EDITOR'S EYE

The World of Software Development.

August 2006

Depending on your point of view, technology transfer programs at public universities are either the best thing since sliced pumpernickel or scams perpetrated by predatory professors.

In a nutshell, technology transfer typically involves the process of turning research into practical (i.e., "money making") applications. Almost all universities these days have offices of technology transfer responsible for identifying the commercial potential of university-conducted, public-financed research. That's Part 1 of their charter. Part 2 is to come up with ways to turn that potential into revenue, usually by patenting the research results and licensing technology to third-parties.

At the heart of public university's ability to get involved in technology transfer is the 1980 Bayh-Dole Act, co-sponsored by Senators Birch Bayh and Robert Dole. This federal law makes it possible for universities, nonprofit research institutions, and small businesses to own and patent inventions developed under federally funded research programs. According to the Bayh-Dole Act:

• Universities can retain title to innovations developed under federally funded research programs.
• Universities can collaborate with commercial concerns to promote the utilization of inventions arising from federal funding.
• Universities are expected to file patents on inventions they own.
• Universities are expected to give licensing preference to small businesses.
• The government retains a non-exclusive license to practice the patent throughout the world and the government retains march-in rights.

To give you an idea of the impact of the Bayh-Dole Act, consider that before 1980 fewer than 250 patents were issued to U.S. universities. In 1999 alone, on the other hand, 3914 new license agreements were signed.

In 2006, Carnegie Mellon University helped establish 14 companies that were based on university technology, ranging from the green chemistry firm GreenOx and the training-simulation game developer SimOps, to the industrial control firm Industrial Learning Systems and robot toy developer Bossa Nova Concepts. At the University of California, more than 200 start-ups have licensed technology from the UC System's 10 campuses and three government labs. UC Berkeley alone licensed technology to more than 60 start-ups, and all in all, more than 100 products have been commercialized under UC Berkeley technology licenses. As of the end of fiscal year 2002, Berkeley had an active portfolio of 681 inventions, 417 U.S. and 303 foreign patents, and 207 license agreements. In the last five years, license revenues have aggregated to over $42 million, and licensing to "faculty" start-up companies has resulted in equity holdings in 16 companies.

So technology transfer programs bring millions of dollars a year to universities. So they're a good thing, right? Well, that depends on who you talk to. In many cases, start-ups that license technology that comes from university research are the professors themselves. In the University of California system, faculty and students have founded more than 25 percent of the start-ups. They benefit from research conducted by their grad students who get a passing grade, and not much else. That's clearly not the case in every start-up, but it has happened--and that's why most technology transfer programs have strict ethical and conflict-of-interest guidelines, like those at the University of California.

Still, there are lots of issues to be dealt with, the least of which involve issue such as intellectual property, open source, and the like. If some unpaid grad student or underpaid researcher comes up with a great idea and turns it into a product that makes them some money and the public who paid for the research benefits in the long run, well, what's wrong with that?

Posted by Jon Erickson at 12:14 PM  Permalink |

Thanks to the headline that Microsoft will be supporting legacy products, I can finally breath a sigh of relief. This is something I've been waiting for for a long time.

In my case, I'm hoping Microsoft is going back to supporting Word 6. Not that Word 6 was such a great wordprocessor, far from it. But it did have a saving grace--to the best of my knowledge, Version 6 was the last version of Word that provided conversion filters for WordStar. And as we all know, WordStar--at least the text-based WordStar 4.0 for DOS--was the best text editor ever written, emacs and xywrite notwithstanding. It's fast, it's small, it's mouse-free. You can sit in the middle seat on an airplane and type as long as your battery holds out, and not have to worry about elbowing people on both sides of you (unless you want to, of course).

But WordStar did (or in my case "does") have an irritating proprietary format that goes back to the days when it only supported 7-bit ASCII. As WordStar moved into the modern world, it added a "document" mode which set the high (8th) bit to force line wraps or word wraps or whatever. (Its "non-document" mode doesn't set the high bit, but then you didn't get any text formatting--italics, bold, and the like.) For details on this (if you care), check out the WordStar Resource Site where other cranks like myself tend to hang out.

The point is the proprietary format. You have to have a filter to convert WordStar documents into something the rest of the world recognizes. And thankfully, Microsoft provided this service up to and including Word 6. Subsequent to that version, however, WordStar filters disappeared from Word, stranding folks like me.

I like WordStar and want to keep using it. I don't have to think about the program when I'm typing, but can think about "what" I'm typing. (Okay, someone will be challenging my ability to "think and chew gum," let alone "think and type", but bear with me.) Everyone around me uses Word or Word-compatible programs. No one is going to switch to WordStar at this point.

Consequently, I end up starting out with a text file, which I work with in WordStar. I then convert it Word 6 to include any special formatting that's required, and pass it on to someone else. I should add that for a number of years, we were forced to using Word 6 because of a Quark quirk. The details and version numbers escape me now, but it had to do with moving Word docs into Quark and losing all formatting instructions unless it was a Word 6 doc. RTF, Word 97, and the like just wouldn't work. That problem might have gone away, but not my wanting to keep using WordStar.

