Green Party of Canada Voter ID database

Distributed Data Natwork

Distributed Data Network

I read a very encouraging article today on, about a greenparty hosted database called GRIMES. the purpose of this database is to provide access to voters lists and the contact information to any green party volunteer with a computer and internet connections. This is potentially an excellent tool to mobilize volunteers to ID the vote from wherever they happen to be.

According to Darcy Kraus,  GOTV coordinator Guelph 2008, GRIMES is a much quicker and more effective tool then the CIVICrm application that was supposed to perform this task in the past.  As he correctly pointed out, in this age of  VOIP service any number of volunteers located anywhere in the world can be targeted on any strategic riding, or list. I particularly like the feature whereby the database presents one contact at a time, which preserves data security.

Having managed a number of ID-GOTV efforts in the past, I have a few suggestions about what lookup, or search capabilities a hosted database should have. I will apologize in advance because I am not familiar with GRIMES’s back end, so some  of this functionality may already be there. The obvious is search by poll number. A second is search by street name.

Search by street name? This speaks to the need for a sign canvas directed at arterial roads and high visibility locations. Another very powerful fuctionality would be the ability to search for ethnic specific names and letter groups that can then be reviewed and tagged for the appropriate mother tongue canvas. For example any surnames with a cz in them are likely to be Polish. Your’ Polish speaking volunteers can visually scan a list of cz names and tag them by ethnicity and mother tongue for the subsequent Polish language canvas. There should also be a field that captures ethnicity/religion to tag, for example Muslim voters, because it can reasonably be assumed that a targeted message could be affective with different sub sets of the electorate. It is reasonable to assume that anybody with a name that is some variant of mohammed is a muslim. Anybody with ‘polous’, as a character group in their surname is pretty obviously Greek. You get the picture I’m sure. Look at your own ethnic background, and you’ll think of a number of ways to search large database, and find people of the same ethnicity as yourself.

There should also be a drop down field, with pre-formatted descriptions of a handful of key issues that may be relevant to the particular voter being canvassed. The value of this for subsequent follow up with soft supporters, or undecided voters should be obvious. Armed with this knowledge, the Candidate can blast through a list of undecideds, and leaning Green/Lib/CPC voters with the most compelling 30 second ‘pitch’ on this issue and win over a lot of votes. Incidentally, this is an extremely good use of the candidates time. An hour or two by the Candidate on the phone every day, converting soft, and undecideds will be a great use of the data, and will reap definite rewards at the advance, and regular polls.

I mentioned soft supporters, and Green/Liberal/Conservative etc. supporters above. The existing layout of Grimes has ‘Strong Green’, ‘Weak Green’, ‘Undecided’, ‘Weak opposition’, etc. This is not discriminating enough. There should be two drop down lists, encompassing strength of support, and Party of support. The first should be ‘Strong’, ‘Leaning’, ‘Undecided’, while the second should record ‘Green’, ‘Liberal’, ‘Conservative’, ‘NDP’, ‘Other’, and ‘Undecided’. The types of uses for this far more discriminating data are varied. Even if the local campaign doesn’t have the knowledge or skills to work this data effectively, the central Party could oh-so-easily make hay out of it. The local Candidates phone canvas should regularly target the undecideds, and leaning towards, and weak supporters and work on them. They are worth the effort, because the whole campaign has been filtering out these targets from the general population, and it is for this that a personal contact with the candidate is very effective indeed. BTW, the Candidate should never waste her/his time on strong opposition supporters. There are plenty of soft oppo’s to convert, so why get into long arguments with totally unlikely targets?

Nationally, the GPC could take on the task of broadcasting targeted messages to the various supporters of the other Party’s. For example, if the Liberal Party campaign should badly stumble mid-campaign, and take it in the neck over bad communication over a wedge issue, it’s time to leap into action. The central Party could  then fund demon dialling machines, and recorded messages ( at pennies per call), which would specifically address this issue. They could deliver the Green message most likely to resonate with these supporters, and invite them to the website, or to; ‘Press 1 to talk to a Green Part rep.’

