DocBug

Art history, optics and scientific debate 

Our Chief Scientist, David Stork, has been doing some side research the past few years in art history. In particular he's been assessing a theory that artist David Hockney presents in his book "Secret Knowledge": that artists as early as 1430 have secretly used optical devices such as mirrors and lenses to help them create their almost photo-realistic paintings.

The theory is fascinating. Art historians know that some master's used optical devices in the 1600's, but Hockney and his collaborator Physicist Charles Falco claim that as early as 1430 the masters of the day used concave mirrors to project the image of a subject onto their canvas. The artist would then trace the inverted image. This alone, Hockney and his supporters claim, can account for the perfect perspective and "opticality" of paintings that suddenly appear at in this time period.

If the theory itself is fascinating, I find Stork's refutation even more interesting. Stork's argument is based on several points. First, he argues, there is no textual evidence that artists ever used such devices. Hockney and his supporters counter that the information was of course kept as a closely guarded trade secret, and that is why there was no description of it. It isn't clear how these masters also kept the powerful patrons whose portraits they were painting from discussing their secret. Stork's second argument is that, quite simply, the paintings aren't all that perfect perspective after all. They look quite good, obviously, but if you actually do the geometry on the paintings Hockney presents as perfect you see that supposedly parallel lines don't meet at a vanishing point as they would in a photograph. And third, Stork points out that the methods Hockney suggests would require huge mirrors to get the focal lengths seen in the suspected paintings: mirrors far far larger than the technology could create at the time.

My analysis is a little unfair to Hockney as I've only seen Stork's presentation, but I must say I'm impressed with his argument. Hockney's argument is quite media-pathic. It's a mystery story that wraps history, secrecy, geniuses, modern science and great visuals all in one -- no wonder it's captured people's attention! Unfortunately, I expect Stork's right about one of the less fun aspects of the theory. It's also probably dead wrong.

For those interested, a CBS documentary on Hockney's theory will be rebroadcast this Sunday, August 3rd, on 60 Minutes.

References:

• Was It Done With Mirrors? (CBS 60 Minutes, 16 January 2003)
• Art & Optics (Conference Web Site, 1 December 2001)
• Further Reflections (Stork's objections from the Art & Optics conference)
• Optical Projection in Renaissance Art: Did the Great Masters Cheat? (Interview with David Stork, Advanced Imaging, June 2003)
• FAQs about claims by David Hockney and Charles Falco on the purported use of optical devices by early Renaissance painters (David Stork's page on the subject)

I got the horse right here... 

The story sounds like something out of The Onion, or maybe a dystopian science fiction short story. As reported widely in the news yesterday, the Pentagon has been planning an electronic futures market for analysis of foreign affairs. The idea is to create a market where people can anonymously bet on things like whether the US will reduce troop deployment in Iraq by year's end, or whether Arafat will be assassinated. The current odds on the bet, so the argument goes, best reflects the actual probability given everything the collected thinkers know. Policy-makers could then use the probability to know where to focus their attention.

By today the firestorm had swept Washington and the Pentagon announced the project has been canceled. Apparently congressmen were not completely aware of what had been planned, in spite of the general plan being up for many months on DARPA's web site and mention of the project in a March New Yorker article.

I can't help but feel sympathy for Robin Hanson, the George Mason University Economics professor who has been spearheading the project. Critics were quick to describe the project as a marketplace where terrorists and mercenaries could make money by betting some horrific event would happen and then causing it. But as Hanson describes in interviews and on his Web site, the idea is more that professors, armchair analysts, and frequent travelers from all walks of life would combine their on-the-ground expertise to come to conclusions even the most expert intelligence worker in Washington wouldn't be able to reach. But interested as I am by the concept I just can't see it working for a number of reasons:

