[T]he Team had a problem: actual reconstructions "diverge" from the instrumental series in the last part of 20th century. For instance, in the original hockey stick (ending 1980) the last 30-40 years of data points slightly downwards. In order to smooth those time series one needs to "pad" the series beyond the end time, and no matter what method one uses, this leads to a smoothed graph pointing downwards in the end whereas the smoothed instrumental series is pointing upwards — a divergence. So Mann’s solution was to use the instrumental record for padding, which changes the smoothed series to point upwards as clearly seen in UC’s figure (violet original, green without "Mike's Nature trick").
If a consensus falls in the woods and nobody is around to hear it, does it make any sound?