I love the `[[likely]]` and `[[unlikely]]` tags since they nicely encapsulate the Modern C++ philosophy.

1. They don't work that well.

2. The intended use case is that you'd label an unlikely branch that you want to speed up as `[[likely]]`, which is confusing.

They are certainly motivated by good intentions (like the HFT use-case as mentioned in TFA, I remember that talk too, I think I was a volunteer that year for CppCon), but even after knowing both of my points above _they were still added to the standard_.