Compute default policy 1 mixture value.
0.5*0.4+0.5*0.9=0.65.
Why does the changed mixture favor policy 0?
Under the changed partner weights [0.9,0.1], policy 0 is worth 0.9(0.95)+0.1(0.2)=0.875 and policy 1 is worth 0.9(0.4)+0.1(0.9)=0.45. Its strong column 0 value 0.95 receives weight 0.9.
Which transfer policy wins?
Policy 1 is a generalist: it succeeds with probability 0.6 against every partner, so under weights [0.2,0.6,0.2] it is worth 0.6, while policies 0 and 2 are each worth 0.2.