There is another potential use of the other Party’s ID’d supporters. I know it stinks. It leaves a bad taste in my mouth, but a well worn tactic for winning the election in a tightly contested riding is to suppress the turnout of the other contender(s). If I were the Campaign manager of a GPC Campaign that was really really close to winning the seat, then I would earn my pay by considering every option. That includes opposition research, in depth, of the other Candidates. Most of the time, the oppo. candidate has gone on the record with an opinion, vote, conviction, or gaffe that really would damage them with their own typical supporter, if only it were publicised. In the runup to EDay, a phone canvas of their own supporters to inform them of this FACT, or even better, a recorded message, that included the oppo. candidate saying the bad thing could be directed at all their identified supporters who would take exception to it. This will suppress their turnout on Eday, and could tip the balance in a tight race. Again, I know it stinks, but ultimately, holding the Candidate to account for their words and deeds is part of the democratic process. Incidentally, this is the purpose of negative advertising in politics. To suppress turnout of the oppo supporters, NOT to convert, or win over.

It is the job of the Central Campaign to do the issue research that will arm the Canvas chair with the most effective message to deliver to each opposition Party’s supporters for each of these purposes. It is the job of the Canvas chair to train, plan, and implement each specific canvas, while the timing of each type of canvas will be jointly co-ordinated by the Campaign Manager, the Canvas Chair, and the Communications Chair. Just remember to manage the timing well, with earned media, paid media, literature drops, and canvas each delivering consistent messages at the same time, that re-inforce each other.

The last point that I will make is that the database being built is a long term asset. GRIMES should take into account that the data will be employed in future Federal, Provincial, and Municipal elections. Each dataset, (election), should duplicate every field for re-recording all the previously tagged data, while presenting data collected from past elections for viewing by the canvasser. Just because somebody voted Green Federally in 2008 doesn’t mean that they will do so in 2009, or Provincially, etc. If an elector displays a pattern of consistent support then they could be targeted in between elections for a canvas recruiting members or volunteers. If somebody voted Green in 5 elections, then an endorsement of a municipal candidate by the Green Party will almost certainly carry a lot of weight with them. If somebody switches votes every time, but consistently voted based on the same issue, then the correct message to deliver becomes very obvious, and they can be converted Green easily, despite the fact that they may never have supported the Greens before.

Did this post sound negative, and overly critical of GRIMES? That is NOT the intention. I am totally enthusiastic about this software. The volunteer developer should be eagerly sought out by the Party, and they should be handed a pile of cash to make the sort of improvements I am calling for here. I can practically read their minds for what they were trying to achieve, and why they put it together like this. They wanted to keep it extremely simple for a novice to log on, and start being very effective with zero training. Kudo’s to them! It worked! Maybe version 2.1 could have a little training presentation, or video at the time of logging on, to ensure that it remains very easily used, and that the broader functionality of the upgrade is properly used by Canvassers.

That’s it for today. I really apologise for such a long delay between posts. My only excuse is that I have been really swamped this month, but I’ll try to post more regularly in the future.

Add to FacebookAdd to DiggAdd to Del.icio.usAdd to StumbleuponAdd to RedditAdd to BlinklistAdd to Ma.gnoliaAdd to TechnoratiAdd to FurlAdd to Newsvine


17 Responses

  1. GRIMEs is great, but there’s one problem. GPC is encouraging use of the central database (well, enforcing it actually) for one simple reason.

    Once a supporter has been entered into GRIMES, they can be turned over to Keys Direct for relentless fundraising. The smarter EDA CEOs have figured out that they should enter these contacts in GRIMES – but tag them as ‘do not contact’.

