How Harvard Economists view the world

The quote of the day comes from Harvard economics professor Mankiw:

exceeding $3 million in such accounts is not very difficult for an individual who is financially successful and frugal. Under current law, a self-employed person can put about $50,000 a year in a SEP-IRA. If he does that every year for 40 years, and his savings earn a return of 5 percent per year, he will retire with about $6 million.

Do you know any 25-year-olds who are self-employed and putting $50K/year into their IRA's?

Okay, that was too hard: do you know any 25-year-olds who are earning $50K a year? There are some, not many.

Heck, it's hard to even find any 25-year-olds who are employed.

Other than the ones that graduated from Harvard, that is.

Reading list for an April weekend

The rainy season came late this year, it seems like it's been raining for the entire month of April.

If you're stuck indoors, and looking for something to read, try some of these:

• As long as Google continue to take actions like this, they still merit the "Do No Evil" brand in my book: Google Takes on Rare Fight Against National Security Letters

  NSLs, which have been in use for to decades but were greatly expanded under the Patriot Act, are written demands from the FBI that compel internet service providers, credit companies, financial institutions and others to hand over confidential records about their customers, such as subscriber information, phone numbers and e-mail addresses, websites visited and more.

  NSLs are a powerful tool because they do not require court approval. Until now, most came with a built-in gag order, preventing recipients from disclosing to anyone that they had received an NSL. An FBI agent looking into a possible anti-terrorism case could self-issue an NSL to a credit bureau, ISP or phone company with only the sign-off of the Special Agent in Charge of their office. The FBI had to merely assert that the information was “relevant” to an investigation into international terrorism or clandestine intelligence activities.

• I love this New York City art project: Jay Shells Drops “Rap Quotes,” His Most Site-Specific Street Art Project Yet

  For this ongoing project, Shells created official-looking street signs quoting famous rap lyrics that shout out specific street corners and locations. He then installed them at those specific street corners and locations.

• A beautiful article at the Smithsonian website about the evolution of the design of chess pieces: How the Chess Set Got Its Look and Feel

  According to the most widely told origin story, the Staunton set was designed by architect Nathan Cook, who looked at a variety of popular chess sets and distilled their common traits while also, more importantly, looking at the city around him. Victorian London’s Neoclassical architecture had been influenced by a renewed interest in the ruins of ancient Greece and Rome, which captured the popular imagination after the rediscovery of Pompeii in the 18th century.

  The DesignWeek article about Daniel Weil's new set design is also quite nice: Daniel Weil redesigns the chess set

  Weil says, ‘The purity of the pieces’ shape is reflected in the way they are held.’ He developed the idea of a ‘north hold’, where the piece is held between the index finger and thumb, and a ‘south hold’, where it is cupped in the hand – Weil says this hold has more ‘theatrical disdain’.

• It's springtime, so it's time to get this year's update on Colony Collapse Disorder (it's still a mystery): Mystery Malady Kills More Bees, Heightening Worry on Farms

  In the valley, where 1.6 million hives of bees just finished pollinating an endless expanse of almond groves, commercial beekeepers who only recently were losing a third of their bees to the disorder say the past year has brought far greater losses.

• Wow; this year's XKCD April Fool's Prank just totally went over my head. If you were in the same boat, the community over at StackOverflow try to explain it, or you can read this nice article from PiCloud: XKCD Hash Breaking

  How many f2 cores running in parallel would it take, on expectation, to generate a better hash within the next 6 hours? It turns out that it would take over 66,200 f2 cores running in parallel over 6 hours before we can expect to find a better hash. Unfortunately, we can’t get you 60 thousand f2 cores and the cost would be nearly $87,500 on PiCloud.

• A little flurry of patent articles:
  • Google Number One: Taking a Stand on Open Source and Patents

    We hope the OPN Pledge will serve as a model for the industry, and we’re encouraging other patent holders to adopt the pledge or a similar initiative.

  • Google Number Two: It’s time to take action against patent trolls and patent privateering

    Privateering lets a company split its patent portfolio into smaller sub-portfolios “stacked” on each other, increasing the number of entities a firm must negotiate with and multiplying licensing costs. This behavior unfairly raises competitors’ costs, ultimately driving up prices for consumers.

