We want to experiment with a system for polling philosophers' positions on key philosophical questions. The aim is to create a repository of individuals' positions which will be sociologically informative, not to determine the truth by vote. Here is how the system might work, in outline. A discussion of tricky issues follows.

The Basics

Claims would be defined and attached to categories. When possible, the claims would be labeled with their canonical names. For example, one could define the following claim:

Representationalism: every phenomenal state is a state of phenomenally representing a certain proposition.

Users will be able to specify their positions by endorsing, rejecting, or being explicitly non-committal about a claim.

Most positions in philosophy are very vaguely defined, and much of the theoretical work there is to be done consists in distinguishing variants on a position. As a result, there has to be a system for precisifying existing claims. For example, one could define a precisification of Representationalism like this one:

Virtualism: there is a nonfactive relation R such that, for every phenomenal state s there is a proposition P such that s = standing in R to P.

For the purposes of this system, we can define the notion of precisification like this:

claim P is a precisification of claim Q iff: a) P is true in all cases in which Q is determinately true; b) P is false in all cases in which Q is determinately false; c) there are cases in which P is determinately true (or false) and Q is not.

There will be disagreements as to what is a precisification of what. Therefore, every introduction of a precisification would introduce two claims, e.g.:

Claim 1: Virtualism is a precisification of Representationalism

Claim 2: Virtualism (as defined above)

I envision a system for navigating claims starting with basic claims which were not introduced through precisification (organized by area or finer topics). Then one could navigate through precisification links and vote on what is true or not, and what is a precisification of what.

Tricky Questions

Who can contribute claims?

Initially, I think only category editors should be able to introduce claims. All claims will also have to be vetted by higher-up editors. This will help ensure that labels are not misappropriated and the system does not become populated with claims which are uninteresting for some reason or other.

Should there be explicit definitions of non-claim terms?

It might be a good idea to have explicit definition of expressions (as opposed to claims). This would allow some 'productivity' in the precisification system. For example, suppose that a precisification of of 'content' is introduced. Then every claim which contains this term could automatically acquire a new precisification corresponding to this precisification of 'content'. An obstacle here is matching all the forms of an expression, but this could probably be managed satisfactorily. The main problem I see is that this could generate many irrelevant precisifications, because different precisification will be relevant in different contexts.

Should the voting be all-or-nothing?

At the moment I'm inclined to think that one's position on a claim should be to endorse, be non-committal, or reject. But we could also have a 'probabilistic' system. The drawback would be complexity for the user. The payoffs may not be significant, because large numbers of individuals making yes/no decisions will in the end result in something like a probabilistic picture. On the other hand, with a yes/no system we might end up getting too many "I dunnos".

Should there be truth value gaps?

I'm inclined to think that we should in fact have four possible positions:

1. Endorse
2. Reject as false
3. Reject as meaningless (there is no fact of the matter)
4. Non-committal

It seems to me that in many cases 3 is the right position, and that it's useful to distinguish between 2 and 3.

How should the results be presented?

The point of this system is not to be normative but informative. So we don't have to decide on the 'right' interpretation of the results. Rather, I think we should make the results available in a variety of ways that people might find illuminating. For example, we could provide raw counts of how many people endorse/rejects/etc what claims. We could also provide weighted sums using a variety of weights, e.g. by how many relevant publications individuals have. We can even weight using impact factors once we have citation counts in our database.

Should individuals' positions be publicly visible?

I'm inclined to think that someone's position should be publicly visible, for transparency. It would also be fun to explore people's positions!

Won't canonical names get abused no matter what?

I agree that a position like representationalism is never going to be defined to everyone's satisfaction. We could allow different claims with the same canonical label, with the equivalent of disambiguating subscripts. However the system will work a lot better if canonical names like this are associated with very vague claims and precisifications are used to tease out various interpretations. This is one thing we can try to enforce with our team of Area Editors through the vetting process.

How about drawing connections between claims?

Yes, I think it could be interesting to allow claims which explicitly relate existing claims, e.g. this supports that. The system will be built with this extension in mind. However I think we should start simple and see how we go.

Precisification or disambiguations?

Instead of precisifications, it might be a good idea to have disambiguations, where P is a disambiguation of Q when P is a reasonable interpretation of Q. On reflection, I'm inclined to think that this is how people will think of precisifications no matter what we say, so this is how they should be treated by the system. There is an advantage to precisifications if people get the system, though: they can enable more automated reasoning once we get there..

Wouldn't it be better to poll authoritative people in a systematic way instead (Leiter-style)?

It would definitely be a good idea to get input from a broader cross-section of the community than the relatively small, potentially biased group of PhilPapers users. We will definitely attempt to poll outside of PhilPapers users in a systematic way as Brian Leiter does for the Gourmet Report (probably by email), and feed the results into the PhilPapers poll. This would kickstart the system. However, I hope that in the long term this won't be necessary, and I think it will also be interesting to have everyone's position.