I think we tried pretty hard to get things done quickly this time. We already had two-three pages of analysis before our turn even began. As much as we want to affect voting, we also don't want to vote prematurely. We need to wait for analysis to finish and we need to hear the opinions of strong players who read the analyses.
It's unfortunate that Bxd5 was SO popular to drive-by voters. We had very strong influence on the votes (We went from 21% for Qd8 to 34% for Qd8).
For what it's worth, the voters' choices (i.e., drive-by and others, including me) have coincided with the recommendations of IM Cheap remarkably consistently -- whether intentionally or not -- for the past several moves: Nb6, Nc4, N:d5, B:d5. That is very different from any other person's recommendations.
This is a useful datapoint to keep in mind!
It shows that there may be easy ways to promote consent (i.e., listen to the authoritative people and follow), and very difficult ways to promote consent (i.e., analyze to death and recommend something that many may not see the use of).
Though this turn IM Cheap supported our Qd8 
(I guess they voted Bxd5 because he supported that last move)
You're right -- there may be exceptions to the "rule of thumb" I suggested, such as when a direct target (e.g., a piece to capture, a check to give, etc.) is available, which has its own appeal to people who don't spend much time thinking or reading.
I'm glad many people were able to justify Qd8 using logic. Stuff like "Our light bishop is better than our e7 bishop so don't trade it for the knight" is more useful in persuading the masses than "Qd8 leads to us having better counterplay based on the analysis..." I think we may have a problem if we ever try to justify a move based on analysis alone in the main discussion. Then again, I think we can always find logical justifications for a good line.
why don't we invite IM Cheap here? Or have we done so already?
He is already in the group
That would be an interesting stat to track, how often the groups moves what a particular player votes. Definitely would lead to more people just picking the leading move though...
Another principle that occurred to me: the fewer reasonable move options, the sooner a recommendation has to be made. With relatively few strong contenders like we had this turn, it's relatively easy for one move to take the lead. If several moves are possible, the drive by votes are likely to be more spread out...thus this voting block will have a greater chance of tilting the outcome and a very early recommendation, while desirable, isn't quite as critical.
Yes, I agree with other that we should think ahead.
In past game with GM Becerra, I had observed that
(1) Drive-Bys CAN be influenced. Most Drive-Bys glance at the FIRST page before voting. 100~ D-Bs vote within the FIRST hour. Total votes are usually 1000+~. D-Bs in staggered time-zone (as savyma has said) come later - 500~ votes are cast within SIX hours.
(2) Sometimes if an important move does not make it, it sometimes makes it to the next move even though THEN it might not be the best one. So D-Bs have a memory effect - they are influenced by discussions in the PREVIOUS moves
(3) The move postings influence many D-Bs - some D-Bs don't even read the comments, look at the move-postings, decide mainly on the top vote, and vote mechanically.
So our motto should be:
(1) come to the consensus BEFORE it is time to vote (what-if scenario, e.g.: if white plays this, our best move is this - parallel processing)
(2) post our UNIFIED arguments (no fighting please), and keep it on the FIRST page from BEFORE it is time to vote till last 8 hours
(3) reject deep positional moves - they are beyond the understanding of many D-Bs
I have seen an unhealthy trend among ourselves which might lead to our defeat:
(1) Many of us consider only ONE thread (of OUR choice) - we are choosing to ignore other lines which WE do not like. We - ALL of us - MUST consider ALL lines simultaneously
(2) We must be LESS STUBBORN, and willing to accept other's ideas
(3) When we do the informal voting, the forum thread MUST NEVER BE LOCKED by ANY admin - doing so might be a subconscious reflection of one's own prejudices. We must NOT CONCLUDE INFORMAL VOTING IN LESS THAN 8 HOURS
People made excellent points above. Thanks to all!
I fully agree with the specific calls for:
As far as the timing of votes, it appears to be fairly consistent between moves -- I've accumulated results from the past three moves, and the resulting graph was posted at the very first page of our present move #17 (as well as the final page for the past move 16...B:d5). Take a look at it; most of what dsarkar shared as experiences from the previous GM vote chess game (against GM Julio Becerra) remain true, but are somewhat magnified (i.e., tougher to deal with) now.
As we have seen this turn (move 16), if we don't have something fast, we may not be of much use in the voting. I would propose that after our turn is complete, we need to rather quickly begin to assess what to do next based on white's most likely moves. For instance, if Bxd5 is selected for our 16th move (which is likely, even though we endorsed Qd8), then white only has two responses, exd5 and Qxd5. I propose the following general guidelines for voting so that we have some chance of actually influencing the discussion:
1. In situations in which white's next move is fairly obvious, we need to start our own voting within the last 4-8 hours of white's turn. When it is our turn, we should have an immediate recommendation (within the first hour). If this is done, we are quite likely to sway the discussion in our favor.
2. In situation in which white's next move is less obvious, we should still have some ideas in the works that we develop during white's turn. Depeding on what move white selects, there are two ways this can go:
A. If votes are evenly divided among a bunch of different things (please monitor this everyone, there are plenty of forum posts of votes), we can take our time a bit more, with our voting between 12-16 hours into our turn and final voting with 8 hours of time remaining. We have enough votes to propel one thing to the top if the voting is close.
B. If votes are leaning pretty firmly in one or two directions (and we think at least one of them is inferior), then we need to accelerate voting to between 6-10 hours into our turn with voting to conclude at the 10 hour mark.
I know this is long and somewhat complicated, but we are going to need to be much faster than we were this time in order to have a real effect on the game with the voting moving so quickly.