Background
The authors introduce their study by stating the obvious (but extremely important) fact that "comparing climate modeling results with historical observations is important to further develop climate models and to understand the capabilities and limitations of climate change projections [our italics]."

What was done
Kiktev et al. analyzed the abilities of five global coupled climate models that played important roles in the IPCC's Fourth Assessment Report to simulate temporal trends over the second half of the 20th century of five annual indices of extremes in surface temperature (annual percentage of days with Tmin < 10th percentile, with Tmax < 10th percentile, with Tmin > 90th percentile, with Tmax > 90th percentile, and annual number of frost days, i.e., Tmin < 0°C), as well as five annual indices of extremes in precipitation, the observational data for which analyses they obtained from the HadEX global data set that contains gridded annual and seasonal values of the ten extreme indices that are calculated from series of in situ daily measurements (Alexander et al., 2006).

What was learned
The international research team, hailing from Australia, Japan, Russia and the United Kingdom, found that "the results mostly show moderate skill for temperature indices and low skill or its absence for precipitation indices [our italics]."

What it means
If we were personally going to use climate model results as the basis for mandating a complete overhaul of the world's energy system - as the world's climate alarmists are attempting to do - we would like to have those models possess considerably more than moderate skill at what they do. We would definitely not want them to have low skill. And to employ models that have an absence of skill would, to our way of thinking, be insane.

Reference
Alexander, L.V. et al. 2006. Global observed changes in daily climate extremes of temperature and precipitation. Journal of Geophysical Research 111: 10.1029/2005JD006290.