Or were judges being too nice? We know there was bias, but we still don't know when bias occurred.

I feel the need to point out that the observation does not necessarily result in a bias. We literally know nothing about the legal system's arrangments for parolees here other than this single data point.

It could be, for example, that there is an understanding that the review boards arrange cases before judges based on the boards' estimate of the potential parolee's worthiness of release; with the 'worst offenders' being later in a giving hearing bracket. This would also take the shape of a linear decline in parole-granting rates -- but would not represent any bias in the judge's reasoning.

(In case this is not clear, as I appear to sometimes have difficulty communicating with others on this site: My only positional claim here is that we do not know if this is actually a bias at all, as we have too little data to make such a definitive statement. All else is merely explanation of this stance.)