Have you checked the assumptions that are involved in radiometrics?
The only assumption I know of is that the half-life of any given radioisotope is assumed to be constannt.
As far as I know there have never been two samples of the same radioisotope found to have different half-lives, so the assumption seems sound.
Do you have evidence that the half lives can be made variable to account for a margin of error of 4500000% ?