Optimal weights for simple rules

Dear Forum,
I need some clarification on the choice of the optimal weights for the optimal simple rule problem. It is clear to me that the objective can equivalently be expressed as minimizing a weighted sum of variances and covariances of endogenous variables.
However, what is not clear is how the optimal weights are found.
Surely, an agnostic approach can be taken as in monfispol.eu/reports/report_1.1.2.pdf ( A complete example on page 5) where
lambda1 = 2;
lambda2 = 0.1;
but my question is what justifies the choice of these weights?
Any help is as always greatly appreciated.

See [Optim Weights in Optimal Policy Rules)