That's not necessarily true in all cases - when we're running studies with computerized trials in the lab, it's very common to run a few test trials first to familiarize the participant with the format, before you can actually get an accurate result.
ETA - last night I think my best score was .24-something, but I was going on less than 8 hrs sleep in two days. Tried it today after sleeping much better last night, and I wasn't *that* much faster than last night, all my scores were around .22. Just tried it now after getting out of the shower, and I finally broke out of the Bobbling Bobcat category. I'm now a Rocketing Rabbit with .1862. Why this makes me so happy, I have no idea.