paint-brush
YC Remixby@j_jason_bell
263 reads

YC Remix

by Jason BellMarch 7th, 2017
Read on Terminal Reader
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

The <a href="https://hackernoon.com/tagged/rational" target="_blank">rationale</a> for <a href="http://ycremix.com/" target="_blank">YC Remix</a> is that there might be a synergy between humans and computers in creativity. Currently, humans are much better at evaluating ideas. This is because humans can interpret ideas rapidly, and because humans can represent ideas in a flexible way and then react to those ideas via thought experiments.
featured image - YC Remix
Jason Bell HackerNoon profile picture

The rationale for YC Remix is that there might be a synergy between humans and computers in creativity. Currently, humans are much better at evaluating ideas. This is because humans can interpret ideas rapidly, and because humans can represent ideas in a flexible way and then react to those ideas via thought experiments.

It seems possible that computers could eventually get a lot better at evaluating ideas, and in fact it’s something many people are working on. However, even though computers aren’t great evaluators, they can still add value in creativity mainly through randomness. Humans have natural constraints on their thought patterns: priming and framing, fluency and taste all constrain creativity. Computers have no such constraints.

YC Remix is a very simple tool. It uses a description template for startup ideas introduced by Paul Graham: <known entity> for <new target>. YC Remix randomly draws from a list of known entities and another list of targets and puts them together. The known entity list is the list of all YC startups that aren’t dead. The probability of picking a given startup is proportional to the Alexa ranking of that startup’s website. The list of targets is comprised of both previous YC startup targets (phrases appearing after “for” in the descriptions on yclist.com), as well as some targets related to the categories in YC’s requests for startups list.

When using YC Remix, some ideas won’t make sense at first. Some may not make sense at all. Some will be ridiculous enough to make you laugh. In my experience, it’s best to think deeply about the idea before moving on to a new one.

If I’ve learned anything in graduate school, it’s that better data are more important than a better model. YC funded startups are good data, so even something as basic as randomly combining terms might generate some diamonds. In addition, quality is not determined solely by the input data, but also by the interpretation of the user. After all, the new conventional wisdom seems to be that ideas are much less important that the humans who develop them. Still, good ideas are not worthless, and hopefully YC Remix can help generate some.