  • My co-worker Don Marti writes about prior art: The America Invents Act: Fighting Patent Trolls With "Prior Art"

    The AIA has a lot of changes, starting with the expansion of what counts as prior art. Prior art is any public information that shows the patented invention was not original. Patent examiners were always supposed to take prior into account when granting a patent in the first place. However, especially in the software field, the understaffed and overworked patent office misses a lot of details.

  • And a lively short article from Joel Spolsky: The Patent Protection Racket

    It is organized crime, plain and simple. It is an abuse of the legal system, an abuse of the patent system, and a moral affront.

    In the face of organized crime, civilized people don’t pay up. When you pay up, you’re funding the criminals, which makes you complicit in their next attacks.

• Felix Salmon pokes his head closer to my part of the world: The Bitcoin Bubble and the Future of Currency

  Still, for the time being, bitcoin is in many ways the best and cleanest payments mechanism the world has ever seen. So if we’re ever going to create something better, we’re going to have to learn from what bitcoin does right – as well as what it does wrong.

• Geeking out, part one: Join-Idle-Queue: A Novel Load Balancing Algorithm for Dynamically Scalable Web Services

  Unlike algorithms such as Power-of-Two, the JIQ algorithm incurs no communication overhead between the dispatchers and processors at job arrivals. We analyze the JIQ algorithm in the large system limit and find that it effectively results in a reduced system load, which produces 30-fold reduction in queueing overhead compared to Power-of-Two at medium to high load.

• Geeking out, part two: Cap'n Proto

  The encoding is defined byte-for-byte independent of any platform. However, it is designed to be efficiently manipulated on common modern CPUs. Data is arranged like a compiler would arrange a struct – with fixed widths, fixed offsets, and proper alignment. Variable-sized elements are embedded as pointers. Pointers are offset-based rather than absolute so that messages are position-independent. Integers use little-endian byte order because most CPUs are little-endian, and even big-endian CPUs usually have instructions for reading little-endian data.

• Geeking out, part 3: Routing and Web Performance on Heroku: a FAQ

  The Heroku router favors availability, stateless horizontal scaling, and low latency through individual routing nodes. Per-app global request queues require a sacrifice on one or more of these fronts. See Kyle Kingsbury’s post on the CAP theorem implications for global request queueing.

  After extensive research and experimentation, we have yet to find either a theoretical model or a practical implementation that beats the simplicity and robustness of random routing to web backends that can support multiple concurrent connections.

  (The Register sniffs in disgust)

• Fast Company explains some of the implications of culture on the way you approach a business model: What American Startups Can Learn From The Cutthroat Chinese Software Industry

  “In China, it’s like this from day one,” he says. “Companies don’t wait until later to figure out who their competitors are, since they have a business model from the start.” The benefit is never having to burn a bridge with a partner which has somehow morphed into a competitor, short-changing your users in the process.

• And lastly, whatever you might think about everything else on this list, here is proof that a boring machine is not a boring machine: World's largest tunnel boring machine lands in Seattle

  Known affectionately as Bertha, this tunnel boring machine has the widest diameter of any boring machine ever built; 57.5 feet. It's being used to dig a highway tunnel under downtown Seattle and it just arrived there today after being shipped from Japan.

Summer concert series approaches!

We're going to see Mumford & Sons at the Greek Theatre on the UC Berkeley campus on May 30th. Should be great!

The Greek Theatre is a wonderful place to see a show; we've probably been to 20 shows there over the years, and we always have fun.

The Greek Theatre has been around a long time (at least in California terms); see if you can spot it on this cool map from 1909!

Perforce 2013.1 is out!

This is already old news, but if you look over on the Perforce website, you'll see that we're now distributing release 2013.1 of the Perforce server.

This is the fifth major release I've contributed to at Perforce, and even though it's a relatively small release from a server point of view, I was pleased to be able to be part of several of the new features, and I'm excited that the company continues to deliver compelling new releases with important functionality.

Are you a Perforce user? If so, try the 2013.1 release and let me know what you think!

What is the truth about true cost?

More than two decades ago, I first heard the adage:

Marketing is the art of telling the truth in the most positive possible light.

I was thinking about that concept as I was reading through the newly-published True Cost of Ownership document from Tesla Motors.