Of course, Microsoft hasn't made it easy to maintain this level of non-compliance. Everytime I got a new computer over the years, I'd have to reinstall Word 6. Of course, subsequent version of Windows wouldn't let Word 6 run because of a missing SDM.DLL. Get that from an older computer, drop it in, and Word 6 runs like a champ--unless you lose your original Word 6 disks, which happened in recent move. So in this case, I ended up buying on eBay the complete Office Suite that included Word 6. What I got for my $5.00 purchase was 37 3.5-inch diskettes. Yes, 37 diskettes. How easy it is to forget what life was like before DVDs or Web-based downloads.

The bottomline is that I now have the WordStar/Word 6 system I like, although the folks in IT support refuse to speak to me. So you can understand why I was delighted to learn that Microsoft will be supporting legacy products. Now all I have to do is convince Microsoft that Word 6, and by extension, WordStar 4.0 qualify.

Posted by Jon Erickson at 12:55 PM  Permalink |

As I've said before, it is a full-time job these days keeping up with programming contests. In many cases, these contests are self-serving, at least for the sponsor. In other cases, they're worthwhile and beneficial. I'd classify IBM's just announced Accessibility ODF Coding Challenge as the latter.

The goal of the competition is to prepare computer science students to develop or adapt computer programs for people with disabilities, the maturing population, and non-native language speakers, so that they can more easily access, navigate, and use the Web and electronic office documents.

Interestingly, to qualify, contest entries must be based on the OpenDocument Format (ODF). While not having anything necessarily to do with accessibility, knowledge of ODF will be an important part of a programmer's toolkit over the coming years, since (according to Gartner) ODF will be required by 50 percent of governments and 20 percent of commercial organizations by 2010.

But back to assistive technology and the competition at hand: According to the World Health Organization, between 750 million and 1 billion of the world's 6 billion people have a speech, vision, mobility, hearing or cognitive disability. Addressing the computing needs of these individuals is unquestionably important.

"This contest brings together three critical ideas that have significant importance in the computer industry today: open standards, open source, and accessibility," said Bob Sutor, IBM Vice President, Standards and Open Source. "We hope that these efforts spark significant uptake in how we make our information and applications available to as many people as possible."

Prizes include everything from laptops and iPods, to t-shirts and a trip to Los Angeles (that's a prize?). Okay, the LA trip is actually to the or the 22nd Annual International Technology and Persons with Disabilities Conference, which just happens to be in LA.

Among other requirements, competitors need to:

• Create an extensible accessibility validation tool for word processing documents created in ODF format (.odt format). The tool you create should parse documents and report output that validates accessibility characteristics against the guidelines.
• Submit a short paragraph or two that identifies an ODF accessibility issue and explains how you would use technology to address the issue. If approved by the IBM panel of judges, you'll have the opportunity to create the code to support the solution proposed in your abstract.

The clock has already started ticking, so if you're a college student and want to make a difference, this contest is worth checking out.

Posted by Jon Erickson at 09:52 AM  Permalink |

Proven fact: If you hang around long enough, you can witness the world change around you. Take "grid computers," for instance.

It hasn't been that long ago that a "Grid computer" was a high-end ($8000-$10,000) laptop, best known for its semi-ruggadized matt-black magnesium case. Okay, maybe it was that long ago, the early 1980s. Still, you could make a credible argument that the Grid was, in fact, the first laptop computer. Moreover, in the spirit of what-goes-around-comes-around, it had features that we're seeing again today. For instance, the Grid didn't have a disk drive, but used 384 KB non-volatile bubble memory. Software could be loaded from a server and external floppy or hard disks. It also had a built-in modem, ran off of built-in batteries, and didn't need a cooling fan. But in the spirit of no-laptop-before-its-time, Grid Systems (the company that manufactured the Grid computer) wasn't able to make a go of it, eventually being acquired by (get this) Radio Shack.

These days, of course, a "grid computer" means something altogether different. Grid computing is a computing model that's based on multiple networked computers that model a virtual computer architecture that distributes execution across a parallel infrastructure. For obvious reasons, grid computing borrowed its moniker from electrical power grids back in the early 1990s.

It is probably no surprise that Dr. Dobb's has covered grid computing for a number of years, with articles such as:

• The Distributed Resource Management Application API by Frederic Pariente.
• Creating Java Grid Services by Aaron Walsh.
• Grid Computing and the Linda Programming Model by Rob Bjornson and Andrew Sherman.
• and so on.

Still, grid computing has generally been relegated to the research arena, although that's starting to rapidly change. Oracle has its Oracle RAC, the grid-enabled version of Oracle, and IBM has an array of IBM Grid Computing solutions. In short, there are real business reasons to adopt grids, particularly for large organizations.

But it is the smaller companies that are starting to make grid computing really interesting. ActiveGrid, for instance, is mixing up grids with Web 2.0, producing tools that let you accelerate Web 2.0 development. (Hear what ActiveGrid's Peter Yared has to say about this in his "Lightweight Architectures Meet the Enterprise" podcast.)

