Wednesday, 5 October 2016

Preparing for the next Independence Referendum: building the software

During the 2014 Independence Referendum campaign, I did a fair bit of canvassing for Radical Independence and some canvassing for Yes Scotland. Both campaigns had similar systems - cards with tick-boxes. The boxes weren't easy to manage if you had a stack of them, and when you got back to base the results had to be manually collated. This wasn't efficient.

I'm told that part of Momentum's success in the recent Labour leadership election was down to a very good phone canvassing app which people could use on their phones and laptops. This allowed volunteers to see who to phone and, equally importantly, who had already been phoned (so people didn't get pestered multiple times).

Of course, Momentum had access to the phone numbers of many Labour members. In the second Independence Referendum, it's very unlikely we will have the phone numbers of voters, but we should have the electoral roll. We could do data lookup against phone book data but in the UK that's not free and I'm not sure how practical it would in fact be. In any case, many younger and poorer voters don't have landlines these days.

So while we should of course use a Momentum-style phone canvassing app where we have access to phone numbers for voters, I believe we also need a door-knocker's canvassing app. Obviously, both apps need to feed data to the same central database.

Dummy map view. For divided households
we'd need 'pie-charted' icons.
What I'm envisaging is something which, when you launch the app, shows you the immediate locality you are in as a map, with households coloured by whether or not they've yet been canvassed. Clicking on a household bring up a page which lists occupants on electoral roll. Against each occupant, there are yes/no buttons (probably labelled with Saltire/Union Jack).

So you talk to the folk on the doorstep, click the appropriate buttons, click 'save', and move onto the next household. If you're out canvassing with a team, the households the other members of the team visit update on your map, so you can see which house you need to go to next.

Many of our target voters live in flats and tenements. So I imagine there that you might click on the building on the map and get a list of all the flats; clicking on the flat gets you the occupants screen as before.

This isn't a complete specification - we need some means of recording doubts, so that we can target mailings on particular issues.

This makes canvassing quicker and easier, and also ensures that all the results of canvassing get back to the master database.
Dummy occupants screen. The
question-mark icon leads to a
per-voter issues screen.

Note that by geomapping electoral registration data we can also identify housing units where no-one is registered to vote, which might help with an electoral registration drive.

Obviously, if we're going to get an app ready for the campaign we need to start now. We should start by investigating existing open source political canvassing apps, to see whether we can adopt or adapt as this may be quicker than building from scratch.

The 'Field the Bern' app from the Bernie Sanders US presidential campaign is quite similar to what I'm imagining - I'm trying to get in touch with the folk who built it to see whether they'll let us use their code (some at least is on GitHub). Something I hadn't thought of which the Field the Bern app does is gamify canvassing - which may help motivate some canvassers.

There's a series of essays on building a canvassing app by a guy called Sam Corcos which looks interesting. His source code is on GitHub. I'll try to evaluate this over the next week. (Update: the code on GitHub isn't complete or documented, and, as far as I can see, doesn't work. That doesn't necessarily mean we can't use it).

But if we're going to get something built and out to volunteers in time to be useful in this campaign, we need to start soon.

Who's with me?




Post a Comment

Creative Commons Licence
The fool on the hill by Simon Brooke is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License