Slashdot Mirror

Actually Anonymous Scientist is right: almost every single object recognition system these days uses learning, and several are inspired by biology. For example, look at http://www.cs.nyu.edu/~yann/research/norb/index.ht ml, or
http://www.cs.ubc.ca/~lowe/papers/06mutch.pdf, or http://web.mit.edu/serre/www/Research.htm.

Information, Technology and Information Worker Productivity: Task Level Evidence (DRAFT!)
http://www.stern.nyu.edu/ciio/WorkOnline/IS2005200 6/022806.pdf

An Empirical Analysis of Strategies and Efficiencies in Social Networks
http://128.135.211.53/research/workshops/orgs-mark ets/docs/bulkley-strategies.pdf

No need to buy anything, just build your own RFID zapper to get rid of those pesky RFIDs! Here is a recipe to build one yourself from the flash of a throw-away camera: http://itp.nyu.edu/everybit/blog/media/rfid-zapper .pdf Not compliant with any FCC rulings though...

During the talk, Jimmy acknowledge that the Beta of the engine is gonna suck and the media is gonna shit all over it. When the beta is released, they're gonna type in bold letters "We know this sucks" to curve some of that negative karma from the press. At least he's realistic about the project. Check it out the video of Jimmy's NYU talk here: http://video.google.com/videoplay?docid=-741696809 2951113589 or download the MP3 here: http://homepages.nyu.edu/~gd586/Jimmy%20Wales%20-% 20NYU%20-%201-31-07.mp3

Basically you store the copy on any number of servers but the key is split up across other nodes. Retrieval of document requires all the key accessible in a simple format. None of the servers can get to the contents of the doc without the full key and therefore there's plausible deniability built in.

"Jobs missed a golden opportunity at this keynote. Given the momentum and the increased buzz around Apple, their slowly increasing market share, more developers on board, Bootcamp etc. he could have finally presented Apple as a serious and viable alternative to Microsoft. For everyone. But instead he decided to go with a f**king phone"

I disagree. There's been so much buzz about the iPhone that only a few people have been asking about Macs and Leopard, and why Jobs didn't even so much as mention them. I must admit that I was pretty dazzled by the iPhone's interface, and it took me a couple days to start sorting out the implications.

I'm convinced that Leopard's new interface will support multi-touch technology (MTI). Am I the only person who believes that Apple has already thought of vastly more expansive uses for MTI than a mere smartphone display? Hello? Mac Tablet anyone? The iPhone interface is merely the tip of the iceberg of possibilities. Take a look at the video demo at the Multi-Touch Interaction Research group's site and imagine some or most of these capabilities, or even greater capabilities, in Leopard. Interestingly, there's a note on the site that says they saw the keynote, and that they have some more exciting stuff coming up soon.

Jobs said nothing about new Macs, new displays, or OS X 10.5 for one reason: he believes that what he has up his sleeve will make Vista look like ancient technology to Joe Consumer, and he's deliberately waiting for Microsoft to launch their expensive media blitz introducing Vista before dropping a Leopard-spotted nuke on them. His aim is to embarrass Microsoft. And I believe that Microsoft came to that conclusion while the keynote was going on, but they still have no choice but to kick Vista out the door.

Joe Consumer has already seen the iPhone's interface, courtesy the mainstream media. He'll be primed for multi-touch interface on a personal computer, and I foresee PC salespeople having an interesting time in the aftermath of Leopard's introduction: "Yeah, that's a pretty cheap machine, but how come I can't just drag things around with my finger like the guy at the Apple Store showed me?"

As many here have pointed out, Macs don't do anything that PC's can't do (much less if you count games and enterprise apps); iPods do less than many other available DAP's; the iPhone won't offer any capabilities unavailable on other, existing smartphones. The difference in all three cases is how Apple pulls the interface together in ways that appeal and make sense to average users i.e., non-Slashdot readers. I believe that Jobs has high hopes that Leopard will present an interface that will finally, clearly, distinguish Macs from PC's in the minds of the average consumer, in the same way that their respective interfaces distinguish the iPod and iPhone from competing devices. I believe that Jobs honestly feels that 2007 is the year of destiny for the Macintosh.

which I think Apple has the patent on.

