reciprocals.)

Here is the question I graded a while back on a  Maryland Math Olympiad.
I request that you do it and post your answer as a comment- I'll be curious
how your answers compare to the students who took it.
I will post the solutions the students used in my next post and comments
on how they were similar or different than yours.
The students had two hours to do five problems.
This was problem 2.

The equalities 1/2 + 1/3 + 1/6 = 1 and 1/2 + 1/3 + 1/7 + 1/42 = 1
express 1 as a sum of three (resp. four) reciprocals.

PART A: Find five distinct positive integers a,b,c,d,e  such that

       1/a + 1/b + 1/c + 1/d + 1/e = 1.


PART B: Prove that for any positive integer k  GE 3 there exists k distinct positive intgers numbers d1,...,dk such that

1/d1 + 1/d2 + ... + 1/dk = 1.

A Theorist Goes to SOSP

Monday I attended the 24th Symposium on Operating Systems Principles, the lead conference for computer systems research. Why would a nice theorist go to SOSP? Trying to recruit a few good systems faculty for Georgia Tech.

I really enjoyed the day in ways I didn't expect. I found several of the talks interesting, even from a theory perspective. Austin Clements, in the first and one of the best paper talks, said he had a theorem and proof (roughly if operations scale there is an implementation that scales well on multicores), though purposely left the formalization and proof out of the talk and focused on implementations. Kay Ousterhout built on some theoretical tools for job scheduling. In a talk after I left, a group from Texas takes a step towards practical proof-based verifiable computing. I never expected to be cited in a SOSP paper.

When I go to a theory conference I see so many people I know that I don't spend enough time meeting new people. At SOSP, I knew a handful of people and just had a great time talking to people I haven't met before, particularly students.

Only thirty papers get presented in single track in this conference held every two years. STOC/FOCS accepts over 300 papers in the same time period. Having an SOSP paper is a really big deal. Despite having only thirty talks and traditionally held in hard-to-reach places (this year an hour and a half drive from Pittsburgh), there were 628 attendees split 42% students, 42% non-student academics, 15% industry and one member of the press.

The 2013 SOSP is the first ACM conference will fully open proceedings and the authors retained full rights to their paper, the gold standard espoused by many in our community. It didn't come cheap, the conference put up $1100/paper to the ACM to pay for the privilege.

My Pope Number is 2: The Smaller World Hypothesis

I proofread Piergiogrio Odilfreddi's book (which is on Lance's List of Favorite Complexity Books) for which I got a generous acknowledgment. I have also
visited him in Italy, though not for a while. 

Benedict.Pope Emeritus (I think that's what he is still called) broke his silence with a letter to Odilfreddi, see here.

Hence I am two handshakes away from Pope Benedict.  It used to be said that there were Six degrees of separation-- for all people a,b there is a path of length at most 6 that links them. The graph varies with you you ask, but it tries to pin down that a and b know each other.

Is six now too big? One measure is how many Google hits
`X degrees of separation' gets

• Six degrees gets 1,760,000 hits
• Five degrees gets 97,300 hits
• Four degrees gets 159,000 hits
• Three degrees gets 605,000 hits
• Two degrees gets 843,000 hits

The last one may not be quite fair- there was an episode of Pokemon
with the title `Two degrees of Separation' and also a company with that name.


How well two people know each other has to be defined carefully.

1. Erdos Numbers- Put an edge between a and b if they have a paper together.
2. Bacon Numbers- Put an edge between a and b if they appear in the same movie.
3. Handshake Numbers (I am not sure its every been called that)- Put an edge between a and b if they have shaken hands.
4. knows-number (likely not defined). Put a DIRECTED edge from a to b if a will return b's phone calls and/or email.
5. Twitter Numbers (Not sure if its ever been defined). But a directed edge between a and b if a follows b on twitter.

Odilfreddi may be an articulation point in the handshake graph or the knows-graph since he is in math AND known to the public (at least in Italy) as an outspoken atheist, so he connects two worlds. Another articulation point might be David Seetapun who has a PhD in computability theory (he worked on Recursive Ramsey Theory which is how I know of him), Finance (Goldman Sacks), Gambling in Las Vegas, and swordfish fishing (he won the Golden Fly Tarpon Tournament). He may be the key to connecting mathematicians to fisherman.

