Thoughts on Telling….

Earlier today, I spent an hour or so Telling outside a poll station on behalf of a friend who’s standing in a local election. Tellers are those party reps (though not strictly so, in my case…) who grab the poll card numbers of folk who’ve voted so that when evening strikes, any folk who promised to vote for you who haven’t turned out can be bundled into the back of a car and driven to the polling station. Or something like that.

The Telling process itself is paper based, scribbling down numbers of voters (or a line to show a vote from an unknown person) as they enter or leave the polling station. (I asked the other Tellers whether the form was to wait until the voter left the polling station, so as not to be seen as attempting to unfairly influence voters on the way into the station. One said he had been originally “trained” to do just that, but the retired Conservative lady tellers seemed to prefer grabbing the number on the way in…)

So I stood there for an hour, writing down numbers on a piece of paper… (I also annotated some of the marks with a simple clock face showing the approximate o’clock of when the vote was cast, just because…; (the Telling sheet instructions advised starting collections from each new hour on a fresh sheet, to give an idea of hourly turnout)).

Just before I went out, I spent all of 2 mins creating a Google form that would capture the voter number information (plus optional info – I’ve no idea what that might be? Maybe definite knowledge of how the voter claimed to be voting or have voted?):

Simple form...

Grabbing the shortcode link for the form meant it was on my phone and available for recording numbers in electronic form… (It also struck me that it should be quite easy to let folk submit numbers to somewhere via a text message; so for anyone with unlimited free texts, there’d be no reason not to send in voter numbers to either a local or national collection point, maybe with sender number reconciled back to a particular ward or polling station?) I didn’t use this “app” in the end, (are there data collection/protection issues with regard to shoving this info into a Google Spreadsheet, for example?), but there was no technical reason why it couldn’t have been used to collect this data in real time.

So what else came to mind? Gaining access to the electoral roll allows (I think) the domicile of voters to be identified by poll number, at least to the street level; which means that it would be possible to generate a crude heat map, in real time, of where votes had been cast from.

Another part of the jigsaw: when canvassing, I believe the theory is that candidates capture the details of voters who might vote their way, so that they can be knocked up (or whatever the phrase is!) if it turns out that by a particular time they don’t appear to have voted (as recorded by the Tellers). Again, this seems to be a paper based activity, although it would be trivial to capture this information, along with exact address or lat/long details, with an app, and feed that info back to a local or national data collection point.

When I mentioned in chatting to the other Tellers that it would be trivial to put tells onto a rough map, the response was: “but why?” My first thought was: well why not, it’s not hard….?! Then followed by the thought: why not “demap” markers? If you have built up a map of possible friendly votes in pre-poll canvassing through a 2 minute to create app, every time one of your voters turns out you could remove their marker from the map (or change the shade of their marker). That way you’d have an at a glance view of where you needed to go encouraging voters out.

So… would the be a Good Thing to do or not? Or would it be a bit creepy? Is capturing this sort of data in this sort of way legal? Does it infringe on areas governed by the Data Protection Act, or electoral regulations?

PS Hmmm… what would be really handy would be a DIFF spredsheet formula that would generate a list of items that appear in one column that donlt appear in another column. If you know of one, or an easy way of achieving this in a spreadsheet, please post a comment below letting me know how ;-)


  1. Stephen Downes

    Interesting. In Canada this is done quite differently. Each political party is allowed a scrutineer at each poll (a polling place, like a school gym or church hall will typically have 5 to 8 polls, and often a single scrutineer will handle all of them, but the number of scrutineers increases as the polling place gets busy).

    Each poll is basically a table with the poll number on it. A poll may be an apartment building, a few city blocks – usually a few hundred voters. The poll clerk, who is hired by Elections Canada, and is neutral, has a list of electors – their names and addresses within the polling place. This list is created ahead of the vote by enumerators who go around knocking on doors and finding out who lives there.

    This voters list is public information. It used to actually be stapled to telephone poles in the poll, but now you have to go ask for it. So each scrutineer also has a copy of the voters list (and, of course, they have master voters lists in the party office, correlated with their own lists of people).

    The scrutineers sit right beside the clerk at the table. When the voter comes to vote, the scrutineer will hand them a signed and numbered ballot, then cross out or check their name, to indicate that they’ve voted (the voter may be asked to produce identification or present their voter’s card they’ve been given by the enumerator).

