Interesting sounding paper here, and I'll just cut and paste the abstract:
Since the end of the twentieth century, global mean surface temperature
has not risen as rapidly as predicted by global climate models1, 2, 3 (GCMs). This discrepancy has become known as the global warming ‘hiatus’ and a variety of mechanisms1, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17
have been proposed to explain the observed slowdown in warming.
Focusing on internally generated variability, we use pre-industrial
control simulations from an observationally constrained ensemble of GCMs
and a statistical approach to evaluate the expected frequency and
characteristics of variability-driven hiatus periods and their
likelihood of future continuation. Given an expected forced warming
trend of ~0.2 K per decade, our constrained ensemble of GCMs implies
that the probability of a variability-driven 10-year hiatus is ~10%, but
less than 1% for a 20-year hiatus. Although the absolute probability of
a 20-year hiatus is small, the probability that an existing 15-year
hiatus will continue another five years is much higher (up to 25%).
Therefore, given the recognized contribution of internal climate
variability to the reduced rate of global warming during the past 15
years, we should not be surprised if the current hiatus continues until
the end of the decade. Following the termination of a variability-driven
hiatus, we also show that there is an increased likelihood of
accelerated global warming associated with release of heat from the
sub-surface ocean and a reversal of the phase of decadal variability in
the Pacific Ocean.