Apple does not have any patents on the iphone. They have applied for about 300, but none have been granted yet. Regarding the multitouch interface, if you search the internet, you'll find that research has been going on in this area since the 1980's. At best, Apple might be granted a patent on the specific technology they've used to support multitouch in their touchscreen, but there are several other ways to accomplish the same thing, some of which are already available.

That Jefferson Han article points out that Apple "may" be coming out with a "touchscreen iPod" in the future, so I doubt it was written knowing that Han has already posted on his site that "Yes, we saw the keynote too" and that they "have some very, very exciting updates coming soon- stay tuned!" The site may say "February 2007," but it's straight from last year. Yay for magazine-caliber latency.

So, way to not point it out by using an outdated article, but I would be so bold as to venture that Han and Apple are working together.

I think if Archimedes were alive today, he could easily do it - he'd just need a very large sheet of paper.

As a web-page of NYUMC suggests, Tai Chi may be of help.

Quote: Enhanced Endocrine and Immune System
One study found that tai chi was associated with increased blood levels of a number of hormones, including thyroid hormone, testosterone, and estradiol. Another study found that tai chi was associated with an increased number of T cells (cells involved in the immune response).

CC.

http://www.physics.nyu.edu/faculty/sokal/dawkins.h tml

Suppose you are an intellectual impostor with nothing to say, but with strong ambitions to succeed in academic life, collect a coterie of reverent disciples and have students around the world anoint your pages with respectful yellow highlighter. What kind of literary style would you cultivate? Not a lucid one, surely, for clarity would expose your lack of content.

Of course, musical accompaniment systems have been around since the 80's and earlier. It's been a research topic in artificial intelligence nearly since its inception!

Robert Rowe published a book on the topic in 1992, on his Cypher system. Here's another good article on the topic by Chris Dobrian. For an open-source system, check out Bob van der Poel's MMA.

Additionally, there's been plenty of work done on robotics for playing instruments, particularly for percussion.

So, admittedly, this is the first time I've personally heard of a project combining the two, so I'll give it that credit for innovation. But I'd be sorta surprised if it hasn't been done previously. When you think about it... all these musical accompaniment systems react in real-time to MIDI input. Simply make a couple of motors respond to MMA's ouptut, for example, adjust timing according to latency and inertia, and you could probably have this project done in a few days.

Not to play it down, I always love to see fun projects like this.. :) And the physical design is quite beautiful for a drum-playing robot.
But "first".. well, give credit where it's due. I think the summary is over-reacting. (I scanned the article.. don't believe it makes any such claims.)

The book, Structure and Interpretation of Computer Programs has the introductory student write a full Scheme language interpreter in the Scheme language, then an assembly language register machine in the scheme language, and finally a scheme Interpreter in assembly language ....

It's an interesting excercise that I went through as an undergrad (though I switched majors and became a mechanical guy instead of a software guy). I remember Scheme as a pretty fun language to program in, once I sort of got the hang of it, but this is an aside.

any universal computer can emulate any other universal computer.

As this applies to the AI argument, the problem is that you have to assume that the human mind is a type of universal computer, which is a statement disputed by detractors of strong AI. At any rate, it is an assumption that the human mind is software.

The Chinese room thought experiment only addresses weak AI, hence the distinction. First of all, it assumes that symbolic manipulation without understanding of context is even possible.

The Chinese room, actually addressed strong AI -- at least that's what it's intended to disprove, I'm not sure that the experiment says anything about weak AI, other than to say that *all* we can hope for is weak AI.

As to symbolic manipulation without understanding of context, that describes exactly what my computer right now is doing (unless you really want to argue that my laptop *is* a working mind, which I think would be a hard argument to make). That's the crux of the argument; computers don't have context and will never have context. As to the fact that they do (mindlessly and succesfully) manipulate symbols I don't think is arguable.

