EDITOR'S EYE

The World of Software Development.

January 2008

I wish I had read this first. Not that it would have changed anything about my recent Coffee, Tea, or C++ editorial. No, but it might have given me more grist for the mill.

At issue is a new study of worldwide technological competitiveness that suggests China may soon rival the U.S. as the principal driver of the world economy. If that happens, it will mark the first time in nearly a century that two nations have competed for leadership as equals.

The High Tech Indicators Technology-based Competitiveness of 33 Nations study suggests that China will soon pass the U.S. in the critical ability to develop basic science and technology, turn those developments into products and services, and then market them to the world. Though China is often seen as just a low-cost producer of manufactured goods, the study conducted by Georgia Institute of Technology researchers shows that China has much bigger aspirations.

The study ranks 33 nations relative to one another on "technological standing," an output factor that indicates each nation’s recent success in exporting high technology products. Four major input factors help build future technological standing:

• National orientation toward technological competitiveness
• Socioeconomic infrastructure
• Technological infrastructure
• Productive capacity.

Each indicator is based on a combination of statistical data and expert opinions.

The 2007 statistics show China with a technological standing of 82.8, compared to 76.1 for the U.S., 66.8 for Germany, and 66.0 for Japan. China’s score was 22.5 11 years ago, while the U.S. peaked in 1999 with a score of 95.4.

Recent statistics for the value of technology products exported put China behind the U.S. by the amount of "a rounding error": about $100 million. If that trend continues, China will shortly pass the U.S. in that measure of technological leadership.

China’s emphasis on training scientists and engineers suggests it will continue to grow its ability to innovate. In the U.S., the training of scientists and engineers has lagged, and post-9/11 immigration barriers have kept out international scholars who could help fill the gap.

China is becoming a leader in R&D, the study notes. For instance, China now leads the world in publications on nanotechnology, though U.S. papers still receive more citations.

Posted by Jon Erickson at 05:40 PM  Permalink |

It's not every day that a famous person crosses my path, but about a year ago I was able to shake hands with Craig Newmark, as in "Craigslist." The event was the SD West 2007 conference, where Craig was co-hosting the Jolt Awards and I was backstage going crazy trying to set a Second Life session with Grady Booch. Alas, with all the craziness, I didn't get the opportunity to sit down and smell the coffee with Craig and I've regretted it ever since.

Fast forwarding, the next I saw Craig, he was a guest on Stephen Colbert's "The Colbert Report". And just like in person, he came across as a nice guy. Darn, I said, that could have been me instead of Stephen Colbert, but then I might have had to wear a tie. Oh well.

Proving again that he is a well-intentioned and all-around good guy, the latest thing that Craig can add to his list of good deeds is that through Craiglist he's donated $1.6 million to establish the first endowed faculty chair at the The University of California at Berkeley's Center for New Media. The donation, which will support research, symposia, and lectures, will be matched with $1.5 million from the William and Flora Hewlett Foundation for a total of $3.1 million. The matching funds come from the foundation's landmark challenge grant that it gave to UC Berkeley to create 100 new endowed chairs.

"The Berkeley Center for New Media and craigslist share a fundamental respect for alternative thinking in the public interest," said its director, UC Berkeley engineering professor Ken Goldberg. "Our mission is to critically analyze and help shape developments in new media by facilitating research with unorthodox ideas, designs, artworks and experiments."

Founded in San Francisco in 1995 by Craig Newmark, craigslist began as an e-mail list of events for the San Francisco Bay Area. Jim Buckmaster became the company's CEO in 2000 and has led craigslist to become the eighth largest Internet company in the world in terms of English-language page views, and the most used classifieds service worldwide in any medium.

The Berkeley Center for New Media supports research and teaching from a diverse community of more than 100 affiliated faculty members, advisors and scholars at UC Berkeley. They work in over 30 departments, including architecture, philosophy, film studies, art history and performance studies, as well as in the College of Engineering, the schools of information, journalism and law, and the Berkeley Art Museum.

"We're thrilled to support UC Berkeley at a time when unprecedented wealth is being lavished upon private institutions," said Buckmaster.

Hear, hear! I couldn't have said it any better myself.

-- Jonathan Erickson
jerickson@ddj.com

Posted by Jon Erickson at 02:33 PM  Permalink |

Sitting here watching my wife shovel snow, I'm wondering why Janko Gravner, a mathematics professor at the University of California-Davis, and David Griffeath, of the University of Wisconsin-Madison, went to all of the trouble of simulating 3-dimensional snowflakes (they call them "snowfakes") when there's plenty of the real stuff being shoveled outside.

I'm also wondering what Professor Gravner knows about snowflakes in the first place. To the best of my knowledge it hasn't snowed in Davis, California for a long, long time. Griffeath is another story. Madison, Wisconsin doesn't lack for snow. And finally I'm wondering what it is going to cost me -- I mean really cost me -- for watching my wife shovel snow instead of grabbing another shovel and pitching in.

According to Gravner, no two snowflakes are totally alike, but they can be very similar. But the mystery isn't so much why they're similiar, as much as why they're different. This is a question that's baffled mathematicians for hundreds of years. To come up with an answer, Gravner and Griffeath are modeling 3-D snowflakes by taking into account how flakes form around a nucleus, the temperature, atmospheric pressure, and water vapor density. Rather than trying to model every water molecule, their software divides the space into 3-D pieces 1 micrometer across. The program then takes about 24 hours to produce one "snowfake" on a PC. The algorithms that describe how they go about all this are presented in their paper Modeling Snow Crystal Growth III: Three-Dimensional Snowfakes. At their Snowfakes page, Gravner and Griffeath reference source code to their "Gravner-Griffeath 2d Snowfake Simulator" program (although it wasn't currently posted), provide videos (avi files) of snowflakes growing, and present some stunning photos of simulated snowflakes. Interestingly, Gravner and Griffeath also generated novel snowflakes, such as a "butterflake" that looks like three butterflies stuck together along the body. Gravner says there's no reason these shapes could not appear in nature, but would be fragile and unstable.

