Random Sampling
l [Gabow] scheme great if m, c small
l Random sampling
» reduces m, c
» scales cut values (in expectation)
» if pick half the edges, get half of each cut
l So find tree packings, cuts in samples
Problem: maybe some large deviations