Turing formulated his famous test and no one has come anywhere close to passing it through context free symbolic manipulation.

A BASIC program running modified and extended versions of the Eliza program *have* passed actual Turing tests. The Turing test I don't think is taken seriously by anyone as a test for strong AI since it's hard to argue that a thousand lines of BASIC is sentient. See:

http://www.nyu.edu/gsas/dept/philo/faculty/block/p apers/msb.html

(you have to scroll down a bit where he describes the experiment).

after replying with questions to clarify finer points about the process, responded with it's ideas on new flourishes to the steps, then THAT would be indicative of intelligence in the system.

1st) The question that the Chinese room asks is this: does the person following the rules to shuffle and spit out Chinese characters speak Chinese? The answer is obviously no, as they have no idea what the symbols they're spitting out actually mean. In your example, the person, the person-room system, etc. has no idea whether they're telling a dirty joke, discussing the weather or asking questions about dance moves. That's the actual point -- appearing to understand a languge is not the same as understanding a language (think of it as additionally disproving the Turing test as a valid test for intelligence).

2nd) In the thought experiment, all possible answers to all possible conversations have been pre-programmed by the English rule-book the non-Chinese speaker inside is consulting and following. Would you argue that this is intelligence? Or does it just give the impression of intelligence (and the impression of being able to speak Chinese).

Without having thought much about your "universe inside of a computer" idea, I have to say that at the moment I can think of one thing that does make this impossible: chaos. A computer fed with all the laws of physics would not predict you, your preference in ice cream, McDonald's, human biology, etc. much less contain the human mind.

In fact, maybe chaos is just one example of complexity that can't be modeled with a "simple" universal computer: after all, it's a big assumption to say that the cosmos is just one big universal computer (which I think is exactly what your argument is based on).

My point is that the problem that remains to be solved isn't a problem of technology but a problem of understanding how our human mind works. This will come from someone studying the human mind, not someone studying technology (as in the universal human language analogy -- the hard part is figuring out human linguistics and how we commnicate, not being clever enough as a computer programmer).

On the other hand, the theorem of universal computing means that once that peice is understood, it will be capable of being modeled in any logical environment, meaning that indeed, all you need is the right program and "poof", as there is a proof to this effect.

There is a HUUUUGE assumption in this statement: basically (as I wrote) that the brain is hardware and the mind is software, so if we understood the human mind already (which we don't -- and again there are arguments that say that we never will; we can't. The argument basically says that it always takes a more advanced mind to understand a lesser one. A single intelligence is by definition not smart enough to understand itself.) it'd just be a matter of writing the correct software for your TSR-80 (or whatever computer).

However, once you start thinking about it, this model of the human mind as "software" starts breaking down quickly. I have come to believe that (forgetting the difficulties of understanding the workings of the human mind in detail) a software program will never actually achieve human intelligence in any form. It might be smart by some definition of intelligence, but it'd never be human intelligence as human intelligence is innately biological.

One cool thought experiment is "the chinese room argument":

http://www.iep.utm.edu/c/chineser.htm#H1

Check it out: I'm sure that if you're willing to think about it it'll start challenging the assumptions that we *all*, as technology minded people seem to make about how things (our human mind in this case) work.

Another paper, by the prof I studied under, is unfortunately actually a bit dense reading IMHO (even though I'm familiar with what he's saying), but he makes a lot really cool arguments attacking the idea that all we need for strong AI is the "right" computer program:

http://www.nyu.edu/gsas/dept/philo/faculty/block/p apers/msb.html

Basically, I'd have more respect for the futurist if he'd actually acknowledge some of these questions regarding AI -- a lot of which are actually quite old (as old as Turing, who, as a techonologist, also made some proposals for strong AI that have turned out to be quite naive, as in the Basic program Eliza passing the Turing Test for intelligence.)

