A Hacker News user is puzzled by conflicting performance metrics between Lichess's browser analysis and their local Stockfish setup. On a Redmi Note 14 Pro, Lichess reports nearly 1 million nodes per second, while the local Python script shows only 600,000. Despite Lichess's higher reported speed, the local machine actually reaches depth 30 in 53 seconds, beating Lichess's 2:30 time.

The confusion stems from how speed is measured and displayed. The user suspects Lichess uses instantaneous readings, while local setups often show averages. Search configurations also differ significantly; Lichess likely uses continuous search with frequent evaluation updates, whereas a local script might restart searches or manage hash tables differently. These factors can skew perceived performance, making direct comparisons difficult.

Ultimately, the discrepancy questions whether 'depth 30' is comparable across different interfaces. Commenters suggested checking for Fishnet offloading or inspecting Lichess's open-source code for engine interaction specifics. The user is seeking clarity on why a setup reporting lower nodes per second can finish a search faster than a browser-based tool claiming superior speed.