Interesting. I'd interpreted it as "we'll start small this time, and adjust things for next time, if there is a next time". Usually an experiment (in science) implies that you don't change the conditions of the experiment while it's running, you let it run and record the results, whether they're positive or negative. Otherwise, you limit your ability to be surprised, which is the whole point of science.

It gets a little more muddled with social experiments, where it's often hard to delineate an "ending point" of the experiment, and there's often real collateral damage in the process. Even with social experiments, though, you get a more accurate signal if you let them run past the point where your gut tells you it's a bad idea.

Surely the point of an experiment is to run it. Observe the results. Then change things in the next run.