But those are unreliable as well...
See, since it seems that you are convinced yourselves, can you tell me how accurately the past temperature can be assessed by using this method? I mean +/- how many degrees?
> But those are unreliable as well...
Can you tell me how many research papers on this specific subject you have read, before coming to that conclusion?
Hint: chances are you're correct (!). Size of the error bars, timeframe(s) talked about, how data was gathered, how results were processed etc, that is what such papers are about. Educate yourself! (Wikipedia is usually a good start)
Can you tell me any answer you'd accept?