How about a metric based on the MVESTS of the accounts (not delegated, but the amount actually owned) that upvoted (by any positive percentage) the option as of the time that they voted? Even with payout declined (I know 2 of the options didn't decline payout because of a bug) I don't think users should have to waste their voting power to vote on a poll. And a MVESTS metric would help the community understand what stakeholders in Steem think about certain issues in a way that actually corresponds to what changes may actually be approved (via witness voting for future hardforks for example). With this metric, however, it would be wise to close the poll within 7 days to avoid abusing poll results by moving Steem Power around to other sockpuppet accounts (for that matter the existing shares metric would also be vulnerable to that if the poll was open for several weeks). Speaking of which, there is no current capability to have the poll end by a specific date (and have the tallies frozen in steempoll.net at that time), is there?
By the way how do you handle the situation where a vote is removed (or downvotes for that matter)? In the case of the MVESTS metric proposed above, you would have to record the MVESTS at the time of voting for each vote so that if they were to later remove their vote you could properly deduct the right amount from the tally.
RE: SteemPoll v2 beta release and improvements poll