Radiometric dating critics simply don't understand how it works. You take a radioactive parent isotope, deposited at an assumed constant to today (See: Uniformitarianism) and as it decays at a known rate, you use the half-life (the time it takes for half of the amount to decay to a stable isotope)to determine how many times it has been halved, which gives you a time frame. Obviously, as with any paleo, the focus decreases as you go back in time. Different types are used for different things- for example, Carbon dating is only good for like 5730 years in the past or something. We use a lot of volcanics in this because volcanics reset the decay clock, so you can measure from when it was deposited originally. It doesn't really work for sedimentary or metamorphic- so once you know the age of a dyke or a sill or something, you can use strat + sedimentology to determine a pretty decent age of the rock. That's why there is always a range and that range gets bigger as you go further back.