Both of those would indeed be possible mechanisms. But the point is that an efficient team would be able to share its members' private knowledge. After all, they all have the same goal, and they all know that all the information is (by assumption) true.
It's probably true that you'd expect private knowledge not to be 100% shared in the real world. But the effect size is surprisingly large, especially in such a simplified setting. Literally, the teams in the private info treatment could have just said "let's all write what we know down on paper and share it", and they'd have been in the same boat as the other treatment!
Participants were given a disproportionate number of positive statements for projects A and C so each participant may be biased towards a more positive interpretation of the somewhat ambiguous statements (sufficiently ambiguous to cause 20% of individuals to make different judgements without that initial bias).
Even if the information was interpreted consistently by each individual, the benchmark of success is still biased against collaborative decision-making. The possibility that the collective judgment is different, and perhaps better, is precluded by the assumption that the group should reach the same judgements as individuals.
If the group produced more balanced or well-informed judgements due to the distribution of expertise, would that be interpreted here as a failure of information efficiency and "decision-making quality"?
It's probably true that you'd expect private knowledge not to be 100% shared in the real world. But the effect size is surprisingly large, especially in such a simplified setting. Literally, the teams in the private info treatment could have just said "let's all write what we know down on paper and share it", and they'd have been in the same boat as the other treatment!