Likewise, Digipede Technologies is providing grid computing solutions on the Microsoft .NET platform. (Digipede's Robert Anderson and Dan Ciruli examine how SOA can scale using distributed computing technologies in the November 2006 issue of Dr. Dobb's Journal.)

Finally, a new version of the freely available Access Grid Toolkit, developed at the U.S. Department of Energy's Argonne National Lab, has just been released. The Access Grid Toolkit is software that uses audio, video, data and text to enable distributed researchers to work together as if they were at the same location.

Posted by Jon Erickson at 09:29 AM  Permalink |

I'm on the road a lot, in large part because it means I don't have to be in the office. Chicago, New York, Seattle, Kansas City, Helsinki. Wherever. It's not that I dislike being in the office, but it's just that other people in the office like it better when I'm elsewhere. My boss, for instance, says it is cheaper to have me on the road than behind a desk. If I need a car, its Rent-A-Wreck. If I need a hotel, it's Ho Jo's, with its free Wi-Fi, free breakfast, and on-site laundry facilities. Starbuck's? Ptui! It's Dunkin' Donuts everytime.

In a few weeks, for instance, I'll be in Boston attending Dr. Dobb's Best Practice Conference, which convenes at the Hynes Convention Center on September 11-14, and it looks to be a lot of fun. Speaking of fun, MIT's Rodney Brooks will be keynoting about "Having Fun and Changing the Future Through Technology," and anyone who has seen the movie Fast, Cheap, and Out of Control knows what he's talking about. (Yes, it's about robots--no, not my life on the road.) Other keynoters include Gary McGraw who will talk about "Software Security: Building Security In", and Murray Cantor who will examine "Agility, Governance and Productivity."

The conference tracks hit all the high points: Build and Deploy, C++, Design and Architecture, People, Projects and Teams, Requirements and Analysis, and Testing and Quality. Sessions will be led by the likes of Dr. Dobb's Journal Senior Contributing Editor Scott Ambler, and contributors such as Robert Seacord, Steven Kelly, Kirk Krauss, and Dan Saks, among others.

I hope to see you in Boston at Best Practices, and coffee is on me--assuming that you're up to walking down Boylston to the Dunkin' Donuts.

Posted by Jon Erickson at 09:57 AM  Permalink |

Congratulations to former Dr. Dobb's columnist and long-time BYTE senior contributing editor Jerry Pournelle who, along with co-author Larry Niven, will be awarded the L. Ron Hubbard Lifetime Achievement Award for Contribution to the Arts at the Writers and Illustrators of the Future Awardsceremony at the San Diego Air and Space Museum, on Friday 18 August.

Both Niven and Pournelle will be in attendance to receive their award as well as honor this year's contest winners, along with veteran NASA astronaut Richard Searfoss.

New York Times bestselling authors Niven and Pournelle each have extensive lists of both fiction and non-fiction works and have collaborated on several very successful novels, including The Mote in Gods Eye, Lucifer's Hammer, and Footfall--with virtually all of their story-telling focused on maintaining that very vital goal of man's drive into space. Beginning in 1980, they served on the Citizen's Advisory Council on National Space Policy to the President, which Pournelle chaired and for which Niven hosted meetings.

"We are very pleased to be able to acknowledge the accomplishments of Mr. Niven and Dr. Pournelle," said Hugh Wilhere, Public Affairs for Author Services, the literary agency sponsoring the event. "L. Ron Hubbard once referred to science fiction as the 'herald of possibility.' They both exemplify this in their life with their continued efforts to see these possibilities once suggested as science fiction be made into science fact."

Niven and Pournelle have been judges for the Writers of the Future Contest, the largest international writing contest in existence, since 1985. They are also very outspoken supporters of the contest. "It's been my experience," said Pournelle at last year's awards ceremony at the Seattle Science Fiction Museum, "that this program is the best there is--no fee to enter the contest, anyone who can write can enter, the winners keep their rights to the stories, and they get an all-expense-paid trip to be taught by some of the greatest science fiction writers there are and a great ceremony in their honor. Plus they get paid! I would estimate that at least 65-70% of the winners of the Writers of the Future Contest have gone on with careers in writing."

Posted by Jon Erickson at 03:49 PM  Permalink |

When I was in elementary school, this time of year meant a new pair of Converse Chuckies, stiff Levis, and a three-ring binder complete with pencil pack. When I grew up and became a school teacher, I traded the Levis for Dockers and Chuckies for Prozac. These days, none of this--Prozac notwithstanding--seems to make the grade in back-to-school.

Consider the Westinghouse Digital Electronics "ultimate" back-to-school package--a "complete Computer/TV system, which includes a laptop computer and 19-inch HD-Ready LCD TV." And why would you want your kid to have this? (Hey, we already know why the kid would want it.) Well, according to Westinghouse, this ultimate package will "help [the kids] achieve those coveted A+ grades and provide a great way to enjoy well-earned free time." Now speaking as a former student, I can relate to the well-earned free time. The A+ grades? Ah, that’s another thing.