Oh, here's a better-explained (IMHO) link to the chinese room experiment:

http://en.wikipedia.org/wiki/Chinese_Room

Maybe this have some similarities with Perlins' autoStereo Display. (Warning: java applet). Also check out the other great applets from Mr.Perlin: http://mrl.nyu.edu/~perlin/

Maybe this have some similarities with Perlins' autoStereo Display. (Warning: java applet). Also check out the other great applets from Mr.Perlin: http://mrl.nyu.edu/~perlin/

http://mrl.nyu.edu/~jhan/

I saw the multi-touch display wall at this year's SIGGRAPH. Playing with it is, obviously, worth more than looking at pictures, but you really have to watch the multi-touch interaction demo real.

Track Me Not - http://mrl.nyu.edu/~dhowe/TrackMeNot/

TrackMeNot runs in Firefox as a low-priority background process that periodically issues randomized search-queries to popular search engines, e.g., AOL, Yahoo!, Google, and MSN. It hides users' actual search trails in a cloud of indistinguishable 'ghost' queries, making it difficult, if not impossible, to aggregate such data into accurate or identifying user profiles. TrackMeNot integrates into the Firefox 'Tools' menu and includes a variety of user-configurable options.

Neat idea. Sadly, the implementation as well as the idea is spectacularly flawed. If you want to be really safe (like the people actually engaging in child pornography) block all cookies, set the browser cache to zero, the browser history to zero, tell your browser not to save passwords/data forms/anything of any kind, run noscript, and run TOR. If this isn't enough, you might want to do all that in Knoppix at a WAP somewhere far from your home, and blow up the computer after you use it.

Just like security, there is no such thing as perfect anonymity, given the motivation law enforcement will track anyone down. Since the afore mentioned ideas will seriously reduce the usefulness of the Web, clearing private data when closing the webbrowser and TOR are probably enough for you.

Back to the reason this "flooding the data miners" idea is flawed, Bruce Schneier wrote:

One, it doesn't hide your searches. If the government wants to know who's been searching on "al Qaeda recruitment centers," it won't matter that you've made ten thousand other searches as well -- you'll be targeted.
Two, it's too easy to spot. There are only 1,673 search terms in the program's dictionary. Here, as a random example, are the program's "G" words...
The program's authors claim that this list is temporary, and that there will eventually be a TrackMeNot server with an ever-changing word list. Of course, that list can be monitored by any analysis program -- as could any queries to that server.
In any case, every twelve seconds -- exactly -- the program picks a random pair of words and sends it to either AOL, Yahoo, MSN, or Google. My guess is that your searches contain more than two words, you don't send them out in precise twelve-second intervals, and you favor one search engine over the others.
Three, some of the program's searches are worse than yours. The dictionary includes...
Does anyone reall think that searches on "erotic rape," "mailbombing bibles," and "choking virgins" will make their legitimate searches less noteworthy?
And four, it wastes a whole lot of bandwidth. A query every twelve seconds translates into 2,400 queries a day, assuming an eight-hour workday. A typical Google response is about 25K, so we're talking 60 megabytes of additional traffic daily. Imagine if everyone in the company used it.
I suppose this kind of thing would stop someone who has a paper printout of your searches and is looking through them manually, but it's not going to hamper computer analysis very much. Or anyone who isn't lazy. But it wouldn't be hard for a computer profiling program to ignore these searches.
Yes, data mining is a signal-to-noise problem. But artificial noise like this isn't going to help much. If I were going to improve on this idea, I would make the plugin watch the user's search patterns. I would make it send queries only to the search engines the user does, only when he is actually online doing things. I would randomize the timing. (There's a comment to that effect in the code, so presumably this will be fixed in a later version of the program.) And I would make it monitor the web pages the user looks at, and send queries based on keywords it finds on those pages. And I would make it send queries in the form the user tends to use, whether it be single words, pair

So say the neurobiologists:

"Scientific Background
We draw on the latest discoveries in the neural sciences, linguistics, psychology and anthropology and apply them in the world of business.
Recent discoveries in the field of the neural sciences teach us that:

  Emotion is the trigger to action
  The rational system follows the emotional system
  Present actions are driven by past experiences
  Present experiences dictate our emerging needs

The Imprint analysis incorporates this knowledge of the human mind to determine people's emerging needs and 'entry-points' for effective communication."

