A new survey relies on the ability to analyse Twitter usage, and Twitter has begun a drive to make money from its data feeds. Both point towards the use of Twitter data as discoverable information.
I wrote an article last week called Tweeting weights and weighing Tweets which described how I use Twitter both as a source of information and as a means of telling readers about my own articles. Amongst the benefits, I said, was the development of ad hoc communities of interest in which formal introductions and agendas were unnecessary to get a discussion going.
I also referred in passing to the inconvenience caused by the UK’s pointless use of the word “disclosure” in place of the term “discovery” used in the rest of the world. That includes amongst its by-products the need to flag tweets and other web content with both terms – to call Twitter’s own search tools “rudimentary” is over-polite.
A strand of correspondence opened up on Twitter in the impromptu way which Twitter encourages. Ron Friedmann @ronfriedmann of Integreon had talked light-heartedly of a “tweet weighter” which, I suggested, could be used to discriminate between a tweeter’s valuable thoughts on e-discovery and his ruminations on his football team, mistress or illnesses. Craig Carpenter @craigrcarpenter of Recommind popped into the correspondence, and I suggested that Recommind ought to run with the idea. I wanted a cut, I said, for “turning Ron’s stroke of genius into a marketable idea and pitching it to a major player in search”.
All this Friday afternoon banter acquired a new focus over the weekend with a couple of links to articles about Twitter posted or retweeted by The Posse List @PosseList. One of them, from the SemanticHacker Blog, poses in its title the question How informative is Twitter? It reports on a study to “characterize different types of messages that can be found on Twitter” involving the download of 8.9 million tweets from 2.6 million unique users and analysing their content, origin and apparent purpose in very broad terms. The statistics which result suggest that any business or objectively informative function remains a minority purpose. 30% of tweets related to the user’s current status and 27% were private conversations, with only 10% involving links to blog and news articles. This actually says little about the ultimate purpose behind the tweets, in that what appears to be a private conversation or a note about the user’s current status may in fact have a business purpose, whilst links to blog and news articles may be about either business or leisure activities.
The point is not so much the metrics derived from this particular survey as the practical application of the idea that tweets can be analysed like any other document. If one were to apply to these short messages the same degree of intelligently-directed processing power as we can now apply to e-mail and anything else, it is not difficult to predict the development of user tools to extract the relevant from the irrelevant and the useful from the useless. The Posse List’s conclusion that “it seems that Twitter really is full of people talking about themselves” is the correct conclusion from the survey data, but does not really do justice to the potential uncovered by the survey’s methodology.
Another Twitter-related article found for us by The Posse List carries this subject on still further. Called Twitter hiring workers to turn tweets into money, it picks up on recruitment advertisements being placed by Twitter for people to work on the monetisation (ghastly word) of its evident success in attracting users.
It has already made deals with Google and Microsoft, licensing the use of its data feed so that tweets can appear within search results on Internet search engines. If feeds are available to Internet search engines (and to the survey referred to above) then they can be made available for anything else, and that must include the possibility of user-driven filtering of tweets for discovery purposes as well as for general search purposes.
We have already seen FaceBook and instant messaging become key components in e-discovery – not something which many people would have foreseen a couple of years ago. You do not need to be Nostradamus to predict the same for Twitter.
My article about all this seems to have brought in a new recruit to the e-Disclosure Twitter discussions. Jonathan Maas @MaasJonathan, Head of Litigation Technology at DLA Piper UK and of the LiST Group, a keen exponent of new technology in all its forms, popped up over the weekend apparently drawn into the ediscovery Twitter community by my reference to the discovery versus disclosure conflict – I had appealed for support in building a group willing to flag their tweets as relating to #edisclosure as well as #ediscovery. Jonathan Maas is one of the few who was already into electronic disclosure (or electronic discovery as it was then still called here) when I first became part of it in the early 1990s. It is good to have his weight added to the discussions from a viewpoint which, like mine, originates in the UK but has a wider international context in mind.
On that subject, the weekend saw Twitter as the cause of an introduction to a new resource in Singapore devoted to ediscovery. Singapore is a particular interest of mine after my conference there in October, and I will write about this new connection separately.