The Westinghouse LTV-19w6 is a 19-inch HD-Ready 1440x900 Widescreen TV includes two HD Inputs and Integrated speakers. The display can be used as a TV, a computer monitor, and a display device for viewing DVDs and playing online and computer games. Of course, "having the two screens promotes increased productivity and allows for the multi-tasking environment the student of today demands." Hmm, sounds like promotional material for vocational school, turning fourth-graders into programmers. Of course, multiple monitors works for Bill Gates, who’s is a believer in multiple monitors . If it worked for Billy, it ought to work for Dick, Jane, and Sally.

Rey Roque, Vice President of Marketing at Westinghouse Digital Electronics, explains the benefits of the program: "The Westinghouse bundle we created for the back-to-school market provides students with the tools necessary to compete and stay ahead in the technology-driven academic race of the 21st century," he says. "The bundle provides a high-end, yet affordable solution that will enable students to maximize their academic pursuits and also utilize the system for entertainment purposes as well." To say nothing of Westinghouse third quarter sales reports.

To be fair, Westinghouse Digital isn’t alone in back-to-school scams. Best Buy lists the back-to-school college essentials as laptop, cell phones, dorm fridges, and iPods. No Chuckies, no books, no pencils. As for high-school students, Best Buy says all of the above, plus a digital camera. And for elementary kids, all they need are CD boomboxes that look like Disney cartoon characters.

Posted by Jon Erickson at 11:22 AM  Permalink |

Let’s see. I knew someone who knew someone who was related to a friend of a cousin of Antsy McClain, lead singer for the Trailer Park Troubadours. Does that make me a Troub groupie or official band member? I don’t know. I can’t keep this Six-Degrees of Separation straight. Luckily, computer scientists at the University of Pennsylvania School of Engineering and Applied Science are putting together some experiment to put this stuff to the test.

Formulated in 1969 by psychologists Jeffery Travers and Stanley Milgram, six-degrees of separation (sometimes called the "Kevin Bacon Game") tried to show that everyone was separated by only six connections from anyone else. (For more information on the Six-Degrees of separation, see "Simulating Small-World Networks" by Mary Lynn Reed, and "Discovering Relationships in Context" by Joe Celko.)

To test a number of standard network theories, Kearns and Penn and doctoral students Siddharth Suri and Nick Montfort gathered 38 Penn students at a time to play a game of color selection on networked computers. The game required each of the students to choose a color that did not match the color of any person who was immediately connected to him or her in the network. The researchers changed the patterns of the networked connections -- that is, who was connected to whom -- in ways that corresponded to the theoretical models.

The findings have implications for many forms of social interaction, from disaster management to how many friends connect to your MySpace page. The Penn researchers have found that some of the simplest social networks function the most poorly and that information beyond a "local" view of the network can actually hinder the ability of some complicated social networks to accomplish tasks.

"Travers and Milgram's classic six degrees of separation experiment was one of the first large-scale attempts at studying a human network, but almost 40 years later the interaction between social network structure and collective problem solving is still largely a matter of theoretical conjecture," said Michael Kearns, a professor in Penn's Computer and Information Science Department. "Our goal was to initiate a controlled, behavioral component of social network studies that lets us deliberately vary network structure and examine its impact on human behavior and performance.

"This coloring problem models social situations in which each person needs or wants to distinguish his or her behavior or choices from neighboring parties", Kearns said. "A good modern example is choosing a ringtone for your cell phone. You don't want to choose one that is the same as a family member or a colleague in the next cubicle. But if there's a limit to the number of available ringtones, you may have a difficult collective problem of coordination. In our experiments, many of the networks were quite dense with connections, and the colors were very few, so they were hard coloring problems."

The first three of the tests began with a circular structure, like a 38-member daisy chain. These networks represent a "small world" network that models a local area, such as a small group in a single town, mixed with the occasional cross-town relationship. The simplest of these, a single circular chain, was actually the most difficult for the subjects, but the more connections made across the circle, the faster the test was completed.

The fourth model represented a more engineered or hierarchical structure: a circle with two individuals that have many more connections than the rest. This model proved the easiest for the subjects: once each of the two "commanders" picked a color, everyone else unwittingly fell into place, despite the fact that nobody was told anything about the network structure or could see anything but the colors of immediate neighbors.

The last two tests studied so-called preferential attachment models, well studied networks in which many parties are highly connected. These models look something like maps of the Internet. Unlike the more circular models, here Kearns found that a complete view of the color selections across the entire network actually led to confusion among members of the network.

"We see that social networks with more connectivity aren't necessarily more efficient, but that it depends strongly on the collective problem being solved", Kearns said. "Less connectivity and less information about the network can sometimes make the problem easier. But now we have an experimental framework in which we can systematically investigate how social network structure influences actual human performance."

Posted by Jon Erickson at 09:22 AM  Permalink |

Ed McBain would be all over this, if he were still around. Lawrence Block probably has a book in the works about it. And Ian Rankin's John Rebus is already a texting fool.

Yes, mobile phone text messaging has moved from college and high-school campuses to criminal investigations, and surely mystery writers can't be far behind.

