There are fluctuations of no more than 5% in C14 production, but over the time period in question—C14 is good for dating back about 50,000 years—the fluctuations even out to a mean, which is actually a very good statistical basis on which to base calculations. Furthermore, the C14 creation in any year can be checked against other data, such as tree ring data, allowing for the C14 content of a sample to be calibrated to known variations.
C14 dating is not a “best guess.” On the contrary, because there is enough corroborating evidence from other sources, it is a very good method of determining the age of a recent (geologically speaking) specimen.
That is incorrect. There are documented fluctuations substantially larger than 5%, such as the one I referenced in the late 8th Century.
C14 creation in any year can be checked against other data, such as tree ring data, allowing for the C14 content of a sample to be calibrated to known variations.
However, tree-ring data is incomplete. And tree-rings are not produced at a rate of exactly one per year. Sometimes there are more, sometimes none. There are gaps and variations and it only goes back so far. Tree rings, however, are the main source for knowing that there are fluctuations.
C14 dating is not a “best guess.”
Don't get me wrong - C14 dating is a very good tool, but it is not as accurate as it is portrayed to be. Significant amounts of calibration are required, and dates measured even against tree rings can have large discrepancies. And even direct measurement of tree-ring C14 doesn't match the 'counted' age of the tree very well much of the time. Aside from trees, artifacts of known age are used to benchmark the levels of C14 for given dates. So we're using C14 to date things and we're using things to calibrate the C14. See the problem? Layered uncertainty. Large error ranges, especially as you go back further in time to before there are a lot of carbon artifacts.