For those who missed it the first time. Seems pretty harmless to me. 
The metaphor is awkward, but we are all so obsessed with H1N1 these days.. the metaphor of Elizabeth as a virus infected my mind. All activists and writers are inherently viruses - they want to infect the population with a New Idea.  2006 the GPC was a rising party (4% in the polls, up from 1% in the last election.) But the party was immature, and had no antibodies to demagogues. 
An astute political operator saw the potential and ran for leader. She received 2,000 votes compared to her opponent who received 1,000 votes. Small numbers for a major political party. It's easy to take over small political parties. 
Since then, the virus has completely infected the party. There is no paid position in the party - no matter why minor - which has not been been infected. All paid staff owe their employment to Elizabeth. Or, at least their continued employment.. 
It will take a lot of work to rid the Green Party of this virus. The party may not survive the surgery, frankly. But we will try in 2010
 

Your hypothesis about the effect of adding Other to the question doesn't really have any basis.
 
Despite the long term trend stability of EKOS polls, there is a lot of variability from poll to poll... wandering in a month 2 or 3%. So its going to be many months of having 'Other' in the EKOS polls before you can make even tentative conclusions about that category possibly drawing away from the Green poll result.
 
There are also lots of reasons to expect that people who answer 'Other' are quite unrelated to those who answer Green. The fact EKOS started doing this is interesting, but it might have just about any implications. It may only mean that they get more respondents considered valid, and/or that they are mostly people who would call themselevs undecided.
 
There is definitely no validity to you saying that the 2008 pre-election polling of 11% would really have been 9% if 'Other' had been included in the surveys then. If for no other reason [and there are others] because having another possible answer changes the dynamic of responses.
 
Here is another example of conclusions that cannot be drawn from the numbers. Nanos has the same consistency for GPC numbers... at about half the EKOS GPC numbers. And Nanos numbers are far closer to the actual ballot box results.
 
What is most obviously different about Nanos is that he does not prompt with the party name. Not surprisingly, that alone substantially knocks down the number of 'Green' responses.
 
Intuitively that makes sense... since its known that a lot of the people who express a Green preference don't vote. So its likely they also are weakly committed to watching politics at all. So that without prompting they won't think of the GPC.
 
But thats just speculation. What we know is that without prompting of party name, the GPC responses are far lower. and that the Nanos results are the closest to the actual results.
 
Does that make the Nanos results the most accurate? Not really. Their surveys may be depressing the Green responses for entirely different reasons than what causes that general drop in Green polling to Green actual votes.

Shamelessly negative comment:
 
Another way to look at is that the GPC non election campaign polling has been pretty stable since well before the 2008 election. Which means that you have a pretty stable cohort who at least seriously consider voting Green.
 
For 2 years before the 2008 election May was much in the media, especially for the year plus before the election. And you went down from the polling levels to 6.8%. And that was with the attention of the debates and the public sympathy from being nearly excluded from them.
 
Now Elizabeth May is invisible to anyone but really dedicated political junkies. The media has lost interest in her. And she seems disinclined to do anyting about it.
 
Nor will there be anywhere close to the same amount of money for putting into the next national campaign. [?Plus less revenue sharing and transfers to target campaigns?]
 
So how much of a drop from that polling plateau might be expected in the next election? 

http://www.ekospolitics.com/index.php/2008/10/seat-projection-october-6-2008/
But we received 6.8% in the vote, not 11%.
EKOS shows the GP at 11.2% in their last ten polls. So, if a vote were held tomorrow, based on 2008, I predict we would get .. 6.9% :-)  

Polls are useful analytical tools when used this way. You are absolutely correct that contrasting polls which share a common methodology, over time can provide real insights.
Seat projection models are always risky though. Because they look like scientific data, it's easy to lend too much credence to them. Campaigns happen on the ground, Riding by Riding, and oine street at a time. There is no way to credibly capture pertinent data like strength on the ground, the hiring of a strong campaign team, and the impact of the current local issues on the local electorate.
 
These types of media commissioned polls are basically done to provide fodder to daily news providers. The really meaty, and truly significant polling data centers on issues, and the factors that influence peoples voting intentions. Last years CBC commissioned polls are a good example of public polls that can actually help with decision making.

Dave, great work on doing some regression analysis on the polling.  It does make for a better argument than posting a single poll result.  I would argue against adding the other directly to the GPC number as you did in your table 2 unless EKOS specifically stated that their methodology before Jan 14th was to lump the others into the GPC vote.  Clarification from EKOS on that other number would help your analysis.
 
As always, the problems with all polls for the Greens is that we need to be able to deliver on the polling intentions.  Nik Nanos pointed out, at the GPC convention in 2006, that the GPC wasn't very good at converting polling numbers into actual votes.  Hopefully, the last few years since the Nanos observation has meant the party brass has put together a game plan for changing that trend.
 
BTW, which blogs are you referring to that gleefully post negative results?