What led to this was a 2002 murder investigation in the UK where an alibi was broken based on evidence that Stuart Campbell--not the victim Danielle Jones--had sent crucial messages from her phone. Campbell's alibi was broken based on text analyses which revealed that the texts had not been written by Jones, but that they had been faked to deflect suspicion from Campbell. There were a number of differences in the texting styles between the victim and Campbell.

With this background, psychologists at the University of Leicester's Forensic Section of the School of Psychology are investigating texting language to develop more precise techniques for identifying individuals by their texting style. The six-month study will assess similarities and differences in texting style, between texts sent by individuals and within and between networks of people who frequently text one another. The researchers are inviting ordinary people -- murderers need not apply -- to help them with the study by completing an anonymous on-line questionnaire. Although forensic authorship analysis is a growing area of research, this is the first study to focus on mobile phone texting.

The researchers are looking to recruit at least 100 volunteers who will each be asked to contribute 10 text messages. The study is anonymous, so participation is confidential and people have the right to withdraw from the study at any time if they so wish. The texts contributed to the study will be analysed using linguistic and statistical techniques and results submitted to the Journal of Speech Language and the Law.

"We are looking for volunteers to participate in a unique forensic linguistic study, assessing similarities and differences in text messaging style," says forensic linguist Tim Grant who, along with forensic researcher Kim Drake, is conducing the study. "This piece of Leicester research will have important applications for forensic investigation -- for example, in the past text messages have been used as an alibi to murder. Being able to say who wrote a particular text message sent from a particular phone has many potential forensic applications.

"As texting is both a relatively new mode of communication and a particularly informal way of using language, there is not a strong expectation that texters will follow linguistic conventions. This freedom therefore allows for significant individual differences in text messaging style, and this can be used to identify the text's authors. Forensic authorship analysis has also been used in cases involving disputed confession, the sending of abusive or threatening emails or letters and in cases of copyright infringement."

The linguistic analysis will concentrate on individuals' texting strategies and vocabulary; for example, the examples (1)-(4) below show four different texting styles and vocabulary that may be analysed.

1. "Came to see K bumped into B 2."
2. "Shall we get last bus to station"
3. "Pls cn i meet. Tried fonin 4 u."
4. "C U L8R"

According to Grant, individual differences can often be seen by just looking at the text; for example, it is likely that the first two texts were created using a predictive texting function. The reason for this supposition is the resultant word deletions (for example, "Came" rather than "I came" or "get last bus" for "get the last bus") that can be seen in texts 1 and 2. On the other hand, the last two texts use more abbreviated text language with vowel deletions ("Pls"), as well as letter and number homophone substitutions ("C" for "see" and "8" for the middle syllable of "later"). What's important to note is that texting strategies are likely to remain relatively constant within individuals and, along with the actual vocabulary used by an individual, may help identify the writer of a text.

The statistical analyses will use methods developed for linking crimes committed by serial offenders. These methods involve the use of, what are known as, coefficients of similarity (an example of which is the Jaccard's coefficient) which will be used to identify the salient linguistic features of texts by a single individual. That is, it will identify the major differences in the texting strategies and vocabulary across individuals. The same statistical techniques will be applied to identify differences between groups.

Posted by Jon Erickson at 09:18 AM  Permalink |

If nothing else, we can always count on our elected officials in Congress to do what's best. The question that comes to mind, of course, is "best for whom?".

Take DARPA, for instance. You can't deny that the Defense Advanced Research Projects Agencyhas lived up to its charter as being the "technological engine" for the Defense Department in particular and society in general. Why, if it weren't for DARPA, which is credited for launching Internet technologies back in its ARPA days, you might not be reading online what you're now reading. Then there's Global Positioning Systems, Multics (on which modern operating systems owe a big thanks), virtual reality, hypermedia, and a host of other technologies that we now take for granted--all of which have their roots in DARPA. In short, DARPA is really about radical innovation and looking beyond today’s immediate needs, instead looking at tomorrow's requirements.

Alas, Congressional leaders in the U.S. apparently don't see it this way, at least when it comes to funding DARPA projects that may be important down the road, as evident by the U.S. Senate Appropriations Committee slashing DARPA budgets. In particular, programs that would suffer cuts include:

• Responsive Computing Architectures -- cut by $3.9 million.
• Security-Aware Systems -- cut by $3 million.
• Automated Speech and Text Exploitation in Multiple Languages -- cut by $6.5 million.
• Integrated Cognitive Systems -- cut by a whooping $60 million.
• Learning Locomotion and Navigation -- cut by $3.8 million.
• Improved Warfighter Information Processing -- cut by $7 million.

Moreover, the committee decided to cut DARPA's Computer Science Study Group -- established to introduce young faculty to DOD-oriented issues in computer science -- from the requested level of $6.6 million to $3 million.

