Is journalism better served by algorithms that aggregate data, or by when humans who assimilate data and postulate? How and why?
Data is undoubtedly important and necessary for journalism. All journalism is supported by/based upon/written around data. The investigative journalist who interviews government officials on alleged corruption, the feature writer who spends days observing in a public setting. These journalists are involved in the process of gathering data. All of them rely on data for their journalistic articles. The journalist collects data from various sources, he organises and analyses them and incorporates them in the story that he is telling. In the last few decades and with the rise of post-industrial knowledge-based economies, the relationship between data and journalism is changing. A few third parties have entered and changed the data-journalist relationship – the computer, the Internet, the algorithm. We have also seen the emergence of a new type of journalism, data driven journalism or what Stray defines as “obtaining, reporting on, curating and publishing data in the public interest.” You may ask, is this not what traditional journalism is already doing? Thus, a second qualifier to Stray’s definition, data driven journalism or data journalism deals with open data or data that is freely available online and which can be organised, analyzed and presented with open source tools. One may already be to see the links between data journalism and the third parties I have mentioned. I however would like to focus on just one third party – the algorithm.
Is the place of algorithms in journalism new?
I would like to argue otherwise. I believe that algorithms have long been with us. Correct me if I am wrong but I would argue that the use of algorithms in journalism probably began somewhat indirectly with the tie-ups between the Gallup Organisation and various US newspaper companies in conducting and publishing presidential campaign surveys. With the appearance of personal computers in the 1980s, algorithms were more ‘visible’ as traditional journalism became more and more computer-assisted. What is new or possibly about the algorithm-journalism relationship today is this: there are an increasing number of them available (even to the public), their complexity, the nature of the data these algorithms deal (open source etc), we are very much more reliant on algorithms than before and finally there is the possibility of algorithms taking over the journalist’s role.
So can algorithms write about real sheep?
In a recent computer game, Deus Ex Human Revolution, an AI (ok fine, an algorithm) known as Eliza Cassan reports the news. She doesn’t report what humans tell her to report. In fact, she decides what to report. She creates the news. Sounds too unreal? Well, take a look at this http://www.guardian.co.uk/media/pda/2010/mar/30/digital-media-algorithms-reporting-journalism
So, it seems the future of journalism is Skynet.
Is that bad?
Why would I argue against algorithms in journalism? To a certain extent, its hard to. Algorithms in fact do a very good job of organising, analysing and presenting data. If I want to understand society as a whole, on a macro level, algorithms do the grunt work of aggregating or averaging masses of data – making a complex whole understandable.
But can the uniqueness and individuality of human experience be completely translated into mathematical language? Does knowing the average income, educational levels of the chronically poor in Singapore equate understanding poverty? I think not. Algorithms need to be ‘put in their place’. They do a very good job of giving us macro/society wide information but they are but one source of telling us about human experience. They inform us but in themselves, they cannot give rise to true understanding (the understandings they promote can be very narrow). For instance, they can tell us about the growing number of casualties in Afghanistan. We may feel horror that in the month of May, some 22 young men and women died violently and that the numbers are not projected to dip any time soon. But they cannot tell us what it is like to be a soldier in Afghanistan. They can tell us ‘what’ but not really ‘why’ (Why is it right to fight in Afghanistan? Why do men and women willingly die for their country?). In a sense, flatness (algorithmic averaging of human experience) is necessary but not everything. Hence the need for human writers and micro articles that algorithms cannot produce (on a side note, it sort of confirms the news can never really be objective, some element of subjectivity is definitely necessary).