IN THEORY, the advent of and explosive growth in social media should be a boon to the social sciences of psychology, anthropology and sociology. Every day, billions of social exchanges occur in the digital wild, each tagged with and linked to the kind of data that many social scientists would give their eye teeth for.
In theory, theory and practice are the same. In practice, however, those data have proved dangerous to exploit. When it emerged in June that Facebook had secretly manipulated the emotional tenor of what a small fraction of users saw, outrage ensued. Even though this kind of experimentation is within the bounds laid out by tick-box user agreements, many column inches were devoted to the ethical considerations of subjecting users to such fiddling.
Online, many people simply typed 'derp'. The word is used as a postscript to a stupid action or statement; it is probably a bastardisation, of the kind that the internet tends to produce, of 'duh'. A new academic initiative aims to reclaim the word, at the same time putting social-media research on a more ethical footing: DERP, the Digital Ecologies Research Partnership.
The effort brings together 18 academic fellows and five social-media partners: Imgur, an image and video repository; Reddit and Fark, two community-driven news and discussion sites; StackExchange, a collection of question-and-answer sites; and video-sharing service Twitch ( recently acquired by Amazon for $970m).
Collaborations of this sort are not new; Facebook's folly was just particularly publicised. Some of the most innovative digital research to date has studied the simple process of Reddit users asking for a gift of pizza. Research by Tim Althoff of Stanford University, in California, and colleagues analysed the sentiments involved in 22,000 posts on Reddit's 'Random Acts of Pizza' (its tagline: 'Restoring faith in humanity, one slice at a time').
Their paper ' How to ask for a Favor', shows that pizza-pie philanthropy was correlated to how early in the month the request was made, and how needy an asker appeared to be (rather than merely how desirous). As mundane as these results might seem, they represent the vanguard of social-network science. This analysis of thousands of real people interacting in a real situation-as opposed to a few dozen underpaid undergraduates in a trumped-up psychology-lab scenario-showed that, contrary to psychologists' expectations, Reddit users rewarded neither requests that sounded upbeat nor those from people who seemed similar to themselves.
Mr Althoff, now a DERP fellow, said that sourcing the data for the study took months of negotiation. DERP aims to streamline and make standard studies such as this, becoming a clearing-house for social-network data and a single point of contact for researchers who want access to them. It says it will only back research that respects users' privacy, uses and stores data responsibly, and meets academic ethical standards. All research will have to be published openly, with the partners learning the results at the same time as the public.
The advantage for researchers is clear. But the partner companies will also benefit, according to Tim Hwang, Imgur's head of special initiatives. His small team manages a sprawling ecosystem of content and interaction that he admits they do not fully understand; now his team can propose questions whose answers lurk in their data-subject to DERP's approval. Mr Hwang says he wants to encourage investigation into what he calles the 'dark matter of the internet'-sites such as his own that get a tiny fraction of the attention of the likes of Facebook and LinkedIn, but that are just as promising in terms of social-science insights.
Early projects for DERP are likely to include cross-platform studies-making use of correlated data from more than one social network-on why internet content goes viral, and research into how people trade and gift virtual currencies. The ultimate goal for researchers would be gaining access to heretofore hidden interaction data that would yield new insights. Scientists might like to know, for example, whether study participants had contacted each other previously via a messaging application, and thus presumably were friends. These are the kinds of data that Mr Althoff says cannot be obtained by traditional means of 'scraping' data directly from the public websites. The link that DERP proposes, an in-plain-sight social-science effort that transcends the public but protects what is private, could finally bridge theory and practice.
Entities 0 Name: Reddit Count: 4 1 Name: Althoff Count: 2 2 Name: Tim Hwang Count: 1 3 Name: Stanford University Count: 1 4 Name: Fark Count: 1 5 Name: Digital Ecologies Research Partnership Count: 1 6 Name: Amazon Count: 1 7 Name: Tim Althoff Count: 1 8 Name: California Count: 1 9 Name: LinkedIn Count: 1 10 Name: Imgur Count: 1 11 Name: StackExchange Count: 1 12 Name: Hwang Count: 1 Related Keywords 0 Name: derp Score: 70 1 Name: reddit Score: 31 2 Name: data Score: 31 3 Name: researchers Score: 30 4 Name: users Score: 27 5 Name: althoff Score: 26 6 Name: social-network Score: 20 7 Name: social-science Score: 20 8 Name: imgur Score: 20 9 Name: theory Score: 20 Authors Media Images 0
Post a Comment for "Social net, working"