News automation in the newsroom | The case of the Washington Post

News Automation and Algorithmic Transparency in the Newsroom:
The Case of the Washington Post

News organizations have turned to news automation to innovate specific processes in the newsroom. Despite the many advantages for news organizations in producing innovative news content, the news automation process is opaque and often not easily accessible for news consumers – undermining the core journalistic ethic of transparency.

This case study examines how institutional dynamics shape internal and external algorithmic transparency practices at Washington Post. Their findings show: While engineering teams exert great efforts to make some algorithmic systems transparent, less information is being shared inside the newsroom. This lack of internal algorithmic transparency is a potential pitfall as it could lead to mistakes in the news production and the reporting process in general.

The findings show that the process of balancing the need for innovation while adhering to the core journalistic ethic of transparency is significantly impacted by the institutional dynamics between members of the newsroom and engineering teams at The Post.

The top five findings from the text:

1. Two Types of Algorithmic Transparency: Washington Post practices two kinds of algorithmic transparency – internal and external. Internal transparency involves limited collaboration between the newsroom and the engineering team, often due to trust in the engineering team’s expertise. External transparency includes sharing information publicly through blogposts or open-source-code.
2. Challenges in Internal Communication: The organizational structure at The Post leads to a separation between the engineering teams and the broader newsroom, causing a lack of interaction and potential misunderstanding between the two groups.
3. (Too) Pragmatic Approach in the newsroom: Most members in the newsroom are primarily interested in the outputs or answers provided by algorithmic tools rather than the inner workings. This pragmatic approach can lead to newsroom staff blindly accepting the results of software tools based on trust in the developer rather than understanding how the algorithm works.
4. Strategic Open-Sourcing and Transparency: The Washington Post’s open-sourcing of code and other transparency practices are driven by a culture of open-source principles and the need to build trust with the audience. However, much of the disclosed information is often technical and not easily understandable to average news consumers.
5. Lack of Transparency in certain projects: While some algorithmic tools, such as the election model, are made almost entirely transparent and open-sourced, other tools are not as transparent due to proprietary concerns and the need to maintain competitive advantage. This selective transparency can create a lack on consistency in how information is shared across different projects.

Key Implications for publishers:

Challenges

• Credibility: News automation might undermine the core journalistic ethic of transparency.
• Lack of communication and knowledge: Lack of internal communication between engineers and newsroom might lead to mistakes in the news production. Also, journalists might blindly accept the results of software tools instead of understanding how the algorithm works.

Opportunities

• Support societal trust in journalism by transparently working with sources, explaining methodology and open-sourcing code of data projects
• Share knowledge: There lies a huge opportunity in filling newsrooms with journalists with a rather technological background.