Defeating Captchas

So if you’ve been on the web recently, you’ve probably seen something like this prompting you to enter the text before the site grants you access:

known as CAPTCHA’s, they are a form of reverse Turing tests meant to identify an entity as a person or a ‘dumb’ machine based on their capabilities. Seems to have originated with Max Levchin when he was trying to combat spammers on PayPal. But I was puzzled when I was prompted to solve one when I was uninstalling a piece of spyware on an old computer – why would they bother to ask me? It was later that Jacob elucidated the motivation for me: malicious bots on the internet can bypass the CAPTCHAs by simply defering it to a human on the other end of some screen wanting access to typically a porn/gambling site on the internet. The human solves it, it takes the input from them and sends it to the site the bot needs access to, and presto, problem solved.

It’s essentially what’s become known as a ‘man-in-the-middle’ attack to cryptography people… but it extends to other disciplines – such as the military. The example I remember is of the South African air force trying to detect whether a plane flying overhead is a friend or foe – because enemy planes can try to act as hidden communication intermediaries between the ground system and other South African aircraft, essentially passing the outgoing communications of one to the incoming channel of the other, and vice versa.

My Summer at the $100 Laptop Project

I was lucky enough to have the opportunity to spend 3 months @ MIT working on the One Laptop Per Child project (OLPC) this summer. I was joined by a cadre of four other students from all over, including:

* Jacob Russ, Harvard, obsessive digital photographer
* Samat Jain, Columbia, taught me half of everything I know about Python
* Aaron Burton, Carnegie-Mellon
* Manu Kumar, Delhi Institute of Technology

I was working under Jim Gettys (co-authored the X desktop system) and Ivan Krstic (a Harvard stop-out) for most of the time; started off doing stress-testing, moved on to develop an email client which was cut off in favor of a webmail client, and then spent the rest of the time learning Python in order to do some crypto library work.

Nicolas Negroponte was the head of the initiative but as his assistant liked to say, he “spends 95% of his time travelling” so I only got to see him twice. Mary-Lou Epstein was the LCD monitor wizard who was in Taiwan half the time, but brought back the first demos of the high-res, non-volatile (meaning you turn off the power, you can still read it) monochrome/color LCD. We spent most of our time working with the prototype motherboards shipped over from Taiwan…

Spent our retreat at Barry’s house kayaking on the Charles River. Played Aerobee with the guys in a government lawn which we were subsequently kicked off of by a European security guard who couldn’t pronounce his negatives (You can play here…-OK… no, no, you can take pictures here). Two-hour lunch breaks at Anna’s discussing everything from spam to Chiapas, Mexico to the Israeli/Palestinian conflict… free passes to the Siggraph conference. Free pizza on Fridays with the rest of the office Spent some time roaming around MIT’s campus, tunneling through their underground catacombs, and got to see the fire truck on top of the dome to commemorate September 11th (there is no way to gain access to the roof – I’ve checked – but my friend Zaina thinks they must have climbed up and used ropes/pulleys).

Lucky Lindy

I never realized how much an achievement Charles Lindbergh’s flight across the Atlantic actually was until I saw this picture (courtesy Corbis):

and the report of his arrival in Paris. As National Geographic put it, “he took off as an unknown boy from rural Minnesotta and landed 33 1/2 hours later as the most famous man on earth… and sent the world into an unprecedented frenzy.” A non-stop 33 hour flight, alone in the bitter cold, with primitive navigating instruments… most people don’t know that people had tried and failed to accomplish the feat before him and had died in the process. it’s a shame he didn’t write an autobiography about it.

Music Notation

I’m taking introductory piano class with Prof. Zerlang this year… he mentioned something rather interesting in class. Traditional music scores are written left to right, mirroring the English language, with notes ranging from top to bottom of the ledger lines and spaces, like so:

yet the piano itself has keys layed out horizontally, so trying to figure out which keys to play then, requires a transformation by ninety degrees clockwise. Why not just write music notation this way? This is a sketch of what I thought it would look like, read from top to bottom….
though notes that need to be held for longer would need to be repeated in lower staffs.
I suppose you could adapt some software to do it for you, perhaps LilyPond rather than Sibelius. But you have convention going against you…