Ks are guaranteed outs. Non-hit contact events have the possibilty of a positive outcome. Errors, moving runners over, and even RBIs. That said, the DP is big negative. Look at the delta from the Expected-Runs matrix for each event. The first number is the expected runs for the situation, the second is the expected runs after the DP (with the lead runner advanced where applicable), the third is the delta:
Runner on 1st, no outs: .816, .105, -.711
Runner on 1st/2nd, no outs: 1.442, .314, -1.128
Loaded, no outs: 2.810, 1.314, -1.496
In the loaded situation, notice how it's 1.314 after the event. That is the same as 1st/2nd, but with a run scored, so even though you are actually scoring there, it's still a big negative. Here are the 1-out scenarios:
1st: .571
1st/2nd: .720
Loaded: 1.788
They obviously all drop to zero.
There are no appreciable benefits in the matrix in any situation where the runners are moved over for an out. Of course, it's important to remember these are theoretical expectancies, and some times you beat it while others you fall short. Due to that theoretical nature, I'll always take a run on the board. It is also important to remember that the alternative to the runner advancing outs is a K with the runners not advancing.
With all that said, it takes several runner-advancing plays to account for one DP in the theoretical matrix, so the overreaction to the K is unwarranted.