What was done
The authors compared, for the overlapping time frame 1962-2000, "the estimate of the northern hemisphere mid-latitude winter atmospheric variability within the available 20th century simulations of 19 global climate models included in the Intergovernmental Panel on Climate Change [IPCC] 4th Assessment Report" with "the NCEP-NCAR and ECMWF reanalyses," i.e., compilations of real-world observations produced by the National Center for Environmental Prediction (NCEP), in collaboration with the National Center for Atmospheric Research (NCAR), and by the European Center for Mid-Range Weather Forecast (ECMWF).

What was learned
Quoting the five Italian researchers, "large biases, in several cases larger than 20%, are found in all [our italics] the considered metrics between the wave climatologies of most IPCC models and the reanalyses, while the span of the climatologies of the various models is, in all cases [our italics], around 50%." They also report that "the traveling baroclinic waves are typically overestimated by the climate models, while the planetary waves are usually underestimated," and that "the model results do not cluster around their ensemble mean," which is another way of saying they are all over the place.

What it means
Quoting once again the scientists who performed the model tests, "this study suggests caveats with respect to the ability of most of the presently available climate models in representing the statistical properties of the global scale atmospheric dynamics of the present [our italics] climate and, a fortiori ["all the more," as per Webster's Dictionary], in the perspective of modeling [future] climate change." Indeed, it gives one pause to question most everything the models might suggest about the future.