Jeff’s Live Nudes

While I’m setting up the foundation for this blog, I might as well mention Kakeya, another project I’ve been engrossed in.

The idea was suggested by Phil a couple months ago in what would become his trademark, an obscenely long email, and it took off from there.  He suggested “a website dealing with transportation costs”, lining out the goals, obstacles and motivation.  We worked through kinks and plausibilities, and surprisingly, came up with no reason not to tackle the project.  The only thing left was to get our hands dirty.

One thing you realize as a college student in an urban environment is that public transportation is one of the toughest problems to tackle in the United States.  Our country’s density is wildly inconsistent — in the cities like Boston and New York, public transportation is a necessity, in the Virginian suburbs where I lived, an afterthought.  Yet it’s often a necessity to travel to and from areas with wildly varying population densities.  Typically speaking, transportation competition is fierce and numerous within populated areas, dropping prices extensively, but not so for long range transportation services (think greyhound, amtrak).

One consequence of this inconsistent price distribution is that there are often multiple routes from point A to point B, each with a cost often disproportionate to the “goodness” of the route (distance, comfort, # of transfers, etc).  This is something we can take advantage of.  Theoretically, transportation route data can be scraped from all the bus, train, subway and plane lines in the country, and one can query the cheapest or shortest route from any location to any other location in a fraction of the time it takes to discover by hand.  This has never been done on such a large scale in our country before (though London has a similar system), and this is the purpose of Kakeya, named after the Kakeya Conjecture, a problem Phil tackled at RSI that nearly destroyed his soul.

Here’s some food for thought: Ashu and I booked tickets a few months ago for a Killswitch/Disturbed concert, one we were both ecstatic about.  That is, until he found out there was a 006 test on the day of the concert.  Essentially our only hope was to find a quick route to the concert so we could make it after the openers.  It took us 2 hours of fishing through piece of shit websites (I’m looking at you, lrta.com) to find a single decent route, and it would have required us to take a 6am bus home.

Here’s another real life example.  Ashu and I (I think he has an aura that causes everyone around him to get owned) went down to Yale to present our Kakeya business plan to potential investors.  About an hour into looking for a cheap route, I turned to Ashu and said, “you know what would help?”  We had the same thought.  KAKEYA.

We have a site up and running in a sort of pre-alpha stage, but I’m not ready to present it yet.  There’s a lot of work to do over the summer, and I’m stoked about our progress.  We have a lot of obstacles in the way, mostly in the form of Google.  But I have reason to believe that we can beat them, or at least offer our own brand of innovation and perspective, and create a free product that can save millions of people heaps of time, money and productivity.  Here’s to making a difference.

As per my OCD, I take great pleasure in categorizing and rating things.  Of all the things I like to compare, I feel strongest about music, which is why I’m introducing Jeff’s Live Music, a chance for me to share songs I cherish, whether from immediate devotion or a rediscovery of a long-forgotten melody.

Our inaugural piece can be said to be a little bit of both.  I downloaded Castles Made of Sand after reading a wikipedia article on Dispatch, in which a reviewer praised them for a sound similar to that of our featured song.  Like most Hendrix songs, I couldn’t fully appreciate the soul of the tune on the first playback.  It was tossed aside for several weeks, incubating in the library until I had the leisure time to revisit it, and I became entranced.

Good music can bring about catharsis upon listening once, great music can upon listening infinitely.  Castles features a creative, challenging guitar part and an uneven tempo that keeps you on your feet.  It’s difficult to pinpoint a single mood Hendrix wants to convey — he dances between a bitter sadness and a wistful, almost hopeful pining in both his guitar and his voice.  The song is not sappy, not sad, not upbeat and not cheerful.

It tells of a story of a deadbeat husband kicked out by his embittered wife, a husband who, in his younger days of lust and love, may have promised the world, but has become disillusioned with it.  It tells a simple story of an indian boy who grows to be chief and dreams of leading his people to war, but is killed in his sleep on the eve of battle.  And it tells of a girl, cripped and mute, who prepares to end her life in the sea when she spots a golden ship over the horizon.  Each story is brief and cut short, like the characters they depict.  It’s a song about the best of plans thwarted, about the ephemerality of all good and bad, about Jimi’s childhood and family.

And as Jimi croons, so castles made of sand melt in the sea… eventually.

So for those who don’t know, this summer, aside from working my usual IS&T job, I’m working with a couple eggheads over at qwobl.com.  It’s a brilliant new paradigm for information classification and search, and, like Wolfram|Alpha, delivers content, not just references.   Want to know the amount of calcium in a gallon of 2% milk?  How about the records of every AFC conference team from the 2008 season?  What about the name of that sexy guest actress playing the paraplegic on House last night?  Google will give you these answers, with a little bit of elbow grease.  But what if you could bypass the step of sifting through webpage content to find these answers?  In principle, shouldn’t data mining be able to solve the problem for you?

Welcome to the Semantic Web, Tim Berners-Lee’s personal vision of the future of information retrieval and compatibility.  The idea: make information online quantizable and classifiable, and unleash the full potential of information automation.  By classifying information based on object-verb-subject relationships, we can construct an intuitive structure on which massive computation is viable.  I realize I’m being incredibly vague, as I’ve just begun to delve into the subject myself.

The power of the Qwobl search engine lies in the strength of the semantic web.  Data is scraped from sources on the web.  Data is converted into RDF form, which is then processed by NER algorithms.  Using this basis, one could retrieve any sort of information desired, without an inkling of human information processing.  A profound concept indeed.

Aside from Qwobl, I’ll be working on Kakeya, which you can read about in the subsequent article, and moonlighting at the help desk for a bit of cash, since I get next to nothing at the former jobs.  I’m both terribly excited and terribly nervous about my summer’s work, and I hope my gamble for experience before income will prove to be less of a blunder than I sometimes imagine it as.