[...]
Both sides could have used sockpuppets to influence the poll, however I think that only one side did use sockpuppets to influence the poll. If you were to look at the posts in that thread you would see a large number of very low quality posts that were all against escrowing for yourself, and many/most of those accounts had no trading experience according to their trust history. I don't have specific evidence to address the results of the poll, however I think it is pretty clear that sockpuppets were being used in the discussion, which in turn influenced the poll (assuming sockpuppets were not being used to vote, which is unlikely if they were being used to sway the discussion), and the working of the OP of the thread was written in a way to encourage people to vote the way that Vod wanted. (even people who had escrowed for themselves were arguing against doing so).
There is an obvious contradiction here. You're saying that you can't tell how your customers would have felt about self-escrowing but at the same time you're claiming that the poll was tainted.
I don't really understand your reasoning for why those who voted against self-escrow must have been sockpuppets or influenced by sockpuppets. For example I don't have much of what you could call "trading experience" in my trust history, does that make my opinion on the subject invalid or half-valid and how would that even indicate (or not indicate) sockpuppeting?
Anyway, how about this - can you create a thread that has a properly (in your opinion) worded question about self-escrowing. If you don't trust pseudo-anonymous polling maybe you can establish your own reasonable voting rules. For example - votes must be posted in the thread, and only certain members' votes count. Perhaps only those who registered on Bitcointalk earlier than you? Or some other rule that would reduce the possibility of heavy sockpuppeting.