  2. John, You’re absolutely right. It is a pretty fundamental problem when there is no trust between the host and the users for one of the most important political tools in the GPC toolbox.
    One ‘workaround’ that I could suggest would be to go ahead and let Keys direct go after the ID’d supporters, but tag all the volunteers and members as ‘do not contact’. That will obviously neccessitate maintaining a seperate up to date database of contactables. That way Keys can work the much larger, and softer lists of green voters without filching from the EDA.
    A second suggestion would be for the EDA’s efforts in populating GRIMES with data to be recognized by splitting all the proceeds of GRIMES based fundraising with the local EDA. if and when they exist.
    A third, and the best solution is to hire a small core of outreach professionals, who can implement a well thought out plan of direct contact. The skills, tools, and resources ought to be conceived from the outset as a scalable program, that can be cranked up with an army of volunteers when by-elections, and general elections are in the offing. The rational division of revenues could be maintained, the large reams of data resident in spreadsheets, and homemade databases across the country could be centralised for long term retention, and systematically built into a powerful political force.
    And then they’ll all live happily ever after. THE END.

  3. A Fool’s Errand!

    Identifying GPC support has as much effectiveness as sending out life insurance applications to the Titanic survivors….they just don’t exist and if they do they are almost dead.

    I listen to you guys talk about effective campaigners in the GPC….what effective campaign has ever been run by Greens? If you don’t win you were not effective.

  4. Deano how much campaign experience to you have?

    I have to ask because your last comment doesn’t make much sense. Identified support has a fairly direct relationship with electoral success.

    More people are becoming Green supporters, not less and while the Greens have not won there have been places where we have done well with much smaller budgets than our opponents.

    Effective campaigners help a campaign outperform it’s budget, there are many Green campaigns that have done that.

    • Thanks Dave,
      You saved me the trouble of responding to Deano. What the hell, I’ll respond anyway.
      Deano, I have spent thousands of hours actually campaigning, canvassing, managing etc. during political campaigns. There’s not much point trying to convince me, or the majority of my readers that unfacts are facts. You’re 7% figure is not supported by anything except your own opinion. If you are interested in a samll wager, or even better, if you are interested in a big wager, I would be happy to oblige you. I will give you 2:1 odds, up to $50,000 that if Elizabeth May runs in a By-election in Caseys riding that she exceeds 10% of the ballots cast. Just be warned, that if you should take me up, for a substantial amount, then I will head down there and make it happen, despite any past differences that may exist between myself and Elizabeth.

  5. My memory fails me as it has been 7 months since I got my hands on GRIMES, but I believe most of the search capabilities you mention do exist, and you can print them out in Excel and share with a campaign team.

    You are correct to assert that there needs to be issue tracking as well. When campaigning for Obama, there were about 10 issues with check boxes that could be ticked, and then his central campaign would mail out brochures to those people with a specific concern (eg. gun rights, Iraq, health care).

    In GRIMES there is a notes section where you could make a note like “Liberal sign on lawn” or “Conservative EDA president”. You have to keep in mind that the canvasser has a limited time to try and extract this info and record it, so WO/SO codes are generally fine. Where we are concentrating attention is WG/Undecided anyways, because we don’t have the resources (even in Guelph with 200+ solid volunteers) to start converting the WO/SO. If we do a good job of converting the soft support, the word will get around to the opposition.

    I would be careful about searching ethnicities. I’m 4th generation canuck, but have a very German last name. I would prefer that nobody made assumptions about what language I spoke or what issues I identified with.

    Sharing the data collected can be a risk too, as someone who identifies with the GPC might not identify with their provincial party, but there is no harm in sharing the basic data (name/address/phone) which can be found in the voter’s list or a phone book.

    As far as fundraising, do not assume that the federal office is working against you or unreasonable. While campaigning in Guelph, we learned that there were fundraising calls coming from head office. I asked the GPC head office to cease and desist that morning, and by the afternoon it was halted. Keep in mind that our team in Guelph actually had fundraising moxy, and could bring in donors on their own.