The following is probably known but I couldn't find it- what is the longest distance between two websites (number-of-links to go from one to the other)?
The average? Are these numbers getting larger or smaller?

ADDED LATER: Christian Sommer emailed me the following two
RELEVENT links:

Diameter of the web and

Tools to study the web graph

The first link claims the avg diameter of the web is 19.

Andrzej Mostowski (1913-1975)

Andrzej Mostowski was born 100 years ago today. While Mostowski worked in many areas of logic, including early fundamental work on model theory, for our readers he's best known for co-discovering the arithmetic hierarchy, sometimes called the Kleene-Mostowski hierarchy.

The arithmetic hierarchy has a few different equivalent definitions but let's use one based on computability. We define inductively there hierarchies, Σ_i⁰, Π_i⁰ and Δ_i⁰. Σ₀⁰=Π₀⁰=Δ₀⁰ are the computable sets and

1. Δ_i+1⁰ are the sets computable with a Σ_i⁰ oracle.
2. Σ_i+1⁰ are the sets computably enumerable with a Σ_i⁰ oracle.
3. Π_i⁰ = co-Σ_i⁰.

In particular, Δ₁⁰ are the computable sets and Σ₁⁰ are the computably enumerable sets. The halting problem is Σ₁⁰-complete under computable reductions, the set of Turing machines that accepting infinite sets are Π₂⁰-complete.

We completely know the structure of the arithmetic hierarchy, for i > 0, Σ_i⁰ ≠ Π_i⁰ and for i ≥ 0, Δ_i⁰ = Σ_i+1⁰ ∩ Π_i+1⁰.

The arithmetic hierarchy inspired the polynomial-time hierarchy in complexity theory. Unlike the arithmetic hierarchy, separations in the polynomial-time hierarchy remain open and any separation implies P ≠ NP. While we have relativized worlds which do quite a few different separations and collapses in the polynomial-time hierarchy the following remains open: Does there exist a relativized world where the polynomial-time hierarchy looks like the arithmetic hierarchy, i.e., for i > 0, Σ_i^p ≠ Π_i^p and for i ≥ 0, Δ_i^p = Σ_i+1^p ∩ Π_i+1^p?

University of Maryland Job Posting Mentions Quantum Computing explicitly!

The University of Maryland at College Park has its job posting up (its been up for a while). You can look at it here. I It lists THREE areas but says that they will take applicants from any area. This is believable since they only listed three. Had they listed (say) seven then I would not believe they are looking at other areas. What is the X such that if they list X then you believe they will take from other areas but if you list X+1 then you don't?

The three areas listed are:

1. Cybersecurity
2. Quantum Computing
3. Natural Lang. Proc.

All three of these seem more particular than I usually see in job postings. That is, I've seen things like  Systems, Theory, AI. SO- is this unusual? I don't quite know--- I haven't been on the market for a long time.

Science and Humanities

David Hollinger, a historian, wrote a recent Chronicle Review article The Wedge Driving Academe's Two Families Apart: Can STEM and the human sciences get along?, one of a number of articles I see talking about the connections between science and humanities and the future of humanities at universities.

Most scientists do find great value in the humanities and I would hope vice-versa. But when funds get tight, different fields talk about their relative importance--it happens between science and humanities broadly, it happens between theory and systems in CS departments with limited slots to hire.

I feel badly for humanities these days. In a tight job market, students and parents think hard about doing a humanities major while universities are trying to find ways to cut costs. I don't have a solution--right now the job market calls for more computer scientists than English majors, but I would hate to see an intellectual core of our academic world shrink away.

Humanities are cheap. A provost once said to me it costs the same to hire five philosophers as one physicist once start-up costs and salary are considered. We should find a way to keep funding the humanities while maintaining the strengths across all fields.

Pushing the bounds of human wisdom is important, whether it be in chemistry or classics. Only when we push in all directions does the ball of knowledge truly expand.

Teaching without a net

As a grad student I was teaching the linear-time Median finding algorithm and I FORGOT
that I needed to solve the more general problem of selection. After less than a minute
of trying to see what was wrong I told them

I am sure that Median IS in linear time. I will consult sources  and redo this tomorrow.

I then did the rest of the lecture (which didn't require knowing the Algorithm for Median) and the next day I did the linear Median Finding Algorithm correctly.

Note that I was teaching well known material. So I KNEW that what I was saying was true even if I couldn't  prove it. I also KNOW that I could look it up. I was TEACHING WITH A NET.