But it isn't like Congress is wanting to save money. After all, it has allocated:

• $1 million for the Waterfree Urinal Conservation Initiative.
• $3.1 million to convert a ferry boat into a crab restaurant in Baltimore.
• $6.4 million for a Bavarian ski resort in Idaho.
• $13,500,000 for the International Fund for Ireland.
• $550,000 for the Museum of Glass in Washington.
• $11 million for a private pleasure boat harbor in Cleveland.
• $500,000 for the Sparta Teapot Museum in North Carolina.
• $320,000 to purchase President McKinley's mother-in-law's house. Funds to rehabilitate the South Carolina mansion of Charles Pickney, a Framer of the Constitution, even though the house was built after he died.
• $2.7 million for a catfish farm in Arkansas.
• $500,000 to build a replica of the Great Pyramid of Egypt in Indiana.
• $13 million for an industrial theme park in Pennsylvania.
• $500,000 for a museum to honor former Secretary of State Cordell Hull.
• $33 million to pump sand onto the private beaches of Miami hotels.

According to The 2006 Pig Book, the total cost of boondoggles such as these is a record $29 billion. By comparison, in 1991 earmarks added just $3.1 billion to the budget.

Think about this: If we would have had the kind of Congressional leaders 50 years ago that we have today, we likely wouldn't have the Internet as a vehicle to kick Congress around.

Posted by Jon Erickson at 09:36 AM  Permalink |

It's not everyday you get to give your alma mater a pat on the back. While a team from the University of Alberta didn't win the curling championship or debate competition, it did win the first AAAI Computer Poker Competition held in conjunction with this year's American Association of Artificial Intelligence conference in Boston.

The U of A computer program defeated all other programs in a two tournament format of one-on-one Texas Hold 'Em. The U of A poker bot won every match it played and amassed by far the most virtual money of any competitor. All in all, there were five competing bots, including three from collegiate teams and two from individuals:

• "Hyperborean" written by Michael Bowling, Martin Zinkevich, Darse Billings, Nolan Bard, Morgan Kan, Michael Johanson, Robert Holte, Jonathan Schaeffer, Neil Burch, Carmelo Piccione, and Finnegan Southey, of the University of Alberta.
• "GS2" written by Andrew Gilpin and Tuomas Sandholm from Carnegie Mellon University .
• "Monash BPP" designed by Ann Nicholson, Kevin Korb, and Steven Mascaro of Monash University (Australia).
• "BluffBot" written by Teppo Salonen from Irvine, California.
• "Teddy"by Morten Lynge in Ikast, Denmark.

Each bot won at least one series.

The game of one-on-one Limit Texas Hold'em was played in a pair of computer tournaments. The first event used a normal pace of about one second per decision, as seen in games played by humans. The second event allowed a much slower pace, in the hope that it might produce a higher level of play. Somewhat surprisingly, the winning program from Alberta made every betting decision instantaneously, even in the slower event.

The AAAI event is the most intensive competition there is for poker programs. To reduce the effects of luck, more than a quarter million games of poker were played in the two tournaments. To make the results even more reliable, every series of deals was played twice, with both competitors getting a chance to play each side of the cards. This is possible because programs can be restarted with no memory of past events. The "duplicate matches" ensure that both programs have nearly equal opportunities, despite the lucky outcomes that can occur in each game.

To level the playing field, every bot was run on an identical machine. The greater degree of control let participants force the bots to "forget", resulting in better results overall. The competition was run on 16 Windows machines located in the at the University of Alberta. There was one server machine, and 14 client machines ( one extra machine in case one of problems). The computers were 3.4-GHz P4 machines running Windows XP Professional, with 1 GB of RAM, and an 80-GB hard drive.

"Poker is a game that involves skill, chance, and many forms of uncertainty", said Alberta's professor Jonathan Schaeffer who I interviewed last year in this brief podcast. "It is a great problem for artificial intelligence, and we stand to learn a lot from competitions like this."

"We've been writing good poker programs for many years", added Darse Billings, the lead designer for the Alberta team, "but we weren't overly confident, because there is still a lot of room for improvement. Poker is a nice well-defined problem for studying some truly fundamental issues, like how to handle deliberate misinformation, and how to make intelligent guesses based on partial knowledge," explained Billings. "Good solutions in this domain could have an impact in many other computer applications."

Posted by Jon Erickson at 08:47 AM  Permalink |

It used to be that "clicks" were irritating sounds, not core business objectives. But that was before the commercialization of the web.

Nowadays, being able to generate clicks, or more precisely "click-throughs," can make the difference between eating at Chez Panis or McDonald's. And that's just one reason why research into click-through patterns and browsing behaviors is vital to online business and therefore a key focus of research.

For instance, Eugene Agichtein, Eric Brill, Susan Dumais, and Robert Ragno, members of Microsoft Research's Text Mining, Search, and Navigation Research Group, have been investigating user interaction with search engines. What they've found, and what they report in their paper "Learning User Interaction Models for Predicting Web Search Result Preferences," is that accurate modeling and interpretation of user interactions with a search engine can significantly improve search-result ranking, the detection of "click-spam," Web search personalization and, ultimately, the overall Web search experience. While user interactions with the Web search engines are plentiful, new robust techniques are required to understand the relationship between user interactions and result quality.

