All posts by Wolfgang Zehrt

“Robot journalism” – how to miss opportunities

Increasingly large volumes of data are being created now that software can “unpick” much faster and more effectively than humans. Computers are running all the time and don’t need breaks like humans. A huge amount of content will be automated in the future, at least content of a certain type and in certain sectors.

The majority of “automated content” specialists started with the wrong focus. They started with soccer, football. Why? Because loads of data exist around soccer! “Real Madrid” meets “Ajax Amsterdam”? Tons of data! Information about games, players, ball speed, transfers, statistics? Tons of data! But – do you really want to read a description of an exciting, breathtaking game? No! And here is the difference -“robot journalism” is data analysis and transfer of findings into a human-like language, but it will not take the reader on a fascinating, emotional trip to a great soccer game. Nevertheless, there are so many other options for the use of automated content and media are not seizing the opportunity. The media industry is too focused on utilizing potential savings; while missing the chance to think about possibilities to get a new type of content on board.

A BBC report on the future of news concluded that, within 5 years, 90 per cent of all news will be automated and written by software. I predict that the BBC is right to take into consideration that we will see new kinds of content, covering a wide range of news including minority group issues or the discover of entirely new findings as well as scientific research results. Taking the latest German election as an example, most of all media missed the chance to get more out of the results. Why did they not look at questions such as: How can you compare voting behaviour over a period of 30 years or more in all 400 constituencies, in less than a second? No matter if you use the generated text for direct publishing or as a draft, machine beats human in that case. That’s not the end of the story: We are working on a solution for a US tech-company to take election data as well as many different other indicators to describe political and social developments in Europe.

Open data policy in most of the European countries is a strong driver for robot journalism. As a consequence, the number, quality, and accessibility of sources are less and less a problem. The devil is in the detail. Convincing a German or Italian governmental institution to provide data in the needed format and in an unusually high frequency is hard work. But, once you have convinced your facilitator in the organisation, realising that his data is no longer ending up in a data graveyard but instead gets public attention, you win the game. Millions of interesting data items are floating around from a global perspective and only software is able to extract key findings and put them together as text. Not an emotional story about an exciting football match, but covering myriads of scientific data-driven topics is the input for automated content. Another idea: Why should farmers not get a story in an agricultural magazine about the prediction of harvest of current crops, based on historical comparisons with similar weather conditions, in consideration of up-to-date market prices and the purchasing power of the most important buyer markets?

The smallest target group of robot journalism is one viewer or reader. What an opportunity! The scalability is immense. Once the data model is created you have a continuous news stream. You can not only update stories in seconds, you can generate out of one set of data different unique stories. It is possible to personalise and regionalise content in a unique way: simple market or weather reports as a part of basic supply of information up to high-quality economic reports as a paid service. Or build an unbelievable variety of portals. We will see it soon. Powered by big US technology companies.

Meet me at the World Publsihing Expo in Berlin and don’t miss his talk:
Tuesday, 10 October, 1.45 pm, Content Stage 2 – Hall 22a:
“Robot Journalism – projects, success factors & challenges”

Reuters Report: Deutschlands Nachrichtenkonsum bleibt traditionell

reuters institute
Bei allen Facebook- und Fake-News-Stürmen: in Deutschland bleibt die Nachrichten-Nutzung weitgehend konservativ und radikale Meinungsportale sind ohne Chancen. Das ist das knappe Fazit der aktuellen Studie des renommierten Reuters Institute Drei zentrale Stützpfeiler der bundesrepublikanischen Nachrichtenversorgung machen die Analysten aus: Spiegel Online für die pointierten, politischen Berichte, BILD Online für Entertainment und Breaking News und, ungebrochen in der Zuschauergunst, die ARD Tagesschau für die seriöse und neutrale Informations-Grundversorgung.

Besorgniserregend empfinden die Medienexperten dagegen die Entwicklung bei den Nachrichtenredaktionen anderer (Online)-Medien: vor allem die in Deutschland mit 28% sehr weit verbreiteten Werbeblocker verhinderten nötige Investitionen in Nachrichtenredaktionen und Nachrichten-Redakteure. Die Nachrichtennutzung über soziale Netzwerke ist laut Reuters im Vergleich zu vielen anderen Ländern sehr niedrig und wird in ihrer Bedeutung von den Verlagen selbst eher überschätzt.

Eine Polarisierung innerhalb der Medienlandschaft, wie sie in den USA weltweit am stärksten ist und laut Reuters die Polarisierung innerhalb der Bevölkerung wiederspiegelt, ist in Deutschland nicht feststellbar. Hier sind die inhaltlichen Ausrichtungen nach “eher links – eher rechts” weitaus näher an der politischen Mitte angelehnt, ähnlich sieht es beispielsweise in Portugal oder den Niederlanden aus. In Italien, Spanien und Polen dagegen ist wie in den USA die Entstehung relevanter Medienangebote an den äußersten Rändern des politischen Spektrums feststellbar.