When I taught Grad Algs a few years ago I sometimes didn't quite know how the PROOF went  BUT I knew that the STATEMENTS I made were correct, and the algorithms and proofs were out there. In one case I emailed the original author with a subtle point I was stuck on. (It really was subtle- the author himself had to think about it). TEACHING WITH A NET

Last semester some of my Ramsey Theory course was taught WITHOUT A NET. Not in termsof the statements of theorems, but in my attempt to find easier proofs of theorems--- sometimes my alleged proof DID NOT WORK. And there was no book I could consult, nor person I could ask, to help me out on these new ``proofs''. One of my attempted simplifications (of the Canonical Ramsey Theory) DID NOT pan out in the end.

This semester  I am teaching an honors interdisplinary course on Fair Division (nicknamed 'Cake cutting'). I've pulled material from a  variety of different subfields (math, CS, AI. Yes AI!). So I have put some things together that are ``new''(not worth-publishing-new but new in some sense). Some of them have been wrong, or to be more fair, not quite right. But WHO CAN I ASK? Nobody! This is truely teaching WITHOUT A NET. I have made about 2 incorrect statements (both of which were prefaced with `this might not be quite right') but the bigger effect is that every day I wonder if what I am saying is correct.
The effect on the actual course is mininal-- but my mentality going in ``will I make a mistake today that I cannot recover from'' is... interesting.

What to do if you are wrong? Own up to it ASAP. Every minute you fumble around you lose the classes interest.

Is the course working? I think so-- they are learning and having fun. It helps that they are honors students who chose to take this course.

2013 Fall Jobs Post

Time again for the annual fall jobs post. As always the best places to look for academic CS positions are the job sites at the CRA and the ACM. Also check out the postdoc and other opportunities on the Theory Announcements site and the Intractability Center. It never hurts to check out the webpages of departments or to contact people to see if positions are available.

I encourage everyone who has a job to offer in theoretical computer science at any level to post links in the comments.

Faculty hiring has rebounded nicely and with computer science enrollments expanding, it should continue to be quite robust. Postdocs will still be down from a few years ago.

Good luck to everyone in the market. I look forward to seeing your names in the 2014 spring jobs post.

Who controls what is taught- the dept or the students?

There is a debate about the questions:

To what extent do we give them what they NEED?  what they WANT?


These questions permeate many other discussions of education.

Rather than discuss this profound issue I will discuss a fictional example.

1. A dept offers one section of Operating Systems (henceforth OS) in the fall and one section in the spring.
2. The same dept also offers one section of AI (henceforth AI) in the fall and one section in the spring.
3. They notice after a few years that the OS tends to underfill and the AI course tends to overfill.
4. Hence they switch to offering OS in the Spring only, and  AI is offered two in the fall and one in the spring.
5. Over time more students take AI and less take OS. Some of this is interest but some is that AI is easier to fit into a schedule since its always offered and has two sections in the spring.
6.  All of the teachers are excellent (remember this is fictional) so the quality of teaching is not the issue. The courses are of equal difficulty so this is not the issue. The courses have the same prerequisites so this is no the issue.
7. The next hiring season they decide to hire someone in AI since they need the teaching help.

The department DID NOT  mean to send the message:

AI is more important than OS.

NOR  did they mean to send the message

 We will let the students decide what is important.

  But the department ended up sending both messages. What should the dept have done? For one they should DECIDE if this is okay with them--- is AI more important than OS? Or more directly, is it okay that students graduate without having a course in OS as  long as they've had a course in AI? They may decide YES- and that would be fine. If they decide NO they could restructure the requirements OR have the advisers give that advice OR just offer less sections of AI.

Does your department fall into this trap--- ending up giving  student's opinions more sway then you intend? 

Shut Down

The NSF core proposal in theoretical computer science, or Algorithmic Foundations as the NSF calls it, has three deadlines this academic year:

• Medium proposals ($500k-$1.2m): October 15
• Large proposals ($1.2m-$3m): November 19
• Small proposals ($0-$500k): January 17

For the most part nearly every core program in computer science has the same deadlines, making it quite an interesting time in CS departments when most of the faculty are all submitting their proposals last minute in January.

Let's talk not about January but about October 15, next Tuesday. Good luck trying to download the proposal call for algorithmic foundations, or the NSF grant proposal guide, or submitting your proposal on Fastlane. All NSF links take you here, where you can read all about what is not happening at the NSF during the government shutdown. So what about October 15?

