Mayor Rob Ford, media kerfuffles and online sentiment

Toronto’s controversial Mayor Rob Ford continues to get himself into all kinds of bizarre tangles with the media. He rarely gives interviews, preferring the unflitered medium of his own talk radio show to get his messages across.

The latest kerfuffle involved a Toronto Star reporter apparently being chased off city land adjoining the mayor’s home by a flip-flop clad mayor. There hasn’t been too much level-headed reporting around this one. Here’s the Globe’s coverage of the incident, for those of you outside the Toronto bubble.

Frankly, I think the Star and the Mayor need to just get over themselves and get on with their jobs.

But just for the fun of it, I thought I’d run a quick test using Sysomos MAP, our social media research weapon of choice, to see how incidents like this can rapidly shift the tone of online conversation.

I asked Sysomos to look at all online discussion in Canada about Mayor Ford – across news sites, blogs, forums, and Twitter – in the four-day period from Monday to Thursday of last week. Then I ran the same search for the first four days of this week.

The shift in the tone of conversation is absolutely startling. Shows you how fast reputations can ebb and flow online these days.

In short – before the kerfuffle, a third of all online discussion about Mayor Ford was positive and 19% was negative. Looking at the results for this week, including the last 24 hours of online controversy, there’s been a big shift: now only 17% of all conversations are positive and fully 37% are negative.

The biggest shift has been in online discussion forums – from 46% positive to 60% negative in the space of one short week.

Here, without further commentary, are the results – click the image to embiggenify:

I don’t think that really needs much explanation.

4 thoughts on “Mayor Rob Ford, media kerfuffles and online sentiment

  1. I think it’s probably more a reflection of the fact that Sysomos isn’t terribly good at assessing the tone in tweets. In their defence, this is something none of the social media research services are good at. It’s just really bloody hard for even the best analysis engine to glean meaning reliably from 140 characters. That’s something best left to humans.

  2. Hey Michael,

    Nice post! And glad to see that MAP is your weapon of choice.

    I can speak to yours and Sean’s question regarding Twitter sentiment. Because of the short nature of tweets (140 characters or less) it becomes a bit hard for our system to make a real evaluation. Our system looks at both specific words and the context of those words. Both engines need to agree before assigning a sentiment value to a post. Because context is sometimes hard to fully evaluate in such a small proportion, both engines don’t always agree and we see a lot of neutral in the Twitter space. We’re very aware of this and working on ways to fix this, but to do it properly.
    On a side note, just being on Twitter and from Toronto, I can tell you that the amount of positive tweets I saw about this (both regarding the mayor and The Star) were very far and few between.

    Sheldon, community manager for Sysomos

  3. Yup, that’s what I figured. I’m not sure any algorithm will ever be able to reliably assess real human tone and intent from a scant 140 (or fewer) characters. I shall now bookmark this comment and make a note to check back in 10 years when our robot overlords have proven me wrong.

Leave a Reply

Your email address will not be published. Required fields are marked *