Good point, but here is a possible rebuttal. Let me know what you think.
It is unlikely that this neutral gene when coupled with another mutation would create such a big difference as to overcome genetic drift. Unless the change causes such a big survival advantage that it survives while the rest of the community dies, both of the mutations in the combination will be subject to genetic drift (which usually maintains the status quo pretty well from what I understand).
If somewhere along the line the neutral mutation hooks up with a second mutation that it combines well with, then that second mutation still has the drift problem to deal with. In other words, assuming this "neutral" mutation is lucky enough to surive the drift, what it ends up being coupled with will also be whatever mutation happened to be lucky enough to survive the drift.
This implies that lab experiments would be accurate in simulations since in both the lab and in nature the combinations in the end result are basically random. The rate of mutaiton may be unusually high in the lab, but that just means we have more possible end combinations to look at.