We’re living in a strange world right now. We’re sort of at a tipping point between broadcast and broadcomment, where we can watch what millions of other people are watching, while at the same time see what they’re saying.
This was brought home to me during the Prime Ministerial debates in the UK. I watched them with my laptop showing tweetclouds, sentiment analysis and Twitter search to get a flavour of what people’s reactions were. So it was a deeply flawed experiment in many ways – take a subset of the population who are interested enough in politics to watch the debates, another subset interested enough to comment online, and find the intersection between them – but it was interesting to see the stats shoot up in favour of Clegg. And ok, so, he didn’t win, but then again no one did. I, on the other hand, did find out a lot about social media monitoring.
Social media in action
So it occured to me then that the best way to demonstrate how social media works is during an event like that. Something that people can relate to what they’re seeing on TV, and reading about in the newspapers. Also – and this is really important – being able to tweet, and then see that tweet – their tweet – appear in the results. It’s what got me into blogging in the first place, when I posted to this blog, subscribed to my own RSS feed in Google Reader, and saw myself pop up a few minutes later. It impressed me. But I think people often don’t quite ‘get’ the idea of cause and effect, that what they blog or tweet about can and will be found by other people.
So this weekend, another event: the six nations Rugby. I don’t play rugby but I do like watching a good game, and this weekend there were plenty (not least because England won). And this time I got quite a few interesting insights using some monitoring solutions that are good for real-time monitoring, namely Twitterfall, Twendz, Tweetfeel and Tweetgrid.
Again, it was a really good occasion to demonstrate how you can set up searches (in this case, for mentions of rugby and 6nations), then tweet something with one of those search terms in it, then see your tweet appear a minute or so later.
Of all the sites I tried, I preferred Twitterfall‘s look and feel. I can imagine it working wonderfully well projected onto a wall during an event, especially in its presentation mode.
But Twitterfall doesn’t really offer any analysis. Even a tweetcloud would be useful and fairly non-controversial, I’d have thought.
Interesting and idiosyncratic Twendz
I found Twendz a little jerky in its presentation, but I did find its analysis tools fascinating. Not least because they’re wrong.
I tested the Twendz sentiment engine a long, long time ago, on the day Jade Goody died (a contestant in Big Brother in the UK). I searched against her name, and saw some tweets come in saying “So sad Jade Goody died” being classed as negative. Presumably this was due to the proximity of ‘sad’ and ‘Jade Goody’ but to my mind, that’s actually in favour of Goody. I asked about this and was told it was a ‘correct response’. Correct in that it’s classing death as negative, but I wouldn’t really be monitoring to find out what people think about death, to be honest.
So this weekend a tweet saying “My two home nations playing but I can’t watch :(” was classed as negative. Again, this is a tweet by someone who I think really wanted to watch rugby, but couldn’t. The proximity of the sad smiley must have classed it as negative. But this is someone who is sad because they can’t be there. It’s a double negative. So it’s in favour of rugby, right? Not against, imho as a human being.
Twendz also has an idiosyncratic way of picking out the main topics people are commenting on. For example, ‘DONT’ came up a few times. When I looked to see what was causing that, it was just two tweets with the word “don’t” in them. Hmmmmm. Maybe this is why Twitterfall steers clear of analysis.
Or maybe it was just having a bad day. Or perhaps it needed the right kind of event to work properly, much like the people of Summerisle needed the right kind of adult.
Tweetfeel, on the other hand, really goes for sentiment analysis in a big way. It even has a big strapline on the home page saying so: “Real-time Twitter search with feelings using insanely complex sentiment analysis.”
And it did seem to work. I was surprised at how well it would correctly classify tweets. It could be that it only classifies tweets that are definitely one way or the other (eg “France are brilliant” or “Scotland suck” – sorry Scotland, but you did, a bit) so I’d have to look into that more closely to compare an unfiltered search with a sentimented search.
Still, it gave me confidence, so perhaps Tweetfeel is good for the sentiment analysis side of things. You could maybe run Twitterfall on one screen and Tweetfeel on another, or maybe even bring them together into a Netvibes dashboard.
Finally, Tweetgrid sounds great and does a decent enough job of presenting tweets in, as the name suggests, a grid, but I found it difficult to get started until I realised the big brown pictures in front of me were clickable icons, and then, well, its presentation leaves a lot to be desired. Given the choice, I’d go for Twitterfall or Twendz any time.
Cause and effect
So there you have it. If you really want to show someone how social media works then speed things up, so that they can see cause and effect, and give them a context. In other words, do it during a live, national event, and show them how this all works with some monitoring tools. They’ll get to see what’s good and what’s not so good about monitoring and social media today.
I’ve already mentioned Netvibes, and I’ve considered building dashboards before big events that bring together live video streams with social media feedback shown alongside. Maybe I’ll do that next time around.
I await the day when broadcasters realise they can integrate this stuff too. I did find it very interesting seeing what people were thinking, but frustrating that I had to look at the laptop, then at the game, and found I couldn’t really concentrate on either.
So maybe, one day, someone will have the bright idea of running a Twitterfall-like column alongside the picture, or running below it like a newswire, together with a cloud. If so, I’ve got copyright on that one.