Beyond chain-of-thought: Scaling reasoning width and depth via parallel thinking | Dark Hacker News