How reliable are tree rings for climate modeling?

[Mk]

As historical thermometers, tree rings are so bad scientists have to cherry pick samples, removing those that do not exhibit a desired response and so inconsistent as to be useless for temperature reconstruction.

I won't say who said it, but does this sound right to you? Dendrochronolgists on TV all say it so easy, you count the tree rings and the space in between, and that will tell you about the temperature and precipitation and how long the tree lived.
Also:

As historic thermometers, tree rings are adequate, because confounding factors like moisture availability, nutrients, etc... are accurately known

This sounds kind of dumb, but what about carbon dating? We assume the ambient carbon (or something like that) stayed the same the whole time. But didn't it change? What about between mass extinctions?

 

[BobK]

I don't see how tree rings can be reliable.

Various factors go into a tree's growth pattern. Fertilization, sunlight, precipitation, temperature, and probably others I haven't considered. All of which can vary in different ways over the years.

I frankly think they are fooling themselves when they say they can extract accurate temperature data from the ring width or density.