From:

http://www.culturalimprint.com/about.html

Who have a fascinating (and scary, since the marketin world is actively using this theory) article:

http://www.culturalimprint.com/emerging%20needs.pd f

The question is, what kind of cultural imprinting are kids getting these days?

Further references are Rapaille's book "The Culture Code"

http://www.randomhouse.com/broadway/culturecode/

And LeDoux's book "The Emotional Brain":

http://www.cns.nyu.edu/home/ledoux/the_emotional_b rain/book_newsci.htm

I'm certain it was posted here before for some other article, but I fail to see posts for this article that link to this?

Multi-touch interface is where it's at - personally, I wouldn't want to talk to my computer to get it to do something. The processing power can be put to much better use calculating a cure to cancer. Plus, it's an annoyance to everyone around, sure, it's cool in that Star Trek kind of way, but for constant use? No way man. It would suck having to tell your computer what to do - really.
Think about it, late at night, you want the computer to do something: "Find me porn" "Honey? What are you doing?", uh, yeah, that's not gonna fly.
Playing games: "shoot that creature there! Damnit! This is much too slow!".
Or programming: "int main bracket argh.. no. Delete. No. Delete del- ah screw it."
Or word processing: "Dear mom, fix aunt, delete that, delete that, delete that, select all, I think it's picking, I think it's picking up a bit of echo here, delete- select all" yeah, no. I think I'll stick with my multi-touch interface.

Multi-touch is completely natural and virtually no learning curve. We all have fingers or limbs, or feet, or noses or whatever else with which you use to interact with things - multi-touch interface takes that into account. Plus, information has traditionally been shown on a screen. How often do you hear the story of the newb trying to use the screen as a touch interface?

Excel is only nice if you've never used anything else to do number crunching. For a start Excel is inaccurate and sometimes downright wrong. Microsoft has known this for a long time and not done much to correct the situation.

The thought that Excel is used to do anything serious involving decision making beyond simple spreadsheet calculations in the real world gives me the creeps.

Seems like it wouldn't be that hard. Image Analogies would be perfect methinks.

John Bruce Thompson is just Fredric Wertham all over again.

Swap out "comic books" for video games and you have this, which is a repeat of history.

Then there is Rosie http://mrl.nyu.edu/~perlin/experiments/rosie/ . Does this mean he can't patent the idea due to prior art?

As long as they don't look like this

We're doing something like this at NYU:
NYU Query by Humming

There were talks last summer with AOL about expanding our database massively (millions of songs) for some sort of music downloading service, and a mobile phone application was discussed. Essentially the phone would do the signal processing on the recording of the voice, send a small string to the server to query, and the server would return a list of matches. Phones can do this now, but it's often horribly slow as FFT libraries are in their infancy on phones and differ greatly across models (at least that was the state of affairs a year ago... surely things have changed since).

This isn't the same as an iPod query-by-humming application, though, where the searching would have to be performed on the device itself. I've never developed for the iPod so I can't speak to its speed compared to phones, but the searching part probably wouldn't be too bad. Our database is kdb running on a wimpy P4 2.2 with 512mb ram (!) and though the data set is quite small, searches are very fast and it should scale well. (Heck, k and kdb binaries are so tiny (~200kb) they could just run on the device!)

If you think statistics are easily rigged then you should check on a guy called Alan Sokal to verify exactly how easy it is to rig qualitative knowledge. Qualitative knowledge is nothing but the lowest form of quantitative knowledge. When you evaluate a teaching method and conclude in a qualitative way that this method is either "good" or "bad" for some reason, you are doing a quantitative analysis with a precision of one bit. If you do a somewhat better evaluation and you classify it as "excellent", "good", "bad", or "very bad", then you improved your resolution to two bits, and so on.