Likewise in their paper "Improving Web Search Ranking by Incorporating User Behavior," the research team discusses utilizing the discovered patterns of user behavior to improve algorithms that rank search engine results. This work demonstrates how to add a new rich set of user behavior features into ranking for large relevance gains.

"Most search engines today use a somewhat two-dimensional approach, matching user queries with the content and link structure of Web pages to return a list of results," says Agichtein. "We’re looking at how to add a third dimension--the users themselves--to improve the search experience. By examining click-through and browsing patterns across a large number of users, we are able to learn a great deal about how people interact with search technologies and can thereby improve our accuracy dramatically."

In other words, not to put bad puns in Agichtein's mouth, we can learn what clicks with users, and what doesn't.

Posted by Jon Erickson at 09:26 AM  Permalink |

"Concurrency in the classroom" is not about Instant Messaging with your pals while listening to your professor drone on about the relevance of Chaucer or explain how peep-hole optimizers work. No, concurrency in the classroom is about the emergence of multi-core processors, the importance of multi-threaded programming, and making the most of resources at hand. It's about the future of software development, in other words.

To drive this point home, Intel has launched global programs to prepare university students for a new paradigm of software development as we transition from single-processor engines to those that will have multiple cores and threads--a shift that transforms software design and requires a new way of thinking. As part of this education program, Intel is providing 45 of the world’s top universities with expertise, funding, development tools, educational materials, on-site training, and sustained collaboration to incorporate multi-core and multi-threading concepts into their computer science curricula.

Participating universities include Carnegie Mellon University, Cornell University, Georgia Institute of Technology, University of Michigan, and University of Washington, along with institutions across Brazil, China, India, Mexico, Russia, Taiwan, and several European countries. The first courses will be offered during the fall term this year and Intel expects hundreds more universities to participate in 2007 and beyond.

"To usher in a new generation of computing technology and bring creative new products to market, it’s crucial to educate tomorrow’s software developers to architect, develop and debug the next generation of software for modern, multi-core platforms," said Renee James, corporate vice president and general manager of Intel’s Software and Solutions Group. "The full potential of multi-core based systems to deliver great performance and expanded usages is unleashed when software is designed to take advantage of the full capabilities of the machine. Working with the world’s best universities, Intel is creating the future for performance computing."

"Intel’s support in multi-core education is critical for two reasons," said Karsten Schwan, professor of College of Computing, Georgia Institute of Technology. "First, getting early access to advanced technology and new equipment is something that always excites students. Second, companies like Intel have a perspective that looks beyond research to see the broader potential for technology."

The curriculum conveniently provides an introduction to Intel multi-core architecture and teaches computer science students how to achieve maximum performance of their programs on threaded, multi-core, and multi-processor systems using Intel compilers and threading tools. It also covers the importance of parallelism, threading concepts, threading methodology and programming with threads (Windows, OpenMP, PThreads).

Posted by Jon Erickson at 08:26 AM  Permalink |

Maybe it's just the idea of spending summer days at the beach, or re-runs of Jaws, or just maybe fact the ocean isn't nearly as much fun in January. For whatever reasons, researchers this summer aren't just heading to the ocean--they're going under it.

For instance, submarines built by students teams from the UK, Norway, France and Spain are right now taking part in the first ever Student Autonomous Underwater Challenge Europe competition at Pinewood Studios where the underwater scenes in Harry Potter and the Goblet of Fire and the James Bond movie Doctor No were filmed. Modeled on a similar competition in the U.S., SAUCE has been designed to advance Autonomous Underwater Vehicle (AUV) technology.

To enter the competition, team had to design and build a submersible vehicle capable of carrying out a range of underwater tasks, such as swimming through a submerged gate, dropping markers onto targets and hitting a floating object. The challenge is made even harder in that the submarine has to be autonomous, meaning that it carries out these tasks without being directed by a human operator.

For example, the submarine built by the Bath University Racing Submarine Team (BURST) is about 1.5 metres long and 1 metre wide, has been built using parts from computers and other everyday equipment, including a controller from a Sony PlayStation. It is powered by four flippers which, acting like a fish, both steer and drive the vehicle along. It also includes two webcams to help the vehicle recognize the targets and navigate through the water.

Meanwhile, in Monterey Bay, California, a fleet of undersea robots are working together to make detailed and efficient observations of the ocean without the aid of humans. The Adaptive Sampling and Prediction (ASAP) project, conducted by a multidisciplinary team of oceanographers, marine biologists, and researchers in control and dynamics, involves underwater robots, known as "gliders," that will take the ocean's temperature, measure its salinity (salt content), estimate the currents and track the upwelling.

Inspired by the behavior of schools of fish, a group at Princeton University has created algorithms that allow the gliders to self-choreograph their movements in a series of rectangular patterns. The patterns span a large volume that the scientists have mapped in Monterey Bay. On a day-to-day basis the control algorithms let the gliders make decisions independently about how to alter their course -- without any input from humans. This day-to-day autonomy enables the gliders to move according to the organized patterns, even as they are buffeted by strong currents.