    (If your name is not on the list, then you need to produce something with your address and your name on it to prove you live in the poll, and swear (or solemnly affirm, if you’re not religious) that you are who you say you are, and your name will be added to the list. The demand for identification is a formality; in fact, a person’s word will be accepted at the poll. It’s rare enough that any attempt to vote twice would be caught.)

    The voter goes to the little cardboard shelter on another table (our entire electoral system is made up of folding-leg tables, cardboard boxes, and paper ballots). He votes and folds up the ballot. Once he returns, the scrutineer checks the number, detaches it from the ballot, puts the now anonymous ballot into the box, and stores the number.

    Meanwhile, the scrutineers then cross out the voter’s name on their lists, indicating that the person has voted. Runners from the political parties come in and out of the polling place to collect the sheets. Back at the party office, more workers call our otherwise round up supporters who haven’t voted yet.

    I’ve taken part in this process half a dozen times or more, as both an inside scrutineer and an outside scrutineer (runner). It is an enormously satisfying process to work in the polling place all day, to see every elector come through the hall and vote, to look at their faces, to see the cross section of a community. Polling places are always happy, uplifting places, even when the parties are bitterly opposed (there’s some back and forth, but the scrutineers from different parties don’t really interact, but they’re not hostile to each other either).

    I suppose this could be automated more – both the voting process and the scrutineer process – but I’m not sure I would want it to be. I like the fact that when you vote you are a person, not a number. I like the fact that you leave a permanent record of your vote in the ballot box. I like the fact that it takes a certain amount of effort to scrutinize an election, but at the same time that you are able to directly observe every part of the process except the actual printing of the ‘x’ on the ballot.

    (Scrutineers remain present in the room during the counting process; for each poll, the poll clerk and the assistant poll clerk open the box and count the ballots right away, in full view of all scrutineers in the room. The number of ballots in the box must match the number of ballot numbers that have been turn off the ballots, which in turn must match the number of voters on the voters list. The tally usually takes up to an hour to produce (it takes longer if a ballot has been dropped on the floor or if a mistake was made with the voter’s list (but everything you need to spot the mistakes is right there)).

    I suppose now the way to do it is to give the outside scrutineer an iPad with web access to the party’s central voter’s list, and have him or her simply check off the names as people vote. That wouldn’t be too difficult to set up, and it would feed right into the list of known supporters in the poll.

    More automation than that would introduce an unwelcome layer of obscurity into the process, and make the system less reliable.

  2. Trish Rosewell


    If you’ll excuse a further off-shoot of your post, not directly addressing your own thoughts:

    In view of the mismatches between demand and capacity at polling stations in the recent election, I’d been thinking that such technology might instead afford the returning officer a means of monitoring voting ‘take-up’ and of ‘recruiting’ or ‘standing-down’ voting machinery on the fly.

    Not quite sure how feasible this would be in practice, though! It would be fiddly and clunky to try to manage the system too economically, but at the very least, it should be possible to see, in the latter hours available for polling, that a large proportion of the expected votes had not been cast.

    Might it be possible to have ’emergency’ polling stations on standby ready to cater for the late rush that caught out so many electoral officials this month? Or to make sure that votes could be cast at another polling station?

    The latter would require different checks and balances from the those used in the current system (although those don’t seem to work infallibly, given the scenario where some MK South voters were given MK North papers, and vice versa). It might represent a half-way house, though, between having electronic voting, which would be mechanically easy but obviously prone to perversion, and the current paper-based system.

    Could an error-checked system be established for validating voting papers from a central database so that votes would be expected to be cast at a particular polling station but could, in extremis, be cast at an alternative location? I recognise that the central validating system might be hackable, but couldn’t local checking of real votes via tellers help to minimise risk?

    Tellers would then become vital to the process, of course, rather than part of the party machinery!

  3. Rob Englebright

    That’s fascinating Tony, I’ve always wondered what the dear little ladies did with their data, and whether I should play into their hands by revealing my secret voter number…

  4. Pingback: Back From the Count… « OUseful.Info, the blog…
  5. Pingback: links for 2010-05-28 « Onlinejournalismtest's Blog