Tamino has been making a strong case that the trend of surface temperatures isn't any different than it has been since 1975, except for some ordinary fluctuations about the trend. He's giving away his graphs for free, and they're worth reproducing.
Here, in 5 short steps, is his proof that there is no pause.
First, take the Cowtan & Way data for surface temperatures, which is HadCRUT4 infilled by kriging. (This is a nice introduction to kriging, if you're looking for one.)
Plot the annual anomaly value from 1975 to 2000:
Determine the linear trend:
and the uncertainty of that trend:
In this chart, the first dashed line on either side of the trendline (which is the solid line) is the 1-standard deviation value, and the second dashed line outside it is the 2-standard deviation value.
Project these trend lines to the present:
Add in the annual Cowtan & Way anomalies up to 2013:
What you see is that the present (last) anomaly is not even one standard deviation below the 1975-2000 trend. No different than it's often been before.
It's hard to make it more clear than this. By calculating the trend only "since 1998" or only "for the last 17 years" or the like, with nonkriged data, you're only presenting a piece of the picture -- a highly selected piece -- a picture where the early data is above the long-term trendline, and the later data is below the long-term trendline. (And not even below it by that much.) It's the SkS escalator graph but looking just at the last step.
By the way, the C&W trend since 1975 is 0.17 ± 0.02 °C/decade (1-sigma, OLS uncertainty, R2 = 0.83). Tamino's value for sigma includes autocorrelation, which is why it's larger than the one here.
As he wrote, "If that’s what you call a pause,' then it’s not a very impressive one."