Lucia at The Blackboard has an interesting article on how a global warming scientists can justify the kind of massive cherry-picking that went into the infamous hockey-stick.
First, an explanation of ‘proxies.’ Proxies are so called because we cannot measure past temperatures directly, but have to use stand-ins. These stand-ins (or proxies) may be growth rings or changes in glaciers or sea level or other indirect measures of temperature. The problem with all of these proxies is that temperature is not the only thing that affects them. So they need to be cross-checked and recorded very carefully.
Lucia points out that you can cherry-pick without even meaning to, simply by removing the proxies (sets of tree rings or whatever) that do not correlate with other records of temperature.
I am sure this is possible, but I am not so sure this is what happened in the Mann / Briffa hockey stick invention. The cherry picking in that case seems so clear it is hard to avoid the notion that it amounted to scientific fraud.
To be fair, Briffa insists there was no deliberate pre-selection of data. He now says there were problems with the methodology. We are working on it, he says. In the mean time, everyone should still believe it.