I was reading a really, really, REALLY old dirt bike test the other day, and they happened to include a quarter-mile time, as well as a 0-60 time in that. I think that's just about the best way to figure out how the bikes do engine-wise. Dynos are great for finding what modifications do to a bike, but unless ALL the variables are held constant in ALL situations, they fall short for comparative data. Seat-of-the-pants is perfect for deciding what you're comfortable with, but it can be deceptive in terms of pure winning. For example, in the first issue of Dirt Bike, they tested three European 250s, and Super Hunky found that one of them felt much slower than the others, but still posted lap times three to five seconds a lap faster. I'm going to leave it open to the imagination which bike that was in this CZ - Maico - Husky test.
But, when you want to win, the clock is ultimately the one who decides the winner, so I think it should be the tool we use to figure bikes out. Get lap times, obviously, but instead of talking about which bike felt the fastest or pulled the dyno off the floor, have someone on the Magazine test crew who is the master of drag racing dirt bikes, and tell him to post some numbers on all the bikes. That would weed out the bikes that post big dyno numbers (like that one year CR250) or the holy-shit-this-feels-faster-than-it-is bikes, and tell the world what can really lay it down, and what can't.