Wednesday, May 22, 2013

Serendipity, Stupidity and Start-up

 It's been almost six months, since I have resolved my life and given it a new turn. I believe now; that I should perhaps catch on to the things in which I am naturally good at, and be done with the reckless experiments that I had been conducting with my life. So here I am implementing statistics in communication systems and practicing guitar regularly for almost three months. But, somehow wacky ideas just keep coming to me; so I decided I would blog about it this time, rather than raise an in-marketable start up upon it again.

 I have been sitting in IIT Delhi's library for seven hours daily now; and it's here that I often have conversations on life, music and the balance required in it, with one of my friends. So these conversations lead to inception of two rather serendipitous and stupid start-up ideas for apps in my mind.

Rating songs, without user feedback: Every song and every genre follows a pattern. For example, a typical alternative rock song would begin with a really low intensity verse, followed by a chorus which is usually of relatively high energy, it usually falls back to the second verse having the same intensity or a medium intensity with a few variations in the rhythm (which is usually increase in notes per beat, or increase in frequency of the notes on the lead guitar). The song appeals to us if the difference in these intensities are significant enough to push us into a mood. There must be a constant rise and fall in the intensities for the song to be laudable by a significant percentage of listeners. A statistical model can be developed on these songs, to identify verse, chorus, bridge and leads. Furthermore a Markovian modeling (predicts what should a following event/mood be, based on the past information/moods) can be used to predict what should be the suitable time for the rise and fall of the intensities (A verse of fifty seconds followed by a chorus of ten seconds can be rather dull to the ears). Realization of this idea can be pretty useful for upcoming artists; who are looking forward to attract an audience for their songs. 

Forecasts: Even though this idea is not to be believed in (or at least shouldn't be believed in); it may push astrology out of business. Our life is also a function of rise and falls; our emotional states, if we are happy today; we know sadness will follow soon. The faster we climb the ladder to feel triumphant, we crash down that ladder equally fast. For those who believe in 'Karma'; bad things happen to people who do bad deeds to others, and people are rewarded fairly if they work towards redemption. I believe that mostly; we are not able to predict what may follow next in our life; because we can't recapitulate what deeds we did in our past.  The idea is strangely heretic, however if a person fills up a questionnaire with basic yes/no questions on his life, daily (for eg. 'did you hurt someone today?'; kind of a psychology questionnaire); there is a discrete data on his life's past which can be recorded. Based on this past data; and a few other external factors a statistical model can probabilistic-ally predict what mood of events should follow next. This application is as ridiculous as astrology, however it's more logical. And a person who wastes his ten minutes daily reading an astrological forecast, can definitely waste two minutes in filling five yes/no questions.