The way to enforce honesty, and declaration of bias, is peer review. Part of your duty would be to evaluate the work of your other peers, and there would be reputations attached to people based on such review. You might also be called upon to review the work of other groups (in a more peripheral way) that you're not involved in, but know something about the field for.

We could even start with an anonymous review process, where you don't get to know the names of the other peers. (Later, you must learn their names to judge their bias declarations.) And where attempting to reach them out of band is punished.

"Punishment" in most cases, for both breaking the rules, and work judged insufficiently diligent by peers or PTO workers, would simply involve not getting full work credit for this particular patent. And thus you must keep going until you earn this work credit. You might have a certain time by which you must earn enough work credit, or if this is done to people who apply for, rather than get patents, you can simply make it that you don't get your patent until you have earned enough work credit.

Yes, all the peers looking at a patent might decide to slack off and declare all the other slacker's work as diligent. But there will always be one straight player in the group, and if there is, none of you will get your work credit. If it's judged deliberate, you might even get negative work credit.

For example, lets say one peer finds 20 pieces of prior art, and others find nothing, and some of the prior art is an obvious google search. Looks fishy.

However, it may simply be the PTO's job to judge this. You're working for them, and like any employer, they know how to judge the quality of a consultant they hire. You could elect to pass on any patent you are assigned that does not fit your skills (to a limit) as there will always be plenty to earn your work credit in the several years a typical patent takes.