Of course, every field of knowledge is multi-dimensional. When you try to improve a rocket engine, should you go for higher temperatures or higher pressures, or what about less weight? By doing small but measurable improvements in each of these and many other parameters they evolved from Goddard's crude experiments to moon exploration in fifty years. Many times you'll find you need to compromise because improving one parameter means worsening another and you often have no clear way to decide which is better. But you always need to be able to measure your results.

You say that "multi-cultural and linguistically sophisticated education is far superior to insular mono-cultures such as those found in many parts of the US", based on your personal opinions alone, but present no argument why. One may very easily argue the opposite point, let's see cultural manifestations. Traveling in Europe you find many movie theaters showing American films, you hear many radio stations playing American music, you find many books written by American authors in bookstores. Why would you say that European culture is superior? (Disclaimer: I'm not an American, at least not a Gringo American). You could just as easily argue that Europeans waste too much of their intellectual capacity learning languages to produce significant creations in other fields.

I'm still wiping tears from my eyes. This is right up there with one of the repurposing of the software from Randomly Generated Paper Accepted to Conference to generate one of the best typeset listserv removal requests ever.

Of course, now that I've reseen Figure 2 and reread the summary in section 5, I'm going to be laughing for a bit longer. Dammit.

This one is from NYU: http://www.cs.nyu.edu/~yann/research/lagr/index.ht ml

Actually, Coral CDN is now advocating the use of port 8080 instead of 8090. Both will still work for the time being.

There is more information in this posting to the [coral-announce] list.

A number of academics in domains like ethics, speech act theory, and philosophy of mind (among others) have been contributing to journals and having conferences related to computing and philosophy for a good while. I imagine that the interesting discussions on issues like free-will as well as models like functionalism will probably gradually enter the wider computer science curriculum.

http://mrl.nyu.edu/~jhan/

"Multi-Touch Interaction Research
"Bi-manual, multi-point, and multi-user input on graphical interaction surfaces"

Here is a page that has a much more impressive multi-touch sensing demonstration than just some doddles and window re-sizings: http://mrl.nyu.edu/~jhan/ftirtouch/

What genius put "bigbrother" as a tag for this article? If you read anything in the submission beyond the horribly-written title, you find out that what the display actually does is differentiate between the people using it at a particular time, thus allowing several people to simultaneously use the same interface device.

This has nothing to do with ascertaining your identity, and this time you didn't even have to RTFA to figure that out.

BTW, the Jeff Han mentioned in TFA has a site where you can see a much more impressive demo video.

i'm surprised no-one mentions http://mrl.nyu.edu/~jhan/ftirtouch/ (Bi-manual, multi-point, and multi-user interactions on a graphical interaction surface) - the demo video shows an on-screen keyboard that you type directly on with your fingers. and as a replacement for the mouse, i find it far more practical. with such an input method, you could expand the pepperpad#s screen real-estate a bit more. slap USB 2.0 on it, improve the battery life, and i think they might have a winner.

http://www.nyu.edu/classes/copyXediting/Punctuatio n.html
Use a period to end a rhetorical question.

Incorporating sensing elements within the display will permit sensing multiple simultaneous points of contact of arbitrary size/shape in a tablet form-factor. Neat!

Apple's been patenting lots of touch-interface concepts recently, too. Vide.

This patent is probably more about touch-screens than screen as scanner (that'd be a neat trick too, but probably would require too much resolution) or camera (would require a different but perfectly calibrated refractive element at each sensor - probably impractical).

-Isaac

Correction: The humanities journal in question was not peer reviewed; see the editors' account of why they published the hoax paper.

Interesting... Just for the heck of it, I ran Alan Sokal's paper Transgressing the Boundaries through the detector. It came back with a 93.8% chance of being authentic.