• First off, critics are right in thinking there's something morally repugnant about the whole plan. The US government should not be hosting a Website dedicated to graveyard gambling, regardless of whether it would actually encourage terrorists to make money from their exploits. (Personally I don't believe there's any chance a halfway-competent terrorist would bet on his own success on a Website run by DARPA, regardless of their assurances that betters will remain anonymous.) In fact, the whole plan bears a striking resemblance to the Assassination Politics plan devised by the Cypherpunk-anarchist Jim Bell. The plan describes how communities of individuals could put a price on a government official's head simply by donating a prize to whoever can predict the exact date of that person's death. (Bell is currently facing a 10-year prison term for harassment of a Federal officer.)
• There has been a lot of talk about how the attack on the World Trade Center could have been avoided if all the information that was distributed around the country could have been brought together in one place. That may be true, but an ideas futures market wouldn't have helped. What we needed was more analysis and communication; the marketplace is too abstract and mediated to allow anyone to put the pieces together. An ideas market won't bring together the CIA agent studying Al Quaeda and the Florida flight-school instructor because neither would have enough pieces of the puzzle to realize what they were looking at. Marketplaces are additive; intelligence requires synthesis.
• Even if the market was a reasonable risk-estimation system, it's not clear what the government could do with that information. As Bloomberg.com, points out, the market would be quite noisy, similar to the stock market. As we've seen from the constant rainbow of alerts we've gone through over the past two years, unspecified and uncoroborated threats aren't all that useful when you're trying to set up a defense.

Update: According to futures sales on Tradesports.com, John Poindexter's chances of keeping his job after this uproar are around 70%.

References:

• Idea Futures (Robert Hanson, first version 1996)
• DARPA Futures Market Applied to Prediction Program
• Decisions, Decisions (The New Yorker, 24 March 2003)
• Pentagon to start futures market for terror attacks / WAGERING: Concept is based on a legitimate theory (San Francisco Chronicle, 29 July 2003)
• Pentagon Prepares a Futures Market on Terror Attacks (New York Times, 29 July 2003)
• Pentagon's threat-bet program to be canceled under fire from all sides (Associated Press, 29 July 2003)
• All Bets Are Off on Government Terrorism Futures (Caroline Baum, Bloomberg.com, 30 July 2003)
• Assassination Politics (Jim Bell, 3 April 1997)
• Cypherpunk Sentenced to 10 Years (Wired News, 27 August 2001)

NPUC 2003 Trip Report 

A couple weeks ago I attended the New Paradigms in Using Computers workshop at IBM Almaden. It's always a small, friendly one-day gathering of Human-Computer Interaction researchers and practitioners, with invited talks from both academia and industry. This year's focus was on the state of knowledge in our field: what we know about users, how we know it and how we learn it.

The CHI community has a good camaraderie, especially among the industry researchers. I suspect that's because we're all used to being the one designer, artist or sociologist surrounded by a company of computer scientists and engineers. Nothing brings together a professional community like commiseration, especially when it's mixed with techniques for how to convince your management that what you do really is valuable to the company.

One of the interesting questions of the workshop was how to share knowledge within the interface-design community. Certainly we all benefit by sharing knowledge, standards and techniques, but for the industry researchers much of that information is a potential competitive advantage and therefore kept confidential. Especially here in Silicon Valley, that kind of institutional knowledge gets out into the community as a whole through employment churn, as researchers change labs throughout their careers.

Here are my notes from several of the talks. Standard disclaimers in place: these are just my notes of the event and subjected to my own filters and memory lapses. If you want the real story, get it from the respective horses' mouths.

Mike Kuniavsky (Adaptive Path): Reverse the polarity

Kuniavsky's first point was that the dot-com era brought new awareness of the need for usability testing and design. What was once an after-thought is now recognized as a primary driver for getting and keeping customers. He describes it as a crisis of purpose that has brought many new players together, including people involved in design, information science, HCI, ethnography and marketing. Techniques have also shifted from those that study single users (psychology, cognitive science) to those that study groups (marketing).

