Carbon dating is about matching radioactive carbon contents in organic matter to samples from trees or other known-age organic matter. As such, it has a limit of ~50,000 years ago.
Historic levels of atmospheric CO2 is mainly based on samples of gas trapped in ice sheets, or estimates based on carbon quantity in mineral sediments that would have absorbed atmospheric carbon in their formation. As such, estimates go back to the entire history of the Earth.
Especially for recent history, where we have bubbles of air from those times still trapped in Antarctic ice, I see this is a remarkably direct measurement, the very opposite of a "farcical amount of guesswork". If we're talking about 1 billion years ago, I start to agree with you a little more.