Unproductive

The last few days have been quite unproductive...I have read a few papers and a cple of books but apart from that i havent had a single idea worth its salt. Am just letting things settle down in my head.

A question thats interesting me is

How does one ensure global consistency of estimates while all the observations that you have are local.
How can global order be perceived as a direct function of local order.
I can see that the bias-variance trade-off of estimators has profound impact on whether local solutions agree globally but am still not exactly clear about what the implications are :))