He then described the methodology used at his usability consulting company, which starts with an analysis of all stake-holder needs. These stake-holders include users of the system or Web site, as well as management, tech support, system maintainers, etc. Through this more holistic approach their solutions are more likely to have long-term benefit for their client because they can be more confident the right problem is being solved.

Colin Johnson (EyeTools): Understanding Users Through Their Eyes

EyeTools is a spin-off company from Stanford that provides user-testing of websites and software using their eye-tracking equipment. Their final reports show exactly where visitors are looking when they read a site, displayed either as they paths their eyes travel or as heat-map displays overlayed on top of the main page. They've shown some clients that not only are users not seeing the important marketing blurb on a page, but that they don't even notice when the blurb is changed to complete gibberish.

One preliminary but interesting effect they've seen is that people have very different viewing patterns when asked to "say out loud what you're looking at" than when they simply read naturally. If confirmed in a full study, that would imply the typical stream-of-consciousness commentary method of evaluating an interface may not get at natural usage patterns.

Bonnie John (HCI Institute, CMU): Data collection in support of US: Moving HCI to Level 5

Many Engineering fields have at least flirted with the Capability Maturity Model, which describes a process as being in one of five stages of maturity: Initial (chaotic), Repeatable, Defined, Managed, and Optimizing. The idea is to standardize the entire process such that methods are repeatable and testable, and the results of any change are predictable. Dr. John's argument was that we, as a field, should strive towards levels 4 and 5.

Arguments against this idea came fast and furious. The first was that our work would become controlled by the "tyranny of the quantifiable." Some research can not be easily quantified, but this does not make the research any less important. It was also pointed out that in HCI it's very hard to tell what to measure, and doubly hard for politically-charged applications such as education. The compromise position was that especially critical niches might still benefit, such as usability of interfaces for NASA's Martian lander. It was also pointed out that the field knows a great deal of rules about ergonomics, but such information has not yet been collected into a single volume.

Genevieve Bell (Intel): Lessons from the field: or, how not to do ethnography in 5 easy lessons

Dr. Bell had just gotten back from six months in Asia, studying daily interactions with technology in urban centers in India, Malaysia, Singapore, Indonesia, China and South Korea. The talk had lots of interesting insights about both Asian and Intel culture; here are a few tidbits:

• Asia is huge and extremely varied. This should come as a surprise to no one, but...
• ...did come as a surprise to some at Intel.
• Religion is hugely important in some parts of Asia. For example, mobile phones in Malaysia and all around the Middle East have a much-loved feature: push a button and they point towards Mecca.
• In South Korea mobile phones are used to manage complex social networks to set up blind dates.
• Technology in Asia is used as a way to show your identity more than here in the US. Phones often are attached to stuffed animals or toys with which the owner identifies. She also showed cellphone covers from Malaysia with pictures of airplanes flying into the World Trade Center, and covers with Bin Laden's picture and a quote from the Koran encouraging martyrdom. All hung out for sale in a normal retail store, wrapped in a Hello-Kitty plastic bag.
• In the US, mobile technology is still more predominantly information, with communication secondary. It is the reverse in Asia.
• Intel culture is such that people really want a big take-away message on a single PowerPoint slide. That's how ideas have an impact there (and many other modern businesses I might add): have it on a slide that others can steal and put in their own talks. These days ideas are traded around like currency, three bullet-points per thought.

