An on-going point of discussion among colleagues is the appropriate composition of expert panels and whether or not experts with clear and obvious conflicts of interest should be permitted to stand on them.
Consensus might be the best way to choose the Pope, but is it the best way to get an objective answer to a scientific question?
While this is important, it is also important to be clear that ensuring a group of people gives you a balanced opinion goes much deeper than ensuring that the representation of interests on the group is balanced (as one solution goes), or that each member of the group is themselves balanced (as is the other option).
Moves in the European institutions to better police conflicts of interest are only one element of the comprehensive programme of reform needed to ensure opinions offered by expert agencies such as EFSA are as objective as possible.
This is demonstrated in an anecdote recently related to me, about the experience of a member of EU expert GM panels, who is sceptical about the safety of GM technology. Often, panel discussion involves one expert questioning the safety of a proposed technology, and someone in favour of the technology providing reassurances that the safety point at issue is not a real danger.
The sceptic said that he has to be very certain of his ground in order to contradict a claim of safety, because if he speaks out and is proven to be wrong, then his credibility is ruined in front of the group. So in practice, it is safer to note the point in question and look it up later.
However, by the time the claim of safety is proven true or false, discussion of the committee has moved on. If the committee makes decisions by vote or consensus, then the claim of safety has carried, regardless of its truth.
Furthermore, because there is so much more relative pressure on naysayers in the group, then discussion can be dominated by a small number of people who occupy the safe position, as they will be the only ones talking.
What does this mean?
It shows that the truth of a claim is only one, possibly minor, causal element in the overall generation of a conclusion by the group in question.
What is also very important in shaping the conclusion of the group is whether or not a small sub-group of outspoken and socially influential people in the group share a particular viewpoint; this can generate a consensus in the group regardless of the truth of that viewpoint.
This can be exacerbated by social pressure being piled on people who disagree with the group, making the likelihood of the introduction of a counter-view contingent on the charisma and confidence of the sceptic to voice it, as much as on the validity of the counter-view.
These pressures will exist regardless of the composition of the group and the balance of the individuals on it. If our objective is objectivity, and not just getting rid of biased panellists, then we need to go further than the composition of review groups, and into techniques which elicit expert opinion as objectively as possible (such as the Delphi method, for example).