Pehaps you were being ironic, but don't cheer just yet: this sword is double-edged. The gap between climate models and the results found in studies of paleoclimates in the (distant) geologic past is one of the key dichotomies in paleoclimate science. Climate scientists shuffle their feet a lot when it comes to the error range of the CO2 concentration levels found when reconstructing geologic paleoclimates, particularly once you move beyond ice-records. This error range forms the key problem for solving the dichotomy: the foundation of correlating (and attributing) paleoclimate temperatures to past CO2 concentrations remains shaky.

While it's true the gap has been narrowing - something is still very wrong, it's just not clear what is. One option is that modern CO2 sensitivity estimates are far too low, which would be horribly bad news, or alternatively, proxies used to reveal paleaoclimate conditions render unreliable results - which would be horribly inconvenient for the hundreds of studies that already have been done. Or, as the article mentions, there's a factor in play not fully realised, such as the role of methane. Which makes the message of the PNAS article the more relevant.

Full disclosure time: I provide PR for a climate change research collaboration that pursues, amongst others, the development of climate proxies that provide greater accuracy of CO2 in the geologic past. Also, the first author of the research publication reported on by The Atlantic was a fellow student when we were still playing in the sandbox at university (though he might only remember me with some effort).