Once normal operations resume, NSF will issue guidance regarding any funding opportunities that have a deadline or target date that occurs during the government shutdown. Such information will be disseminated via a FastLane Advisory and other electronic methods.

In principle, the government could reopen for business Tuesday morning and the proposals would still be due Tuesday 5 PM. I would guess the deadline would be extended but there is nobody to ask, no one at the NSF to answer the phones and NSF employees are forbidden from responding to or even reading email. At least those that already have grants can keep spending their money, most importantly continuing to fund their students.

These are short term problems, the government will re-open at some point and the NSF will get back to business. But all discussions seem to lead to budget cutting and even just erasing the sequester of last year seems unlikely. The budget crises hasn't stopped Eric Cantor and Lamar Smith from continuing to trash some NSF grants.

Science is too important to be a pawn in politics. Investments in science have given incredible value back to America in terms of jobs and economic growth. Yet somehow science never gets mentioned as a tragedy in the Washington money battles.

P vs NP is Elementary? No-- P vs NP is ON Elementary

As I am sure you all know, the TV show Elementary  (Premise- Sherlock Homes in Modern Day NY. He emails and Texts!  Watson is a female! and...) had an episode that involved P vs NP in a big way. I think they would have been better off with a fictional problem (Bourbaki's conjecture in Recursive Algebraic Topology?) rather than a real problem that they could say rather odd things about.

1. Sherlock Holmes doesn't know what the Mill. Prizes are. I thought most educated people did. Everyone I know knows about them. Could be the company I keep.
2. The show indicates that `Solving P vs NP' means `showing P=NP' It never seems to dawn on them that maybe P is NOT NP.  
3. The show  assumes that once P=NP is proven it will take a very short time to write a program to  use it.  If P=NP is true then I suspect taking the proof and making it work on real world problems would take several years.
4. The show focuses on P=NP's implications for crypto. As Lance has pointed out in his book if P=NP then the benefits for society are GINORMOUS, and would dwarf the relatively minor problem of having to switch to private key  (I agree with Lance for the long term, but I think the short term would be chaotic for security).
5. The show refers to seven Mill problems. While this is technically correct they really should mention that one of them (Poincare's conj.) was already solved. 
6. They seem to think that algebraic geom would be used on P vs NP. If they were claiming it was being use to prove P NE NP then I would think of the Geometric Complexity Theory Program and be impressed. Since they were using it to work on P=NP I'm less impressed. If Alg Geom really is used to prove P=NP then I'll be impressed.
7. How was the episode- I am a fan of the show in general, and this was a solid but not outstanding episode. I wonder if I knew less about P vs NP would I enjoy it more.
8. They are talking about P vs NP on National TV! That's kind-of nice. Only danger is the overhype. If  P NE NP is shown and this has no real world applications then the public may be confused.  I suspect we won't have to worry about that for at least 300 years.

Celebrating Maths in Oxford

This week I'm in Oxford for the opening of the new Andrew Wiles Mathematical Institute building and the Clay Research Conference including on workshop on New Insights on Computational Intractability.

The building is beautiful with two small towers, one each for pure and applied math, and a common room joining them. The downstairs, where the talks are being held, is separated from the tower by its own glass dome, so, I was told, that the noise of the students don't disturb the great thinkers above.

The Clay Math Institute, best known in our circles for the millenium prize problems, has moved to Oxford from Cambridge (Massachusetts) and will take up residence in this new building. Unlike Jim Simons, Landon Clay didn't have a particular math background but was looking for a purpose for a charitable foundation. He met Andrew Wiles and loved the story of his proof of Fermat's last theorem and Clay realized the need for basic math research. Besides the millenium problems, the institute sponsors a number of research fellows and the move to Oxford reflects a transition to funding American mathematicians to a more international base.

What about the new insights into intractability? Lots of great talks on connections to physics and economics, on proof complexity, information theory, algebraic and circuit complexity. On the other hand I watched some talks on other millenium prizes and while difficult to follow, it looks like they have measured progress towards resolving their problems. In complexity, we're still searching for that true path towards P ≠ NP.

Long Tails and Fat Heads

Sometimes words or phrases are used in MATH and then spread to the REAL WORLD. I have blogged about how the terms Prisoner's Dilemma has become a real-world-phrase here and speculated about the terms Venn Diagram, Zeno's Paradox, and n+1 here.

