An experiment can be a comparison, or a gate, depending on the intent, not its mechanics

An experiment run is just that - a run. How you use its results defines the purpose. It’s a gate if the answer it gives is yes or no (ex. should we carry on, or move on to a different architecture?) It’s a comparison if it produces a ranking “model A > model B > model C etc”. This distinction defines the next steps.

see Unlike inter-hypothesis, running intra-hypothesis experiments allows comparing models ceteris paribus for a co-existing independent dimension of experiment classification