I do believe most of the above posts certainly answer the question at hand, especially those with real experience.
The crack thump method is too much of a 'vacuum' idea. In a perfect world, where all snipers used the exact same caliber rifle, loaded to the exact specs, shot from the exact same weapon system, under the exact same environmental standards, with no human error, hypothetically, yes, the crack thump method would work SHOULD work, to an approximate amount.
However, when looking at the above statements about being in the pits, the fact that a sniper who misses one shot will be DOPED in for the next shot before you finish counting, all specs are not the same for all rifles, bullets, and projectiles, and human error is a significant issue in every task in life, then no, the crack thump is not truly a probably method of ranging. A human is not capable of counting to the .xx of a second at a definitive point in time. To elaborate, no human is going to see the splash of a bullet and instinctively begin to count by the .xx of a second. The 'that sneaky mother fucker' comment, after just dropping a load in your skivvies will have already taken up a split second. Or the natural instinct of to get further behind cover will take a split second. Then you're stuck at 1.5x of a second, rather than the 2.xx of a second it truly was. According to your hypothesis, of approximately 300 yards per second, this leaves you with a disadvantaged 450yard SWAG compared to the true 600y range. Another fatal flaw added to the list.
Do you see all of the points being made above? In a vacuum, with all variables known, calculations already started, and a robot to begin counting to even the .xxxx of a second, and with the agreed upon "I shoot once and won't move, then you shoot once and don't move" method by both snipers, then yes, it is highly likely that the method you are question would work; however, in real life, when we are not in a vacuum, all variables are NOT know, and human error is playing a key factor (i.e. The sniper shooting at you missed due to human error. The system was capable of making the shot, but due to a miscalculation somewhere (distance, wind, etc) he missed his shot because of human error) then no, the method is not likely to be successful.