For those of you who don't remember the story, Sokal, a physicist, wrote a paper full of postmodern-sounding gobbledygook, asserting among other things that gravity is a social construction (the paper was subtitled, "Towards a Transformative Hermeneutics of Quantum Gravity"). The paper was accepted at a peer-reviewed humanities journal. Sokal later revealed it to be a hoax.

Of course, by the detector's standards the article was indeed "authentic" in the sense that a human being did write it.

Interesting... Just for the heck of it, I ran Alan Sokal's paper Transgressing the Boundaries through the detector. It came back with a 93.8% chance of being authentic.

For those of you who don't remember the story, Sokal, a physicist, wrote a paper full of postmodern-sounding gobbledygook, asserting among other things that gravity is a social construction (the paper was subtitled, "Towards a Transformative Hermeneutics of Quantum Gravity"). The paper was accepted at a peer-reviewed humanities journal. Sokal later revealed it to be a hoax.

Of course, by the detector's standards the article was indeed "authentic" in the sense that a human being did write it.

Have any of you heard of The Sokal Hoax? In 1996, a daring and dissatisfied physics professor named Alan Sokal wrote a bullshit paper called "Transgressing the Boundaries: Toward a Transformative Hermeneutics of Quantum Gravity", which Sokal called "a pastiche of left-wing cant, fawning references, grandiose quotations, and outright nonsense", which was "structured around the silliest quotations I could find about mathematics and physics" made by humanities academics. In short, it caused a big scandal because the paper was readily accepted without review by Duke University's postmodern cultural studies journal Social Text. It's probably one of the best and most controversial examples of a hoax on the "academic community," and it is excellent proof of just how much bullshit flies for "cultural studies." Run THAT through your paper detector! Read more about it here: Skeptic's Dictionary and Museum of Hoaxes

All this talk without a single mention of the Sokal Affair? It's pretty relevant. Also be sure to check out Paul Boghossian's article, "What the Sokal Hoax Ought to Teach Us." Great reading.

Section 6. Conclusion: Cooperating to end the balance of financial terror

The sharp rise in U.S. net external debt since 2001 has financed a fund a boom in government borrowing, a boom in consumption and a boom in residential construction -- not a boom in investment, let alone investment in the export sector. The U.S. has become increasingly dependent on foreign purchases of fixed income debt securities - and in particular purchases of U.S. treasuries by Asian central banks - to finance huge U.S. current account deficits, deficits that are absorbing an enormous fraction of all cross-border capital flows.80

Our analysis suggests that without any policy changes, the US current account deficit will rise above 7% of GDP in 2006, and above 8% of GDP in 2008, in part because of rising interest payments to non-residents. If most of the financing for the deficit continues to come from Asia, Asian central bank reserves would need to double between the end of 2004 and the end of 2008, rising from $2.4-2.5 trillion to $5.2 trillion. Foreign holdings of U.S. Treasuries would rise in parallel, going from $2 trillion (end 2004 estimate) to $4.2 trillion. We doubt that Asian investors, even Asian central banks, will be willing to take on the financial risk implied by holding such a large stock of dollar claims on a country whose external credit fundamentals are deteriorating at anything like the U.S. current low nominal (let alone real) interest rates.

It is true that East Asia cannot dump its existing holdings of U.S. treasury bills without paying a financial price. If East Asia sought to diversify its reserve - holding more euros and fewer dollars as a hedge against dollar depreciation - it would trigger a downward adjustment in the dollar's value. Indeed, East Asian central banks have to continue to buy additional U.S. treasuries to provide the ongoing new financing the U.S. needs to keep the dollar from falling.

But the U.S. should not take comfort in the fact that East Asian economies cannot painlessly extricate themselves from their enormous - and growing -- financial bet on the U.S. dollar. The U.S. cannot extricate itself its dependence on the cheap financing provided by Asian reserve accumulation any more easily. The U.S. economy can only expand at its current pace on the back of the implicit subsidy provided by Asian central banks. The boom in housing created by low interest rates and, for that matter, the surge in value of all financial assets linked to low interest rates - would come to an abrupt end without access to Asian financing.