I recently came across a pair of words that are related--- one of them seems to be (like Prisoner's Dilemma) going from MATH to THE REAL WORLD. The other one is very odd in that I've seen it in the REAL WORLD but it SHOULD be in MATH.

Long Tail: A Probability distribution has a long tail if there are MANY items that have a SMALL but NON-ZERO prob of happening. This is a term in probability. However, I have seen it used in the REAL WORLD as in Amazon has a long-tail strategy meaning that they will sell LOTS of DIFFERENT things even if the number of people buying some of them is small (like this which is ranked 9,578,520- though I doubt they can be that precise). This article from the Atlantic Monthly points out that ESPN used to have a long tail strategy (e.g., showing Billiards and others sports that are not that popular, but ALOT of them) but then abandoned it for... see next definition. Note that the term Long Tail is used for both a type of Prob Dist and a marketing strategy related to it. How common a word is Long Tail? It gets 66,500,000 hits on Google. The first page has the definition above only. The 10th page had about half of the hits with the def above.

Fat Head: A strategy where you concentrate on just a few items. ESPN is doing that by covering just a few sports, but the most-watched ones (too bad, I was hoping they would cover my favorite sport, chess boxing). This SHOULD be a math term for a Prob Dist with just a few points of high prob. I asked my friends in the ML community and he assures me that NO its not a math term--- but it SHOULD be! How common a word is this? It gets 2,300,000 hits on Google. The first page seems to have NOT have ANY reference to the definition above.

SO- this COULD be a case where a term used in the REAL WORLD migrates to MATH with essentiallythe same meaning. This isn't that uncommon (the term Continuity comes to mind) but this timeI will have predicted it! Maybe I should do Machine Learning.

Complexity and FOCS

The Conference on Computational Complexity Call for Papers is out, deadline November 27.

The deadline for early registration and hotel for the FOCS conference in Berkeley is October 4. Student travel support is available.

Dealing with Death

Mary Jean Harrold, a professor of software engineering at Georgia Tech, passed away last week. Mary Jean was 67 and still quite active before the cancer struck.

Computer science is still a relatively young field and most of even the early computer scientists remain quite alive. So a death in the field, particularly a colleague, makes a mark because it typically is happening at a young age. I've lost five co-authors (Avner Magen, Steve Mahaney, Andrej Muchnik, Nick Reingold, Carl Smith) all well before their time. Each death is a stark reminder of what's important in life, what does the next theorem mean when life seems so short?

We're nearing a time in computer science that many of our ranks will die after living to a ripe old age. Those remembrances will be of a life well lived. But there will always be lives cut short. The best we can do is remember them and move on and continue to build on the research tradition they left behind.

Crystal Math- What NUMB3RS and BREAKING BAD both get wrong

The TV show Numb3rs  had as a premise that a GENIUS mathematician
could help solve crimes. Is this true? I rather doubt you need a GENIUS-
though of course some prob, state, data mining,  the math behind forensics, and a few other things help. And it may help to know some number theory if a mathematician who is working on the Riemann hypothesis has his daughter kidnapped.  But I don't think you need someone on the level of Charles Eppes.

The TV show Breaking Bad  (see Honest Trailor for Breaking Bad and/or
Idiots Guide to Breaking Bad if you've seen the first 4.5 seaons at least)
has as a premise that a GENIUS chemist can make really good crystal meth. And as a by product it's blue. I know nothing about the crystal meth business; however a chemist friend of mine (who has never made the stuff) tells me that YES, being a careful chemist is good, and certainly better than a meth-head who is more likely to blow up his lab than to produce any, a GENIUS chemist would not be any better than a good chemist.

The TV show Elementary  (Sherlock Holmes in modern day New York) and many other shows (Monk, Perception, Psyche, The Mentalist, Columbo, and others I am sure) has as a premise that a GENIUS observer could help solve crimes. This may be more true then the above, but there are other tools available today (e.g., DNA).

All of these shows, and others, make the  FALLACY OF EXTRAPOLATION. Taking a good idea and extrapolating it to absurdity.

Here is a non-TV example: If blogging is part of my job, and I can deduct job expenses for Tax purposes, then I should be able to deduct the cost of the DVD's for Numb3rs that I bought because of this post.