A cascade of interest greets Taproot’s election matching engine

We are buoyed and blown away by the response to the Taproot Survey, our matching engine to help Edmonton voters discover how well they align with the candidates running for mayor and city council.

As I write this, we’ve received more than 6,800 responses, with more coming in every minute. That’s an order of magnitude greater than the goal we set when we conceived of the idea. A tremendous amount of work has gone into this project over the past year, and it’s gratifying to see it pay off to such a high degree.

Responses have been pouring in since we launched the Taproot Survey, a matching engine for Edmonton voters and candidates running for mayor and city council in the 2021 election. (Mack Male/Flickr)

How did we get here?

The survey is the culmination of the People’s Agenda project that we started more than a year ago. We asked Edmontonians what key issue they wanted the candidates to talk about as they competed for their votes in the 2021 municipal election. Those responses roughly coalesced into eight questions, which then formed the basis of a series of listening sessions to help us further understand people’s concerns.

The next step was to figure out how to determine where the candidates stood on these issues. Taproot co-founder Mack Male came across the Meet Your Mayor app from The City in New York. Because he is a developer himself, Mack said with his customary confidence that we could build something like that for our election. So we set about making it so.

How did we make it so?

Based on the input from our initial question and the subsequent listening sessions, as well as our own knowledge of civic affairs, we came up with a list of questions aimed at sussing out where the candidates stood and how they might be different from each other.

For the matching engine to work, we needed the questions to be multiple-choice. (Maybe machine learning will allow us to analyze unique sentences and match them with voters’ responses by 2025, but that wasn’t an option this year, much to some candidates’ chagrin.) We wanted those choices to stay away from motherhood statements — everyone wants a better city, no? — and we tried to make sure the answers were mutually exclusive and grounded in reality. We wanted the survey to be substantial and wide-ranging but not prohibitively long, so we landed on 10 topics with three questions each. Finally, we wanted to provide a short paragraph of context with links to further information to ensure that both candidates and voters understood the question we were asking.

This took a lot of time. Probably more than our small team could afford. But the prospect of having an easy but meaningful way for voters to find out who agreed with them on the issues spurred us on (along with a little bit of sunk-cost fallacy, if we’re being honest). We also benefited from excellent data-crunching from Madeleine Stout, and sage advice from Prof. Jack Lucas at the University of Calgary; our People’s Agenda steering committee members, Rob Houle and Elise Stolte; and Speaking Municipally co-host Troy Pavlek.

It’s been a long journey, but the train is finally coming into the station on our People’s Agenda project. (Mack Male/Flickr)

Then what happened?

Once the survey was finalized, we had to get the candidates to answer it. To me, this was perhaps the scariest part of this endeavour. Without a significant response rate from them, all of that work would have been for naught.

Step 1 was to make a well-considered, professional product. Step 2 was to make the case for the candidates to make time for our survey amid the plethora they had already received from others, not to mention the many other duties associated with campaigning. Some took more convincing than others, but by the time we were ready to open the survey to voters, we had half of the candidates, and we’re up to three-quarters now, including all but one incumbent.

Meanwhile, Mack and our technical intern, Meenakshi Chaudhary, were building the rest of the engine and website. That was a lot of work, too, but it will pay dividends, not only now but for future elections. We try to build replicable systems at Taproot, and this was no exception.

We soft-launched on Sept. 23, and opened it up to the wider public on Sept. 24, hoping but not knowing it would be well-received. It was. Deepest thanks to everyone who has shared the survey on social media, in newsletters, or by email. The magnitude of the response so far is because of you.

What’s next?

We’re going to keep collecting responses, from both voters and candidates, until election day on Oct. 18. Along the way, we’ll be writing stories arising from the data. Be sure to subscribe to The Pulse so you don’t miss a thing. (P.S. You can have more fun with the data yourself if you want — here’s how.)

Speaking Municipally, our municipal affairs podcast, will also be diving into the survey every Friday until the election.

Mack is planning to feed some of the data into our election night dashboard, which means Taproot will have not only the city’s best visualization of the data from Edmonton Elections (if past years are any indication), but also unique insights that you won’t find anywhere else.

Once the mayor and council are elected, you’ll be able to see at a glance where most or all of the winning candidates stand on these issues, which are likely to come up again over the next four years.

Finally, we’ve learned so much during this election campaign that we can apply to future ones. So watch for us to continue to build on this foundation in the years to come.

One thought on “A cascade of interest greets Taproot’s election matching engine

  1. Great to see the responses! This is proving to be a difficult decision for a lot of voters and it helps to have these resources to work with.


Leave a Reply

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

WordPress.com Logo

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

Facebook photo

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

Connecting to %s