The problem is that we cannot just measure temperature over a time span that suits our argument that would incredibly irresponsible.quote]
The reason why 1000 or 2000 years is typically chosen for this sort of analysis isn't because there is much evidence that temperature variations were any greater in the holocene (up to around 10 000 years ago) but because the temperature evidence becomes much more suspect. (to be technical the proxies diverge, leading to a much wider confidence interval). Of course a sceptic would turn around this argument in a simplistic way ignoring evidence that suits their belief. In contrast the graph that you posted suits a sceptics scale of events totally ignoring the relative stability of temperatures on a 1000-10000 year timescale which is far more relevent to determining anthropogenic effects. It is the relative stability of the climate evidence that provides brute raw statistical evidence to man's effects, without any technical knowledge of greenhouse gases, solar radiation and the like.
However, for the sake of rigour we cannot accept anthropogenic effects on data alone, we need the theoretical and modelling to back it up and it generally does. So I would say that the raw data provides something of the order of 90% certainty, the calculations and models only add to this, that is why I think the 90% certainty statement by the IPCC is arm twisting by the politicians, it is more like 99%.