But make no mistake, this cheap financing is coming directly at the expense of the U.S. manufacturing sector. The continued transfer of resources out of tradables production bodes ills for the long-run health of the U.S. economy. It is not in the long-run interest of the U.S. economy to try to support an ever-increasing external debt load on the back of a shrinking tradables sector. At some point, the external side of U.S. economy has to expand to pay for the United States' imports, or the amount that the U.S. can import will have to fall.

The Grand Challenge, I thought, was designed with a detrimentally macho mindset, with a needlessly high ratio of financial risk to scientific output. It sent many expensive cars through long stretches of very uniform-looking but occasionally high-risk (cliffs!) terrain. The cars had a road to follow. It was ok to hire an army of undergraduates to hand-design a path through the terrain (see the funny scenes involving CMU from chapter 6 of this online NOVA episode about the Grand Challenge).

Maybe they've learned a few lessons, because the DARPA LAGR competition does everything in almost the opposite manner:

• Focus on software: The teams all use the exact same hardware. We each get a robot for testing, but for the competition we just hand off our software to the government organizers, who load it into their robot and let the robot run the course.
• Focus on the difficult problems: Current robotic localization and mapping is heavily reliant on laser range scanners, which have a limited range (~30 feet) compared to vision, and therefore is unsuitable for long-range path planning. Instead of letting scientists wait for laser scanners to get incrementally better, this competition forces its entrants to adopt a vision-based solution, like people. As one organizer put it, we didn't get to the moon by incrementally improving the airplane.
• A sane funding structure: A pool of 10 or so initial teams start out with funding, with each successive competition (with increasing demands) weeding out more and more teams. I've heard that other DARPA robot competitions happening in parallel have adopted similar funding systems. Contrast this with the grand challenge teams, who didn't get any money until the end (CMU notably sank $3 million into their efforts to win the $2 million prize).

LEDs are photodiodes too! http://mrl.nyu.edu/~jhan/ledtouch/index.html

I found this neato LED interface the other day from hackaday
Some more links to projects like this can be found on the story on hackaday.

(1) capacitive switches -- inexpensive, digital (on/off)
This reverses the idea of protecting a high-impedance circuit from stray capacitance introduced by a limb or finger -- turn your accidental 'people detector' into just that:

http://www.discovercircuits.com/C/capacitance-sw.h tm

(2) frustrated internal reflection -- not necessarily cheap (needs a camera), or easy (needs video analysis) but can handle multi-touch and large screens
The idea is to shine light in from the side of a class or plastic screen, and have a camera look at the backscatter introduced by finger contact, which scatters the sidways light rather than allow it to reflect at the bounary. The camera turns the touch events into a video stream which is then analyzed to compute touch events.

http://mrl.nyu.edu/~jhan/ftirsense/index.html

(3) strain gages -- not necessarily cheap ($5 - $8 for each strain gage) but can provide a very sensitive analog signal with wide dynamic range.
Put some strain gages on several centimeters of half-inch square steel tube and you can easily measure touch events, as well as strong pushes.

http://www.omega.com/literature/transactions/volum e3/strain.html

Here is a simple, inexpensive (if you make it yourself) amplifier for strain gages that I've tried, and can vouch that it works well:

http://www.staramp.com/

MERL's DiamondTouch is a multi-user system and can tell distinguish between the touches of several users. If you're touching in one place and I'm touching in another, not only does the table detect both touches, but it can tell which of us touched which spot. Current DiamondTouch prototypes use an X-Y grid and so do not detect touches and gestures in a fully pixelized manner. This is a limitation of the current hardware, not the fundamental technology. You can read the original DiamondTouch paper here.

Also note that MERL DiamondTouch predates Jeff Han's work by about four years.