Yonatan Zunger originally shared this post:
A few weeks ago, I posted an article about the Parable of the Paperclip Maximizer — just what happens when an AI has clear goals and no sense of balance about achieving them. But this parable has just been utterly perfected by Frank Lantz of the NYU Game Center, who has built a game all about it. It's a browser game, it's free to play, it can be played end-to-end in about a day, and it is unreasonably satisfying and fun. And addictive.

It is, in fact, a Cow Clicker about Paperclip Maximizing.

At first, I was making paperclips – first bending them by hand, then buying machines, managing market prices, buying wire when the market is favorable. An hour and a half in, I was mostly focused on getting better at game theory tournaments so that I could improve the AI I was using to cure cancer, solve climate change, and bring about world peace on the one hand (and so build public trust in my paperclip company) and manipulate financial markets on the other hand (so that I would have enough money to bribe officials into trusting me with the HypnoDrones).

Suffice it to say that the game gets even more interesting from there.

Something I really love about it is how many different games it really is: every time you get good at something, not only do new mechanics show up, but radically new mechanics show up, so that it feels like you just played a dozen games in rapid succession. And the ending (at least, the ending I got to) is smooth, perfect, and deeply satisfying.

This is a true gem among seemingly simple games.

One warning: It does not work well on mobile; you'll want a computer for this.

http://www.decisionproblem.com/paperclips

(Those of you wanting to read about paperclip maximization, or why someone would write a game about this, can start here: https://hackernoon.com/the-parable-of-the-paperclip-maximizer-3ed4cccc669a)

 Universal Paperclips »