I mentioned the discovery of a hypergeometric combinatorial enumeration formula with a useful decomposition that could be used to find a limit probability distribution for the objects I was enumerating. I edited out the following: Richard Stanley had previously derived a rational generating function for enumerating the almost-injective functions, however, I could find no obvious way to manipulate this to obtain the limit probability distribution, nor did it suggest the decomposition of a non-machine summable formula into machine summable, asymptotically significant and non-machine-summable, asymptotically insignificant parts. The two-term formula was obtained by classifying the almost-injective functions using a regular grammar; the closed term arose from a sub-language closed under concatenation. Computer experiments suggest that a class of enumeration problems arising from regular monoids have closed form hypergeometric enumeration formulas.
Finding an algorithm for computing closed forms of indefinite binomial coefficient sums was an open problem in Knuth's Art of Programming; it was solved by Wilf and Zeilberger. In the example I considered, there is no closed form solution; however there is a way of finding a decomposition of the sum with good properties. Finding the asymptotically significant term required an idea: the point is that the algorithms I used don't help you find this. One of those algorithms I used was due to R. W. Gosper, a legendary hacker.
I actually care a lot about the last question. It often makes me give an application a second look. I don't think the other 3 YC partners care very much about it though.
I am way more curious to hear about the best answers to this one you've heard from this or previous rounds. Someone should remember to ask again after the 10th.
"In 2005, Matt discovered the best way to deal with Nigerian 413-scams on CraigsList: when someone offers thousands of dollars for your stereo via some phony money-wiring organization, just request a receiver-pays-shipping slip and FedEx them 40 pounds of bricks."
I talked about choice and the sensory overload it often causes. We have reached a point where the only type of cell phones we can't buy anymore are cell phones that are just a phone. The suppliers, however, cannot be held responsible for this. Afterall, they are simply trying to satisfy the users constant demands. So eventhough this mental toxin often makes us feel miserable, it is preferable by consumers over having less choice. And this fascinates me.
I am contemplating writing an article about choice and relating it to Web 2.0. Here is an excerpt from another article I wrote that relates to this: "...if a choice is present, perfectionists spend a ridiculous amount of time over simple decisions. Such is the mentality governing users of today's social networks. We are caught up in a web of choices; indeed, the core reason behind MySpace's success is that users rule. MySpace might suck, but it gives its users ample of choice. Once a user has invested an hour of their life customizing their profile, they are not likely to leave the site anytime soon."
I crawled the MIT facebook and built a friendship predictor that will take guesses at who, out of the people you are not currently listed as being friends with, you actually are friends with in real life. People's traits (age, sex, home state, etc.) are significantly less predictive than how close you are on the [ friendship / class / photo / group ] network. In descending order of predictive power, the most important traits when considering two people as potential friends are:
how many friends you have in common
how many pictures you simultaneously appear in
how many classes you have together
relationship status
looking for (e.g. friendship, dating, random play)
how many facebook groups you have in common
I couldn't think of anything that didn't sound trite when I first filled out the form, so I left it blank and decided to come back to it later.
(As far as I'm concerned it's a long shot. I don't think my idea and my goals for my business line up that well with what YC is really looking for, based on what they've done in the past. So I filled out the form more for myself than for them.)
But then I thought of the best answer to that question -- the most personally meaningful thing I've discovered, something that I don't think anyone else has ever noticed -- over the weekend, and only realized this morning that the deadline had whooshed right by while I was sleeping.
There is great knowledge in the classics. Patterns of human nature always repeat. WhenI know the mind of the other person, I know who their allies are. I also know their team and the position they stand in. Now take this point of view and apply it in your software app. It is a paraphrase from Sunzi AoW Chapter 3.
I thought that "discovered" was very ambiguous (was it intentional?)
I wanted to write something offbeat, like, "The Dawson City mayor stole cable TV for all its citizens in 1996" or "DNA-polymerase can synthesize 10000 nucleotides/second" (I can't find the actual number on the net, but it's crazy high)
Vervet monkeys have three main predators, and can alert the rest of their group to the presence of each. The rest of the group responds appropriately to the type of call given. If, for example, a monkey gives a loud, barking call, signifying the approach of a leopard, the other monkeys run to the trees.
I had the chance while in France to visit Futuroscope by luck. When I found my self in this Cosmos attraction, a huge sphere room watching the universe as a spectator like you are among the stars, has caused me one of the most undescribable, fascinating feelings in my life.
I advice it highly, if you ever find your self in France, go to Futuroscope!
Most of the attractions are in AMAZING 3D!
Looking up at the nigkt sky has that effect on me too. What makes it worse is that whenever I mention it to anyone else they clearly don't understand why.
Finding an algorithm for computing closed forms of indefinite binomial coefficient sums was an open problem in Knuth's Art of Programming; it was solved by Wilf and Zeilberger. In the example I considered, there is no closed form solution; however there is a way of finding a decomposition of the sum with good properties. Finding the asymptotically significant term required an idea: the point is that the algorithms I used don't help you find this. One of those algorithms I used was due to R. W. Gosper, a legendary hacker.