Merissa Mayer (Google): The science behind Google's UI

Google has six basic steps to designing and testing their interfaces:

1. Use existing bodies of research. For example, they mined all the psychology literature about what kind of fonts and color combinations were most readable.
2. User-test existing products. Initial tests look at around 15 users to get feedback. Testing in early 2000, for example, revealed that users kept waiting "for the page to finish loading" because the layout is so sparse. This led to Google adding the copyright notice at the bottom, as an indicator that the page was complete.
3. Internally test products in development. Ms. Mayer kept talking about feedback from "googlers," clearly an extremely valuable internal testing resource. Eventually I asked what exactly googlers was, and whether it was some specific Q&A group. She laughed and said no, it was just the company-wide mailing list. Whenever they have a new interface or feature to test, they just send the URL to the entire company, and the next day find some 100+ emails with feedback. Google clearly has a fun and open company culture; it shows in everything they do. One nice quote: "Delightfulness has a value. People like to know the personality of the people behind a service."
4. Experiment on live site. Google will sign up sets of users (e.g. broadband users in a specific area) to try out a beta site. They used this technique to try out displaying thumbnails of resulting web pages. The feature wasn't well-received: the thumbnails forced text down, so you could only fit four results above the fold instead of 5-7. Also, thumbnails were generally not useful unless you had seen the page before. However, they are trying the thumbnail experiment in Japan now, where usage is different. Due to differences in the keyboard, the Japanese use Google much more often to go to simple URLs like Yahoo!, rather than type the URL in directly.
5. Analyze the logs. Google is constantly checking their logs to check things like how often the spell-checker feature is used.
6. Lather, rinse, repeat. Iterate after each step. Google does a site-wide test once a month, and also tests other products like advertising products or the Google Toolbar.

As an interesting side-note, she says the purchase of Blogger wasn't as strategic as it is seen to be outside of Google. They were purchased mainly because they were good engineers who needed resources, and it seemed to be a good match. There might be a move in the future to have them help with a service where blog owners can resell targeted ads on their site, should they want to monetize their publications.

Electronic Voting Gets Burned 

Electronic voting is getting slammed this week. First, Dan Gillmor's Sunday Column took election officials to task for not insisting on providing physical paper trails that can be followed should the results of an election be in doubt. Then on Wednesday several computer security experts at Johns Hopkins University and Rice University published a scathing analysis of the design of the Diebold AccuVote-TS, one of the more commonly used electronic voting systems, based on source code that the company accidentally leaked to the Internet back in January. Exploits include the ability to make home-grown smart-cards to allow multiple voting, the ability to tamper with ballot texts, denial of service attacks, the potential to connect an individual voter to how he voted, and potentially the ability to modify votes after they have been cast. The New York Times and Gillmor's own blog have since picked up the report. Diebold has since responded to the analysis, but at least so far they haven't addressed the most damning criticisms.

There are several lessons to be learned from all this:

1. Paper-ballot voting systems can fail and can be rigged, as anyone who was awake in November 2000 can attest. However, they have two advantages over electronic systems. First, they are easily understandable. Every voter and poll worker understands that ballots should be kept in a locked box to keep them from being stolen or modified. Far fewer people know how to prevent man-in-the-middle attacks on protocols that don't encrypt their passwords. Second, paper ballots are localized. If I want to rig a paper-ballot election I need to do a lot of leg-work to get all those physical artifacts changed. With some of these electronic security exploits I just need access to a single voting booth and I can silently change every vote it registers for the rest of the day.

   The solution lauded by both the Johns Hopkins team and Gillmor is to have a "voter-verifiable audit trail" as a backup for the electronic system. Whenever a vote is cast, a paper ballot is printed and checked by the voter for accuracy. If the print-out reads correctly, the ballot is stored as a record of the vote. If it is incorrect, the vote is invalidated and the paper ballot destroyed. Should the electronic record be questioned, the paper audit can be counted to confirm the results.

2. To be secure a system needs openness, not secrecy. Security is hard to get right, and there are a lot more bad guys in the world than there are good guys working on a single programming team. The best way to secure a system is to publish the full specs and source code and let the world's experts look for loopholes. Diebold claimed their system was secure, and even had outside certification by "independent testing authorities." But once the code was shown in the light of day the warts became visible. Vendors who sell "secure" software without publishing their source code are like used-car salesmen who won't let you do a test-drive first.

