I think OP is on to something here. In particular the study design makes it hard to tease out issues specific to common knowledge verses anchoring or bias towards ones own knowledge. Notably each individual's information set makes B look worse than A or C.

I think the study would have been more compelling if the common knowledge favored A and C but each individual's total knowledge was neutral between the A,B and C. If results favored not B in that set up it would indicate that folks were specifically anchoring on the common knowledge itself rather than anchoring on their initial hypothesis.

This is a group effect quite often seen in other places.

Tornado chasing is a good example. Take an experienced tornado chaser near a storm, and they will commonly back off at a safe distance when things are looking sketchy.

But if you take that same chaser and put them in a group of people of varying skill levels they can get complacent and stay in a dangerous area much too long. Because the low skilled people are commonly worse at identifying dangerous situations, they don't begin to worry/panic in situations they should. The higher skilled people will commonly ignore their own feelings in that case, probably because of some kind of innate human group dynamics, where we see those people being call and think they are misjudging the situation.