    If I ran things federally (which I don’t) I would ask provincial organizers to discuss individually with each EDA setting target numbers of monthly fundraising calls, and get going on monthly reporting. If the call targets aren’t met monthly, then the GPC should be doing it, via Keys Direct or whomever. This would build campaign funds AND compel EDA’s to use GRIMES before an election campaign starts.

    The campaign in Guelph opened my eyes to the possibilities of a real victory, and GRIMES will be the tool that makes our normally ‘soft’ vote actually show up to the polls.

    Happy organizing,

  6. Thanks for your’ comments Darcy, you did a great job in Guelph. 5,000 ID’d supporters is an impressive number. Hopefully this will double in the next election.
    As far as ethnicities, the local EDA ought to know which ones to go after. I agree that German surnames are not differentiated from the general population. Muslim names however are not. A mother tongue canvas for a group that is cohesive can really lock down an entire slice of the population. With some prep work, like seeking out the endorsements of key community figures, then phone canvassing in the mother tongue can really work. Obviously, your canvassers are bilingual, and most second generations, unilingual english will recognise their parents mother tongue, and not be offended by it.
    Volunteer hours are ALWAYS in short supply, so here’s a suggestion for you. Provided on the initial contact, the data you have recorded is clean, you can outsource specific conversion tasks. I have used a telephone messaging service in the past. It’s a couple of years, but I think their name was protus? You upload your list, then either upload a digitally recorded message file, or record one over the telephone. They will charge you pennies per message drop, which will be played either live, or to an answering machine. This way, you can target lists, large or small, with a tailored message. This can be part of your’ GOTV arsenal as well.
    As far as sharing lists, provided you get back better (meaning more detailed) data than you leant, every election is a gift of fresher, more complete data. No list is perfect, but re-qualifying past Federal Green voters for provincial elections is way better than making a comparable number of calls to the general population. When the data comes back to the Federal campaign, then those who voted both federally and provincially are your’ rock solid core.
    BTW, did you scrutineer the polls in Guelph on eday? If so you have one of the most valuable datasets possible. The lists of people who actually habitually vote. Your’ updated voters list from the advance polls absolutely must be cross referenced against your supporters lists to identify those supporters who actually voted. Ditto for every poll you scrutineered, it is very important that you cross reference the electors who actually voted against the complete electors lists. In future elections, it will help you to target your’ canvas towards people who actually vote, which will squeeze a lot more juice out of your future canvassing results. In the next election, you filter your phone canvas to ID Green supporters from amongst people with a history of voting. That means that the supporters you ID are way more likely to actually vote than the general population.
    Thanks again for visiting my humble blog, and for your’ valuable comments.

  7. We avoided the ‘robo-call’ messaging, but we did try a phone poll – press 1 for candidate x, press 2 for candidate y, etc. It didn’t work well for a variety of reasons, the main one being that we didn’t know exactly who was picking up at the other end. (Homeowner? Babysitter? Child?)

    We did a large scale scrutineer, and we did cross reference to the polls we thought we worked hard and would do well in. Due to the fact that the campaigning in Guelph went for well beyond 60 days, I made an error and those records were mostly destroyed during the office cleanup. Now we know better for next time!

    The key thing in all of this (as was stated in the article) is that we are building towards election day, and then reading the tea leaves for subsequent elections.

    • My personal opinion is that if you’re careful how you couch your’ message, the message drop will on balance have a positive impact. A good approach could be to intro with explanation that telephone messages have no environmental impact, or some such. I believe that even without the explanation, the messages will do far more good than harm.
      I know that not all campaign teams will concur, so you could do a message drop that is strictly for answering machines, and doesn’t deliver ‘live’ messages. There are services that will drop messages directly onto Bell Canada customers message centre without even ringing the phone. These options are a little more expensive. The fact is that there is no other way to reach many thousands of people for hundreds of dollars, and deliver a solid 30 second pitch.
      I have actually run a double blind test of this before, but like you, the election day data was lost before being preserved. I was pretty pissed because we had set up a control group, and had big samples for message drops, drops plus direct phone call, and control group with no GOTV done. We randomized the data across all the polls being scrutineered, and had big samples of ID’d voters. After all the work of scrutineering the polls, the scrutineers data was all discarded, and 2,000 volunteer hours were completely thrown away. It’s a shame, because if it had been done properly, the efficacy of different GOTV methods would have been absolutely proven.

