[arghwhat]

This is neither a micro-benchmark nor a scam, but it is click-bait by not mentioning jq specifically.

Micro-benchmarks would be testing e.g. a single library function or syscall rather than the whole application. This is the whole application, just not one you might care that much for the performance of.

Other applications will of course see different results, but stuff like enabling LTO, tuning THP and picking a suitable allocator are good, universal recommendations.