This Hacker News thread discusses an Ars Technica article about creating fair dice from random objects, focusing on designing physical shapes with specific probabilities of landing on different sides. Commenters discuss the long-term stability and robustness of these non-traditional dice compared to standard ones. One commenter provides a method for creating a fair coin from a biased one by flipping it twice and only accepting results where the outcomes differ. Others point out limitations of this method with extremely biased coins. The thread mentions Keenan Crane, a highly regarded researcher known for communicating complex topics effectively. One comment references a Roman rock crystal icosahedron die. A commenter mentions a classic quant interview problem involving creating fair outcomes from biased events by exploiting the commutative property of multiplication, but this is corrected as unrelated to the article's focus on physical shape design. The discussion touches on both physical and mathematical approaches to generating randomness.