   Diebold is in an interesting situation now. The Johns Hopkins analysis found security holes big enough to fly a starship through, but they had to make a lot of assumptions due to not having the complete specification. Diebold is defending their software in part because of those holes in the team's knowledge, but unless the whole system is brought out into the light for a full and informed debate to occur there's no way it can be trusted.

3. The Johns Hopkins analysis notes that there was a large amount of additional source code they could have used to analyze the security of the system, but did not. The problem was that the extra source was protected by a PKZip password. PKZip doesn't provide any real security (there are plenty of password-breaker programs online) but the team didn't crack the minimal encryption because they were worried about being arrested under the Digital Millennium Copyright Act. Yet again, the DMCA has a chilling effect and leaves our Nation worse off and less secure in the process.

Luckilly, if you live in California there's something you can do. Secretary of State Kevin Shelley is soliciting public comments on a recent task force report on touch-screen voting machines, until Aug. 1. Comments can be sent to Secretary of State Kevin Shelley, attn: Touch Screen Report, 1500 11th St., Sacramento, CA 95814. E-mail comments to taskforcecomments@ss.ca.gov or fax at 916-653-9675.

References:

• Technical Response To The Johns Hopkins Study On Voting Systems (Diebold, 25 July 2003, updated 29 July 2003)
• Computer Voting Is Open to Easy Fraud, Experts Say (NYT, 24 July 2003)
• Security Experts: Electronic Voting Machines Threaten Democracy (Dan Gillmor's Blog, 23 July 2003)
• Analysis of an Electronic Voting System (Kohno, Stubblefield, Rubin and Wallach, 23 July 2003)
• Electronic Voting Machines Need More Safeguards (Dan Gillmor in Silicon Valley, 20 July 2003)
• Ad Hoc Touch Screen Task Force Report (California Secretary of State, 2 July 2003)

US to add RF-ID to passports by October 2004 

Frank Moss, US deputy assistant secretary for Passport Services, announced at the recent Smart Card Alliance meeting that production of new smart-card enabled passports will begin by October 26, 2004. Current plans call for a contactless smart chip based on the ISO 14443 standard, which was originally designed for the payments industry. The 14443 standard supports a data exchange rate of about 106 kilobytes per second, much higher than that of the widely-deployed Speedpass system.

The Enhanced Border Security Act and Visa Entry Reform Act of 2002 requires that countries in the US visa waiver program include machine-readable biometric data. This new passport project would update US passports to meet that standard as well. Moss says the goal is global interoperability, and plans to adopt standards from the International Civil Aviation Organization (ICAO). The EU recently made a similar announcement.

The new passports would encode cryptographically signed copies of the passport photo and passport information, making such information more difficult to forge. Moss did not mention the use of biometric data not currently included on US passports, such as fingerprints or retinal scans, though The Register suggests that the EU is looking into these systems as well. Given that the ICAO has not yet ruled out such additional biometrics it is probably too soon to tell, though the current ICAO blueprint does favor face images over fingerprint or retinal-scan technology.

References:

• Government Cites Policies and Standards Behind Broader Use of Smart Cards (Smart Card Alliance, 22 July 2003)
• Passports to get facial biometrics (Washington Technology, 22 July 2003)
• TI Embraces Prox Card Standard (Texas Instruments RFiD Journal, 6 March 2003)
• Enhanced Border Security and Visa Entry Reform Act of 2002 -- ALDAC No. 1 (US State Department, 15 May 2002)
• MRTD - Biometric Technology Identity Confirmation (ICAO Tech Report Exec Summary, March 2003)
• Other ICAO Technical Reports
• Biometric Identification To Provide Enhanced Security And Speedier Border Clearance For Traveling Public (ICAO Press Release, 28 May 2003)
• EU backs biometric passports (The Register, 23 June 2003)

IEEE deciding short-range wireless standard this week 

Nearly six years to the day after the process was started, it looks like the IEEE is honing in on a single standard for a fast (around 100 Mbit/s), short-range (< 10m), low-power, low-cost wireless communication. The standard, which will be IEEE 802.15.3a, comes out of the IEEE Wireless Personal Area Network (WPAN) working group. Unlike cellular or Wi-Fi networks, the point of a personal area network is to communicate with other devices that are there in the room with you. For example, a high-speed WPAN would allow your PDA to stream video directly to a large-screen TV. Alternatively, your core CPU could wirelessly communicate with medical sensors, control buttons, displays and ear-pieces, all distributed around the body. The standard fills much the same niche as Bluetooth (the first standard adopted by the working group, also known as 802.15.1), but the new technology is significantly faster than Bluetooth (up to 100 times faster, according to champions of the technology).

Trade news columnists who know more than I do about this are picking Texas Instruments' proposal for OFDM UWB (that's Orthogonal Frequency Division Multiplexing Ultra Wide Band, thank you for asking) as the likely technology to be picked. Assuming it does, TI's UWB business development manager says we can expect to see the first UWB products hitting the marketplace in 2005.

Update: The standard did not receive enough votes to pass, and will be voted on again in mid-September.

References:

• IEEE 802.15 Website
• Wireless raises the final standard (ZDNet, 22nd July 2003)
• Basis of UWB Spec Could Be Decided This Week (Electronic News, 21st July 2003)
• Military's 'sneaky wave' out of hiding (C|net, 11 March 2003)
• Texas Instruments' UWB page

We control the horizontal... 

I can hear it now:

Exec #1: "Members of the Word Media Cartel, we are against the ropes. We've tried imposing draconian penalties for even trivial piracy. We performed a perfect end-run around the fair use doctrine with the Digital Millennium Copyright Act. We've sued into bankruptcy anyone who might have a business model more survivable than our own. We've even sued down-and-out college students for $97.8 trillion dollars each, as an example to others who would stand in our way. And yet the peer-to-peer networks continue to thrive."

Exec #2: "If only our industry had a way to convince people that piracy was wrong. You know, change how people think about copying music and movies."

Exec #1: "Yes, yes, but there's no point in wishing for... hey wait, say that again!"

And so it came to pass: the Motion Picture Association of America launched an unprecedented media blitz to convince the American public that by using Gnutella you hurt not just Disney stock-holders, but also Jerry, the man who fetches coffee for George Lucas every morning at 5am.

The sheer power of this blitz is daunting. The kickoff this Thursday will have thirty-five network and cable outlets all showing the same 30-second spot in the first prime-time break (a "roadblock" in ad-biz terms). Then every major theater in the country will play daily trailers on all screens in more than 5,000 theaters. Whew. And all that time is donated, which would be incredibly impressive if the spots weren't essentially being donated to themselves.

And now the $97.8 trillion-dollar question: is the American people so pliable that their morality can be changed by a media blitz? (Could that be the manic laughter of of thousands of ad executives I hear in the distance?)

References:

• College students sued
• Background on the DMCA

DVR Manufacturers Capitulate 

The Associated Press has an overview piece on how the makers of digital video recorders are capitulating to (excuse me, "voluntarily cooperating with") Hollywood and other members of the Content Cartel. Not too surprising given the shots fired through the bow of Sonicblue (makers of ReplayTV), forcing them into bankruptcy after paying millions in legal fees. In line with the rest of the industry, ReplayTV's new owners say they will be good little boys and remove the ability to auto-skip commercials or send recorded programs over the Internet to other Replay users.

In the short run this means consumers get fewer features, but in the long run it's just more sand thrown against the tide. DVRs are just commodity hardware, some standard drivers and a little bit of interface software. If UltimateTV, TiVo or ReplayTV doesn't provide the features people want then a whole host of small manufacturers, kit makers and do-it-yourself modification kits are all more than willing to fill the gap.