The results on that page suggest that some of the posterior estimates (e.g., alpha2) are very far from zero. So constructing a prior with a mean of 0 and an SD of 1e-6 suggest that that those results are not very credible a priori. Another way to think about it is that those priors suggest that the observed data is (nearly) impossible. Once you set the SD to 1, the data is much more plausible as far as the model is concerned.