Most times I encounter this random metaphors for metaheuristic optimization they seem like a modified genetic algorithm to me. What do you think? Is there a way to remove the analogies/metaphors to better compare algorithms?
If you want to compare the ideas involved on the various metaheuristics, a good starting point for the comparison is to think of them as Population based vs Trajectory based. Population based metaheuristics works with the concept of create a solution that mix components of good solutions. Trajectory based metaheuristics works with the concept of creating a solution and iteratively do improving modifications(moves) on it. With the image below, you can visualize this dichotomy and other possibles parameters for comparison.
To compare performance, a common approach is to use time to target. Given a known solution and its objective value, how fast my metaheuristic achieve this value? This site provides a tool that helps in this task.
The new metaheuristics are pretty much genetic algorithms and local search in disguise. And this is actually a problem and hinders progress. There is a good write-up on this topic by Sörensen, Kenneth. "Metaheuristics—the metaphor exposed." International Transactions in Operational Research 22, no. 1 (2015): 3-18.
The argument is that the "new" metaheuristics are not, in fact, new. They just reheat and repackage concepts, ideas, operators, subroutines from classical metaheuristics under a new name. While among all those repackaged ideas there is some "truly innovative research of high quality", it is hard to filter them out from all the noise, and hard to ascribe credibility to those new ideas.
Also have a look at The Evolutionary Computation Bestiary.