Gravner and Griffeath aren't the only people interested in snowflakes. Kenneth Libbrecht of the California Institute of Technology is another scientist who's been unraveling snowflakes. (Hmm, Cal Tech is in Pasadena, California. And how much snow has been shoveled there this winter?). In his recently published paper The Formation of Snow Crystals Libbrecht briefly describes some of the subtle molecular processes that lead to differences in snowflakes.

But while Gravner, Griffeath, and Libbrecht are worried about fake snow, I'm still worried about the real stuff -- and how much it's going to cost me that my wife shoveled the driveway. Let me think -- dinner or a snowblower? Any suggestions are welcome.

Posted by Jon Erickson at 11:25 AM  Permalink |

Always one for a cup of joe, it was the headline "coffee-break criterion" that led me to stop and read Brian Hayes' article Accidental Algorithms in the current issue of American Scientist. You know what the coffee-break criterion is, right? According to Brian "a computation is slow if it's not finished when you come back from a coffee break." Clearly Brian has never gone on a coffee break with me.

But the point of Brian's article isn't coffee, tea, and most certainly not me. Rather it is about holographic (or "accidental") algorithms, which are a relatively recent phenomenon in computational mathematics. Invented by Leslie Valiant and described in his paper of the same name, holographic algorithms are "a new kind of reduction that allows for gadgets with many-to-many correspondences, in which the individual correspondences among the solution fragments can no longer be identifed. Their objective may be viewed as that of generating interference patterns among these solution fragments so as to conserve their sum." Valiant adds that "their computational power comes from the mutual cancellation of many contributions to a sum, as in the optical interference pattern that creates a hologram." Got that? In otherwords, holographic algorithms deal with the boundary that exists between easy and hard (NP) problems.

As for the "accidental" part, Brian Hayes explains that Valiant "refers to them as 'accidental algorithms,' emphasizing their capricious, rabbit-from-the-hat quality; they seem to pluck answers from a tangle of unlikely coincidences and cancellations."

Don't worry, I'm not going to attempt provide in this limited space a complete summary of what holographic (call me "accidental") algorithms are. It wouldn't be justice to Hayes, Valiant, or the algorithm. And in any event, I'm not smart enough to summarize it in 25 words or less. But don't shy away from Brian's article. It is interesting and likely important to a familiar class of problems.

Posted by Jon Erickson at 04:25 PM  Permalink |

Everybody loves a survey, right? And that's just one reason why we're inviting you to participate in Dr. Dobb's 2008 Process Framework Survey on or before January 15, 2008.

The survey explores your organization's adoption and experiences, if any, with several common process/architectural frameworks such as CMMI, COBIT, ITIL, and TOGAF in both Agile and non-Agile situations. The survey should only take 4-5 minutes to complete and 10 people who participate will receive a Dr. Dobb's DVD Release 3, compliments of your's truly (and Scott).

Scott will examine the results of the survey in an upcoming column, so don't miss this chance to share your experiences with the rest of us. We -- Scott, me, and your fellow developers -- really appreciate your participation.

Again, the URL for the survey is www.surveymonkey.com/s.aspx?sm=GdRGGRhqeXGwxGsoxSQsew_3d_3d.

Posted by Jon Erickson at 03:26 PM  Permalink |

It's common knowledge that datacenters are hogs, at least when it comes to energy consumption. But it is only recently that we're finding out how big a hog they really are.

According to a recent AMD study conducted by Jonathan Koomey using IDC data, the electricity used by servers in the U.S. -- including servers, cooling, and auxiliary equipment -- was approximately 45 billion kWh, resulting in total utility bills amounting to $2.7 billion in 2005. The study estimates that the total worldwide datacenter power and electricity cost $7.2 billion annually.

But that's not the scary part. The study goes to project that, based on current trends, worldwide server consumption from 2005 to 2010 would require an additional capacity equal to more than 10 additional 1000 MW power plants. Now that's a lot of electricity.

The report reveals that electricity used by servers in the U.S. and Europe comprise about 2/3 of the world’s total, with Japan, Asia/Pacific, and the rest of the world each between 10 and 15 percent of the total. However, it appears that the rest of the world is catching up with the U.S. and Europe in server electricity consumption. Asia/Pacific (excluding Japan) consumption is increasing at a 23 percent annual rate, compared to a world average of 16 percent a year. That doesn't mean that the U.S. and Europe are using less electricity, but that the rest of the world is using more.

Datacenters throughout the world are designed and operated in similar ways to those in the United States. Accordingly, if the 20 percent savings estimated by the U.S. Environmental Protection Agency are applied to the AMD report, projections for global datacenter electricity use in 2010, total savings would equal approximately five 1000 MW power plants. In other words, says AMD, relatively modest changes in the way datacenters are designed and operated could offset approximately half the expected growth in global datacenter electricity use in 2010.

And I might add: Don't forget to turn off the lights when you leave the office at the end of the day.

Posted by Jon Erickson at 12:55 PM  Permalink |