Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Statistics serves as a tool to overcome our cognitive biases. But what if these biases are at the center of learning?

Take for example the Gambler's Fallacy where a player believes she can predict the outcome of a coin toss with greater certainty than is possible. Obviously she cannot. But if I had to design a Machine Learning algorithm, I would certainly want it to always assume that a pattern existed. That way, if the data were predictable, the algorithm would be able to take advantage of it.



A smarter algorithm would refrain from betting on the coin.


That's great if you know a priori that you cannot predict the outcome of what you are betting with.


I don't think it would be necessary to know a priori that something is unpredictable; failing to find a pattern after some number of observations should allow an algorithm to determine that the event is not predictable by that algorithm.


The trouble is that you can't reliably distinguish random data (or data determined by factors you're not considering) from patterned data.


How do you think random number generators are tested?


A smarter algorithm would consider the payoff as well as the odds. Then it may bet or refrain.


What if the coin had the best odds compared to all other games?




Consider applying for YC's Fall 2026 batch! Applications are open till July 27.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: