We don't need to use proxies to estimate the termperatures when we have actual temperature records. It seems like you're trying to make something sinister out of using the best available data.
You kind of do need to if you want to make claims like the one made by the GP that I was responding to. Specifically saying:Never before in history has the temperature changed with more then 1 degree over 100 years.
According to the proxy data that shows that historic stability, the temperature hasn't changed by that much since 1900 either. That is a very important observation to make as it shows the claim is lacking in substance.
In addition to the availability of actual direct measurements since the 1900s, which greatly reduces the value of more recent proxy values, there are also some problems with getting recent data from some of the proxies. For instance, there is the divergence problem in dendroclimatology which seems to show that pollution (or other factors) may be inhibiting tree growth since the industrial revolution. I imagine there's probably some difficulty with ice cores as well, since it may be difficult to get recent temperature measurements from glaciers that are shrinking.
Yes, Mann's paper from 2007 notes the same problem you mention. The calibration phase has a consistent problem with early calibration late verification. If you calibrate to the instrumental record from 1900-1950 and then verify the resulting proxy data from 1950 onwards you get observed evidence for a systematic bias in the underestimation of recent warming. Mann's words, not mine. The proxy data isn't registering the recent warming that we see in the instrumental record. I suppose that's no reason not to continue assuming that the same proxy data would have picked up similar warming in the past in your opinion. I'm less convinced. If you look closer at Mann's paper, he also observes the EIV methodology is least susceptible to the bias, and is thus the best method, again in Mann's opinion, not mine. If you check the reconstruction graphs though the EIV goes higher than the 1850-2006 average in 600AD, 1000AD and 1400AD. Mann doesn't graph the EIV from 1900 onwards, but in the calibration problem discussion he shows some portions of it, and it sits no higher than the 1000AD records for EIV from 1950 through 2000. It's all there in Mann's paper and please correct me if I'm in any way misrepresenting things. I've been trying to follow and understand this correctly for a long time now.
Of course, there are actually people studying these problems and even the quality of the calibration period which is used to match instrumental records to proxy values, you might want to try searching with the "proxy calibration" keyword, or reading up on the divergence problem in dendroclimatology.
As I noted above, I've done searches on that and come up with the observations above. The literature basically agrees that the proxy data, for reasons unknown, has a systematic bias in the underestimation of recent warming.
Forgive me for pointing that out when people make the claim that the historic record shows the last century is an anomaly. The instrumental record is anomalous compared to the proxy data. Attributing that entirely to climate change and not the fact those are entirely different data sets and methodologies is dishonest in the extreme. There's a reason the scientists writing the papers have a lot of caveats on the results regarding this, dropping those to make a blanket statement to scare folks is misrepresenting the facts.