I was just reading a paper from First Monday about Twitter, and I can’t help but think the authors don’t really understand Twitter[1]. The article is about the difference between followees and friends. Followees are defined as the people you follow, whereas friends are those you interact with at least twice (by their definition).
I think what the author’s missed is the difference between connections and interactions. Most of their discussion is about interactions – the one-on-one kind of exchange, but the flow of information that I witnessed during the Toronto Blackouts doesn’t factor. I’m sure that #darkto wasn’t the first such incident and definitely won’t be the last, but to only count the direct interactions leaves an awful lot out.
For instance:
- Celebrity: They didn’t even mention celebrities on Twitter. There are a HEAP of them and they all interact differently but probably account for a large percentage of the upper users (well, the active celebs at least). I could write a whole paper on this.
- Direct Messages: They mention “direct public posts” by which they mean replies or @messages, but there is nothing about direct ‘private’ messages. That’s a whole nother level of interaction.
And then there’s the problem with their ‘research’:
- Data Collection: Where did they get the data? I know there was a release of data from Twitter which has since been pulled due to privacy concerns and will probably be re-released with specific terms of service, but that was after the date of their initial submission. They do not even mention where it came from, just that they have it. Creepy? Yes. As far as research goes, it’s not research if you don’t mention your methods. Very shoddy indeed.
I’m sure once the twitter dataset is re-released there will be tonnes of ways of analysing twitter interactions, but this paper is barely worthy of being published with just the holes I’ve mentioned from a single reading of it and the superficiality of the analysis.
- not that I have a full and complete understanding of it [↩]