Efficient estimation asks how tightly estimates can cluster around an unknown parameter. Fisher information measures how much the observed data reveal about that parameter: when small parameter changes produce large changes in the likelihood, the information is high. section 3.2.2
Fisher information also sets a theoretical limit on the precision of unbiased estimators through the Cramér–Rao lower bound. An estimator that reaches this bound is efficient; maximum likelihood estimators often approach it as sample size grows. section 3.2.2