Have you read this document? If not, I encourage you to stop right now, and go visit their site, and have a read.

To their credit, the document is short and plainly-written.

But what does it say about America, about our morals and values and behaviors?

As I understand it, from reading their document, it is actually surprisingly affordable to buy a $75,000 Tesla S, assuming that:

1. You make a $7,500 down payment
2. You qualify for a $67,500 car loan, at an interest rate (2.9%) that's a full 1% lower than what you can get a mortgage for
3. You live in a state where you don't have to pay sales tax, nor registration fees
4. You live in a state where, furthermore, you qualify for a $10,000 tax incentive for buying an electric vehicle
5. You own your own business, and use this car for your business, and deduct the interest, depreciation, and operational costs you're paying on this car and its 70K loan from your taxes.
6. You currently drive a 19 MPG clunker, for which you pay $5/gallon in gas, and you go through 15 gallons of gas a week, and you have a way to charge your Tesla without having to pay for the electricity (say, you have your charging station at your office), or at least can arrange to deduct your electric bill from your taxes.

And even then, the numbers don't work, unless you figure in some additional fudge factors such as:

1. Assuming that not having to go to the gas station 4 times a month is worth $100/month to you
2. Assuming that you'll now get to drive in the carpool lane, and that that's worth another $150/month to you

Now, tell me, and be honest: is there anyone you know for whom all these things are true? (Bonus points if that person is honestly using this car for their business, is replacing a 19 MPG vehicle with this car, and drives at least 20 miles in a carpool lane everyday for their business.)

I guess what's so depressing about this is how brazen Tesla is about just up-front admitting that everyone who buys one of these vehicles is just flat-out cheating on their taxes.

Magnus loses ...

... but yet, Magnus wins!

What a crazy last round!

Aronian solidly beat Radjabov, and Gelfand and Grischuk drew after 27 moves.

Then, in a spectacularly wild position, Carlsen came through the time control with Svidler down a pawn, and with his pieces badly scattered. Scarcely 8 moves later, Carlsen was lost, and resigned.

But then, most unexpectedly, Ivanchuk played his best game of the tournament, defeating Kramnik, and so the tournament ended with both Carlsen and Kramnik level on points, at 8.5 points each.

It turns out that the rules of the tournament then proceeded to consider various tie-breakers, and although the pair were level on the first tie breaker, Carlsen prevailed on the second tie-breaker, which was the number of victories.

So Carlsen it is, and a Carlsen-Anand match will be played!

Hooray!

Go Magnus Go!

We're down to the final round of the Candidates Tournament, and it's as close as could be: Carlsen and Kramnik are tied for first, with Svidler and Aronian tied for third.

In the last round, Carlsen has the White pieces versus Peter Svidler, while Kramnik has the Black pieces vs Vassily Ivanchuk. Having White for the last round gives Carlsen a slight advantage, but Svidler has been playing a marvelous tournament and will be a very tough opponent.

Meanwhile, Ivanchuk has been playing rather poorly, losing a number of games on time, apparently unused to the time controls used in this tournament, which provide no per-move increments.

(A slight aside on modern chess time controls, for those not familiar with how they have developed over the years:

In more serious tournaments, games may be broken up into multiple time controls that help dictate the pace of play, while also giving players plenty of time with which to think. This ensures that while a game might last six hours or longer, players are forced to reach a certain point of the game after just a few hours.

One multiple time control format seen frequently in major tournaments is 40/120, G/60. This time control requires players to make at least 40 moves in their first two hours of playing time, then gives each player another hour with which to finish the remainder of the game.

...

With an increment, players have time added to their clock after every move that’s completed. In the long international tournaments mentioned earlier, players often receive a 30 second increment – either throughout the entire game, or (more commonly) only in the final time control, thus ensuring that they’ll always have at least 30 seconds to make a move.

)

In this particular tournament, the time control is 2 hours for 40 moves, then an extra hour added for the next 20 moves, then 15 minutes more with a 30 second increment to finish. Recent top-flight tournaments have generally featured an increment throughout, I believe, so Ivanchuk's loss of several games on time has been credited by many observers to a failure to adapt to the slightly different control rules of this tournament.

At any rate, the games are underway, what will happen?