  8. I’m heartened to hear that I’m not the only one who made that error. Let this be a lesson for all who run a large campaign. We could easily have kept the data for a while and done the analysis down the road.

  9. […] Even better, hire a Call Centre Professional to set up in-house capabilities. Polish up the GRIMES voter contact database, as suggested in my previous post. Prepare your’ list of target ridings. You may want to […]

  10. I was working as a database administrator during the 2008 federal election at the GPC. I had a lot of data manipulation work during the election.

    Search & ehnitity
    – Users can be searched according to street name, first name, last name and address
    – There currently no ethnicity tag

    There should also be a drop down field, with pre-formatted descriptions of a handful of key issues that may be relevant to the particular voter being canvassed. ”
    —> Comment field accomplishes this relatively well

    Supporter history
    —-> This would probably be the most difficult point of what you suggest. The dataset isn’t normalized, although I wouldn’t totally normalize it. Keeping the voter data provided by elections Canada and the supporter table separate was a good idea (the EC data is orders of magnitude bigger), but the data set for supporters is further denormalized than that.

    The problem with Grimes is that it hasn’t been open sourced for other volunteers to easily participate in its development. It is using open source technology, but the code itself for Grimes is not open source. It could be debated if the software belongs to the GPC or to its original author (+ possibly Evan Hughes and me, who contributed tiny bits of code, if we wanted to follow the law to the letter).

    There could probably be an independent source-control server set up and managed by the GPC so that the code isn’t totally open source. But you’re adding overhead where none is necessary if you put it on, say, .

    Its too bad the CiviCRM application for canvassing was dropped, because the Drupal license obliges one to publish code. But if Grimes is better and faster, then it should be open sourced. Having it open sourced would have easily help us in the past referendum campaign when the PopVox occasionally failed us.

  11. Cool comment.
    Much of the purpose of my suggested shopping list was to enable data users to quickly and easily abstract data sets for specific tasks. Ethnicity is a simple one, because such a tag allows a list for an ethnic specific canvas.
    The tag for key issues is for the same reason. As a drop down, then you would have consistently formatted data to search on. If the issue was entered free-hand, then it would be next to impossible to abstract an accurate list of electors with a specific policy bent. I’m sure you could see how valuable it might be to quickly pull a list of Liberals who are most interested in a Green tax Shift if their local candidate were to come out publicly saying Green tax shifts were a stupid idea.
    Supporter history would be tough, and I’ll add another even harder one for you. To add in a means to display all unique voters at a given address and phone number. That way the canvasser can ask for everybody by name without dialling the same phone number repeatedly. I would guess this is hard because it’s always taken twice as much time to build this feature as it does to create, and prepare the whole darn database interface when I’ve run past campaigns. As far as open source goes, I’m more interested in outcomes. I guess it comes under the category of ‘many hands make short work’. I’d still say though that this is so important that if it takes serious dollar resources to make it happen, the GPC should spring the cash rather than rely on volunteer efforts. Please understand, I’m not detracting from the efforts, just emphasising how important a good database is.


  13. An homeléss information technology workér whose job was given to an H1 B livés out oƒ his cár 4 10 months beforé finally déciding to trý one of thosé páyday loans and is now able to work from home buying and selling Fx. great to hear a success storý from the realm of poverty


  15. […] Conservatives have CIMS, the Green Party has GRIMES, (and civiCRM), which is but a pale imitation, but directionally sound. The NDP has NDP Votes, with […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: