Posted on June 29, 2010


I’ve recently heard a prominent venture capitalist say that what’s missing in NY’s start-up scene are “grand approaches”.  Something revolutionary. I think Hunch is an example of people trying something ambitious. 

@cdixon, one of its founders, explains that Hunch is trying to tap connections among people based on their personalities and personal preferences, instead of based on their existing networks.   This is more or less interesting, but I don’t think it is what Hunch is ultimately about.  Instead, I believe that Hunch’s true value lies in its ability to collect some pretty nice data.  And data is increasing dramatically in value, while computation power is becoming less important.  Data is the fuel for artificial intelligence.

If we look at Moore’s law, we know that there is a good chance that sooner or later computers will have the computation power of human brains (I think the most recent estimate expects this to happen in the next couple of years.)  Now, think what else is necessary to form something akin to an intelligent being:  Vast amounts of information.  In the case of humans, this data is usually being fed through various means: Parents, schools, experience. Imagine the amounts of information a 25 year old has been exposed to and processes every day.

Humans collect data through all five senses.  Smells are great for recalling specific memories.  Music can bring us to tears.  But in order to become conscious, evolution had to come up with a system of reflecting about the data collected through these senses and, of equal importance, to communicate it to others.  Language.  Language is a means of bringing structure to our thinking.  Some argue that depending on the language you think in, the outcome of your thinking will actually differ significantly.

Language is a great means for feeding data to artificial intelligence.  Language, that relatively simple, one dimensional system of various signs (if you want to reflect on how simple it is, just get aware of all the other stuff that is going on in your brain besides the more or less coherent sentences you  constantly form), can be used as a means to feed our thoughts into computers.

Many of the recent successes among start-ups do nothing else but get people to translate their thoughts into coherent language and then input it into computers.  Think of Twitter, for example.  Twitter gives us (and computers) direct access to people’s stream of consciousness.  Finding out what the mass of people is thinking at any moment is of incredible value.  Remember the scene in Waking Life where Ethan Hawke suggests that ideas often come into existence almost simultaneously in the brains of many people who are far apart from each other? (Or think Leibniz and Newton.)  Twitter makes prevalent thoughts accessible to artificial intelligence in real time.

Think of the preponderance of blogging.  Compared to published work, blogs are much nearer to the stream of consciousness of their authors than books.  The universe of blogs out there can be immediately be accessed by computers.

Think of Google’s book-scanning project.  People already have access to all books worldwide (for books published in the US, all it takes is a trip to the Library of Congress in DC).  Hence artificial intelligence is the main beneficiary of Google’s project.

And even the iphone and its friends.  Think about them as a devices to digitize human thought.  iphones and other “smartphones” are phones only secondary.  More important, they are digitalization devices.  As more and more people carry their smartphones with them, they start tweeting, emailing and taking electronic notes.  As you know, all these communications are already heavily analyzed by computers, if only to target you with personalized advertisements.  In other words, smartphones constantly collect human thought, digitize it and make it available to whoever wants to use it.

Enter Hunch.

Hunch could be the next step of data collection, because it goes beyond listening to people’s communication with each other.  Instead it tries to look directly into its users’ brains.  Hunch users tell its website about their most inner feelings: How they prefer to break up with significant others.  What philosophies they like.  Whether they think they are smarter than most other Hunch users (60% think they are.).  I.e., thoughts that they wouldn’t tell other people, that they wouldn’t communicate.  Thoughts that therefore cannot be collected by existing services like Twitter and such.  Hunch’s servers then try to find patterns, i.e. they analyze what attitudes usually go together.  Are republicans breaking up with their girlfriends differently than libertarians?  In return Hunch users get suggestions of where to buy there shoes  and where to eat – a service that is necessary to induce users to talk about themselves, but one that is truly only a by-product.

If Hunch kicks off, then it is a magnificent data collection device.  The data Hunch collects will therefore, without doubt, be of great value once computers are powerful enough to deal with it intelligently.

Addendum: Let me add Quora to the list of relevant businesses.  All they are doing is collect data that can be fed to AI. Would love to see how much of the valuation of the reported $86MM in VC money that flowed into this company was based on the usefulness of their data for AI at some later point.