Soames Represents!

In Soames' paper "The Unity of the Proposition" he takes up the task of figuring out what the '<', the '>' and the ',' mean when we say . He gives what I take to be a sound refutation of Frege and Russel, and then sketches a view of his own. If I understand Frege and Russel right, I believe they are correct about something that Soames denies. I'll start by reiterating the sketch that Soames lays out.
(S) Propositions are structured entities, the consituents of which are individuals and relations. These consituents compose a special complex C (which is, for Soames, the proposition).
(R) C is the consituents' standing in the R relation to each other.
Soames commits himself to (S) on page 16:
"We retain the idea that propositions are structured complexes, the consituents of which are objects and properties"
Two sentences down he commits himself to (R):
"The complex is, in effect, the standing of the consituents in those relations"

Soames thinks that we use the relation R to represent the way in which the constituents are put together. The idea here is subtle, so let's see if we can tease it out a bit. I'd like to say that for this account to work, the consituents of a proposition must actually stand in the relation R to each other.
1) my bed, the inside of relation, and chaos do not stand in relation R to each other (assume for reductio)
2) (1)->(3)
3) my bed, the inside of relation, and chaos are merely represented as standing in the R relation to each other.
4) ~(3)
5) ~(1)
The idea is, if there is a proposition then if Soames is right it somehow involves them being related by R. However if they do not actually stand in relation R, then they must at least be represented as doing so (premise 2). However, I think if R indeed had this representational capacity, we could do away with it altogether and place that representational capacity on whatever it is we take R to represent. For instance, if R represented predication and R didn't actually hold of the consituents of a proposition, we would be representing the consituents as bearing R to each other, which in turn would represent them bearing the predication relation to each other. It would be much easier just to represent them bearing predication to each other and do away with R. Therefore, I take (4) to be true, and I take Soames to imply that the consituents of a proposition actually do bear R to each other.

So we have that is R(my bed)(in)(chaos). We interpret R to be the predication relation, and thus R(my bed)(in)(chaos) represents that there's chaos in my bed.

So assume (according to this picture) we interpret R to be R' (for instance, R' could be the predication relation). That would be a good band name: "The Predication Relation". Anyway, Soames doesn't assert this, but I would presume than this sort of account would take a proposition Rab to be TRUE iff R'ab. This has some unintuitive consequences. For instance, if we weren't around to represent R as being R', then Rab would have no truth value. Thus, if we weren't around, it wouldn't be true that snow is white etc. This is similar to King's account, and while controversial, not everyone would find it bad. It would however have to exploit the true in/true at distinction.
Also, to avoid some more unintuitive results, this R relation would have to hold of the constituents of propositions necessarily. If it did not, and if the proposition that Fa really was just RFa, then the proposition that Fa couldn't be RFa at a world in which ~RFa for the same reasons that we need RFa to be true here for it to be the proposition that Fa. That that world at which ~RFa could be similar enough for us to use language in pretty much the same way. If this is right, it would be odd that we would use a different R to represent R' in that world.
Admittedly, these are sort of weak jabs at Soames' sketch. The more knee-jerk reaction I had was "I don't use R to represent anything, I don't even know what R is!". So I propose an alternative that more suits my intuitions about what's going on:
I use 'There's chaos in my bed' to represent chaos standing in the inside-of relation to my bed. The sentence is true iff the representation is accurate. 'Yesh balagan b'mita sheli' means the same thing in virtue of representing the same thing. It may be objected "what is this representation you speak of. Surely it's not chaos standing in the inside-of relation to your bed, for in fact chaos does not stand in any such relation." My response would be to say that it's surely possible to have a representation. Soames has one, namely Rcib, where R stands for predication. I would simply have the sentence itself represent, and the proposition be the object of representation (a platonic entity I suppose). So 'There's chaos in my bed' would represent which would be chaos being in my bed. It's false since there's no chaos in my bed (at the moment), though one could say that chaos being in my bed exists (though is perhaps uninstantiated, or doesn't obtain, or something).
Since my view is contrary to that of Soames, I'm sure I've gone wrong somewhere.