The crux of the argument is that the error bands on the last 10-15 years of temperature data show that the trend isn't cooling, and that in fact it's just a slow decade and about to be followed by a warm one. And we all know that can happen.
Problem is, the error bands are significant relative to any cherry picked "things are warming up" 8-15 years OR any "things are cooling down" 8-15 years and as such, no rigorous conclusion can be drawn.
We'll know more in a decade. Sadly that's how it works when you're measuring stochastic processes.