[invisal]

The problem imho is that people aren't able to interpret it correctly,
because they don't understand what the time complexity actually says.

I am not saying that the time complexity using the big O notation is useless. Just it
is broken as a mean to measure algorithm performance. Lets say that I want to
choose between "insertion sort" and "selection sort". The average complexity of both
algorithm are O(n^2). Without seeing how the algorithm works, it is hard
to make decision which one to choose to suit my current data.