Some doubters have tried to dismiss geologic dating with a sleight of hand by saying that no rocks are completely closed systems (that is, that no rocks are so isolated from their surroundings that they have not lost or gained some of the isotopes used for dating). Speaking from an extreme technical viewpoint this might be true--perhaps 1 atom out of 1,000,000,000,000 of a certain isotope has leaked out of nearly all rocks, but such a change would make an immeasurably small change in the result. The real question to ask is, "is the rock sufficiently close to a closed system that the results will be same as a really closed system?" Since the early 1960s many books have been written on this subject. These books detail experiments showing, for a given dating system, which minerals work all of the time, which minerals work under some certain conditions, and which minerals are likely to lose atoms and give incorrect results. Understanding these conditions is part of the science of geology. Geologists are careful to use the most reliable methods whenever possible, and as discussed above, to test for agreement between different methods.
Some people have tried to defend a young Earth position by saying that the half-lives of radionuclides can in fact be changed, and that this can be done by certain little-understood particles such as neutrinos, muons, or cosmic rays. This is stretching it. While certain particles can cause nuclear changes, they do not change the half-lives. The nuclear changes are well understood and are nearly always very minor in rocks. In fact the main nuclear changes in rocks are the very radioactive decays we are talking about.
There are only three quite technical instances where a half-life changes, and these do not affect the dating methods we have discussed.
1. Only one technical exception occurs under terrestrial conditions, and this is not for an isotope used for dating. According to theory, electron-capture is the most likely type of decay to show changes with pressure or chemical combination, and this should be most pronounced for very light elements. The artificially-produced isotope, beryllium-7 has been shown to change by up to 1.5%, depending on its chemical environment (Earth Planet. Sci. Lett. 171, 325-328, 1999; see also Earth Planet. Sci. Lett. 195, 131-139, 2002). In another experiment, a half-life change of a small fraction of a percent was detected when beryllium-7 was subjected to 270,000 atmospheres of pressure, equivalent to depths greater than 450 miles inside the Earth (Science 181, 1163-1164, 1973). All known rocks, with the possible exception of diamonds, are from much shallower depths. In fact, beryllium-7 is not used for dating rocks, as it has a half-life of only 54 days, and heavier atoms are even less subject to these minute changes, so the dates of rocks made by electron-capture decays would only be off by at most a few hundredths of a percent.
2. Physical conditions at the center of stars or for cosmic rays differ very greatly from anything experienced in rocks on or in the Earth. Yet, self-proclaimed "experts" often confuse these conditions. Cosmic rays are very, very high-energy atomic nuclei flying through space. The electron-capture decay mentioned above does not take place in cosmic rays until they slow down. This is because the fast-moving cosmic ray nuclei do not have electrons surrounding them, which are necessary for this form of decay. Another case is material inside of stars, which is in a plasma state where electrons are not bound to atoms. In the extremely hot stellar environment, a completely different kind of decay can occur. ' Bound-state beta decay' occurs when the nucleus emits an electron into a bound electronic state close to the nucleus. This has been observed for dysprosium-163 and rhenium-187 under very specialized conditions simulating the interior of stars (Phys. Rev. Lett., 69, 2164-2167; Phys. Rev. Lett., 77, 5190-5193, 1996). All normal matter, such as everything on Earth, the Moon, meteorites, etc. has electrons in normal positions, so these instances never apply to rocks, or anything colder than several hundred thousand degrees.
As an example of incorrect application of these conditions to dating, one young-Earth proponent suggested that God used plasma conditions when He created the Earth a few thousand years ago. This writer suggested that the rapid decay rate of rhenium under extreme plasma conditions might explain why rocks give very old ages instead of a young-Earth age. This writer neglected a number of things, including: a) plasmas only affect a few of the dating methods. More importantly, b) rocks and hot gaseous plasmas are completely incompatible forms of matter! The material would have to revert back from the plasma state before it could form rocks. In such a scenario, as the rocks cooled and hardened, their ages would be completely reset to zero as described in previous sections. If this person's scenario were correct, instead of showing old ages, all the rocks should show a uniform ~4,000 year age of creation. That is obviously not what is observed.
3. The last case also involves very fast-moving matter. It has been demonstrated by atomic clocks in very fast spacecraft. These atomic clocks slow down very slightly (only a second or so per year) as predicted by Einstein's theory of relativity. No rocks in our solar system are going fast enough to make a noticeable change in their dates.
These cases are very specialized, and all are well understood. None of these cases alter the dates of rocks either on Earth or other planets in the solar system. The conclusion once again is that half-lives are completely reliable in every context for the dating of rocks on Earth and even on other planets. The Earth and all creation appears to be very ancient.
Short-lived isotopes can be slightly affected by external conditions, but they use decay processes that longer-lived isotopes do not. Short-lived isotopes are not used in determining geological ages, only longer-lived isotopes are used.
Also, the changes induced are very small. I once calculated the percent error required to make a 10,000-year-old rock falsely appear to be hundreds of millions of years old. It worked out to be multiple thousands of a percent, not a mere 1.5%, which is within measurement error.