In addition to gliders, the ASAP ocean-observing network also includes research ships, surveillance aircraft, propeller-driven vehicles, fixed buoy sensors, and coastal radar mapping. The ASAP field operation is one of four marine research initiatives taking place during the summer in Monterey Bay.

Posted by Jon Erickson at 09:35 AM  Permalink |

Distributed computing is all about the sum of some parts being greater than the whole, or something like that. An easy illustration of this is the University of California at Berkeley's SETI@home project which uses Internet-connected computers in the Search for Extraterrestrial Intelligence. Stardust@home is a similiar distributed computing project.

Stardust@home is a space science project in which volunteers from the general public help scientists locate particles from distant stars. The particles were captured by the spacecraft Stardust and are embedded in collector plates made of aerogel, which were carried by the spacecraft. The aerogel plates, and the particles within them, were returned to Earth in a sample return capsule on January 15, 2006. The purpose of Stardust@home is to find the particles within the aerogel plates.

Stardust@home is looking for interstellar dust grains -- particles from distant stars that are now free-floating in space. They are extremely small, the largest being only a few microns in diameter, and are therefore very difficult to detect. In the past decade, spacecraft traveling beyond the orbit of Mars discovered that a steady stream of such particles is continually moving through our solar system as the solar system itself travels through the Milky Way galaxy. Using a Web-based virtual microscope, volunteers are looking for the fewer than 50 grains of submicroscopic interstellar dust expected to be there.

On August 1, the first day the project was open for business, Stardust@home drew nearly 115,000 volunteers searching for these interstellar motes within the millions of scans of the Stardust Interstellar Dust Collector that eventually will be put on the Internet. Traffic was so heavy, in fact, that the system was brought to a crawl. According to a note on the web site, project engineers were scrambling to add more servers.

The digital microscope scanner was developed by Stardust@home director Andrew Westphal, a UC Berkeley senior fellow and associate director of the campus's Space Sciences Laboratory, based on his previous experience scanning glass detectors for cosmic ray particles.

Interestingly, the Stardust@home project uses the Amazon Simple Storage Service (Amazon S3) to store and deliver the tens of million of images that represent the data collected from the dust particle aerogel experiment.

As for dust, if Stardust@home needs a little extra, come on over to my office--there's plenty to spare. But bring your own Endust.

Posted by Jon Erickson at 11:58 AM  Permalink |

Different things mean different things to different people. Take "underground radio," for instance. To me it means alternative music, low power, and an outlaw mentality. But to the folks at Los Alamos National Lab, underground radio means just that--through-the-earth communications.

Developed by the Lab as an emergency communication for first responders, rescue and security teams, underground miners, and the like, through-the-earth communication provides two-way voice and text interaction and provides a data link to robotic machines. Unlike conventional radio signals which are line-of-sight limited, the technology can send information through the Earth for several miles, or through concrete, steel, rock, and the like.

Underground radio uses very low frequency (VLF) electromagnetic radiation and digital audio compression technology to carry voice and text data. The VLF signals also can transmit tracking and location data for radio users in the case that they are unable to respond.

"This is a technical solution to the problem of voice communication in underground areas. It is also inexpensive to build," said David Reagor, the principal investigator of the Los Alamos team who originally developed the technology.

To make the underground radio more widely available, the Lab has licensed the technology to Vital Alert Technologies.

"The new technology is a breakthrough in digital and wireless communications," said Joe Miller, of Vital Alert. "As a pre- and post-emergency warning, evacuation, and rescue communication system, it solves RF [radio frequency] radio failure problems and eliminates systems downtime complications in difficult environments such as subways, tunnels, skyscrapers, and mines. The new technology will also greatly enhance the ability of mining companies to protect their workers."

Posted by Jon Erickson at 11:43 AM  Permalink |

One man's redundancy is another man's multimodal, or so I've come to learn. What brought this to mind is the National Institute of Standards and Technology's release of its Multimodal Biometric Application Resource Kit.

According to NIST, most biometric security systems are "unimodal" in that they rely on a single distinguishing physical characteristic--a fingerprint, for instance--for authenticating identity. But relying on a single biometric feature can lead to problems--poor illumination can make a face image unrecognizable, dirty or damaged sensor plates can affect fingerprint equipment. Luckily redundancy can be built into systems to check multiple sources of information. Such redundant systems are referred to as "multimodal" because they examine fingerprint, face, and iris data. But most biometric equipment, including the sensors that capture data and the database that stores the information, are not interoperable. Organizations must either purchase a complete system or develop middleware--custom integration software--to link together applications.

Enter MBARK, an open source Multimodal Biometric Application Resource Kit. Originally developed as a tool to build a large database of face, fingerprint, and iris images for performance testing of biometric systems, MBARK has evolved into a standardized, flexible middleware package that lets you plug in sensors from different manufacturers. The package, which includes example applications and public-domain source code, can help reduce the complexity of building multimodal biometric applications. MBARK also can used to develop standards and tests for biometric system interoperability and usability.

Posted by Jon Erickson at 09:35 AM  Permalink |