“This is a breaking news story and will be updated” is a familiar message that typically precedes news reports on significant and rapidly changing topics, such as natural disasters, where the facts are still being ascertained and the situation is not well understood. And, while much of the reporting on COVID-19 (the disease caused by the novel coronavirus SARS-CoV-2) falls into the category of breaking news, readers are not always prompted to consider this.
As with other rapidly changing stories such as disasters, there are a number of challenges when reporting on COVID-19. For example, scientific knowledge is rapidly developing, the duration of the pandemic is also unknown, and it is uncertain when things will be stable.
What has always made breaking news difficult for news quality is that conditions change rapidly, and the context of changing conditions is not fully explained. But recent reporting on COVID-19 demonstrates how the challenge of establishing sufficient context for breaking news — by both online ranking and recommendation systems, and by news organizations that publish on the Internet — poses additional issues for news quality in the distribution and consumption of information online.
The Case of WHO’s January 14 Tweet
Consider news articles on the World Health Organization (WHO): WHO communications from January 2020 became newsworthy in March and April after U.S. President Donald Trump announced on April 14 American funding to the organization would be halted, pending a review of whether the WHO was “severely mismanaging and covering up the spread of the coronavirus.”One particular tweet has come under scrutiny. On January 14 at 6:18 AM ET, the WHO tweeted:
Preliminary investigations conducted by the Chinese authorities have found no clear evidence [emphasis added] of human-to-human transmission of the novel #coronavirus (2019-nCoV) identified in #Wuhan, #China.
At issue is WHO’s possible intent and responsibility related to the message, one reading of which seemed to indicate that there was no risk of human-to-human transmission and no risk to global health. Instead, the novel coronavirus first detected in Wuhan, China has become a global pandemic. interpretations of WHO’s seemingly erroneous January 14 tweet range from a mere relay of information from official Chinese medical channels to potential complicity in delaying awareness of the disease’s severity.
These narratives are important when attempting to determine how the United States government did or did not prepare for COVID-19. This WHO January 14 tweet has been noted throughout April in multiple news outlets, including The Guardian, Washington Post, The Atlantic, CNN, Fox News, and an opinion piece by the Wall Street Journal among others; The WHO’s January 14 tweet is currently included in the Wikipedia timeline of the pandemic as well.
Contextualizing a Tweet
The tweet illustrates the challenge of contextualization. Tweets as we know are short (280 characters) and are effective for quick, broad notifications. However, the brevity of tweets makes nuanced communication difficult. Some tweets can seem conversational, and others serve as the official “public address system” of organization such as WHO. By their very nature, tweets aren’t built for longer discussions (though sometimes people try, as in this recent series of tweets by biologist Carl T. Bergstrom).
The brevity of tweets demands contextualization for fair interpretation, so what are ways that one might add context to the WHO January 14 tweet above? Trying to understand the main takeaways from an organization on a complicated topic starts with an examination beyond a single tweet, alongside other tweets and concurrent reporting.
First, Examine Additional Tweets
One way to consider context is to look at other tweets by the WHO on the same day. About 7 hours later on January 14, at 1:50 PM ET, the WHO tweeted about Thailand’s first case of COVID-19 imported from Wuhan. This tweet from later in the day on January 14 announced that “additional investigation on the novel #coronavirus (2019-nCoV) is needed to ascertain […] the presence of human-to-human transmission” in addition to several other then-unknowns about the disease.
Additional investigation on the novel #coronavirus (2019-nC0V) is needed to ascertain:
📂the presence of human-to-human transmission
📂modes of transmission
📂common source of exposure
📂the presence of asymptomatic or mildly symptomatic cases that are undetected pic.twitter.com/ArQFHIDubc
— World Health Organization (WHO) (@WHO) January 14, 2020
Or, Examine Concurrent Reporting in Order to Present More Context
Another way to provide context about newsworthy tweets is to examine other reporting on the WHO from the same time period. On January 14, several outlets published concurrently these articles:
- Reuters, “WHO says new China coronavirus could spread, warns hospitals worldwide” (current, archived Jan 14);
- The Straits Times, “Wuhan virus has limited human-to-human transmission but could spread wider: WHO,” (current, archived Jan 14);
- The Telegraph (UK), “WHO refuses to rule out human-to-human spread in China’s mystery virus outbreak” (current, archived Jan 15).
These stories were not based on WHO tweets but instead from a statement at a January 14 Geneva news briefing by Maria Van Kerkhove, acting head of the WHO’s emerging diseases unit, who said:
From the information that we have it is possible that there is limited human-to-human transmission, potentially among families, but it is very clear right now that we have no sustained human-to-human transmission.
Both contexts show how tricky communication can be. The “no clear evidence” of the WHO’s widely scrutinized 6:18 AM ET tweet might mean “no evidence exists – don’t worry,” or could mean: “no clear evidence…yet.”
Accurate interpretation of a tweet depends on the context provided by the larger body of communication taking place around it.
Further Questions About Gathering and Updating Online Context for Breaking News
Both context examples do not factor significantly in discussions that include the WHO tweet at issue. The tweet from the WHO later in the day on January 14 at 1:50 PM ET has not received much notice compared to the 6:18 AM ET tweet, if going by the meager 37 retweets and 76 likes (as of April 30) is any measure. The concurrent articles (see above) were however discovered through Wikipedia’s timeline.
It’s not that all news articles have to include every bit of context — different WHO communications may make the same point as the two examples discussed in this blog post. Instead, the examples here demonstrate the challenge of a breaking news context, which is how to keep up with past context as the situation continues to evolve, all the while with potentially high stakes.
News reporters, fact checkers, and Wikipedians all know how time consuming it can be to appropriately contextualize an issue (take this the length of this post from Factcheck.org, or from Wikipedia’s history of the timeline here, here, and here). Sufficient time is required to adequately sort through what evidence may exist. It can be intensely demanding to address questions immediately about events that have suddenly captured the public’s attention. Ultimately, the challenge of building a robust timeline is a historical task that includes many considerations.
For the ever-living nature of information published on the Internet, however, there are additional issues. Neither news ranking, recommendation and search systems nor news outlets that publish online how to handle the problem of surfacing context. The newest information and reporting collide with past evidence every day, but does this make context easier? For both news organizations and for algorithms trying to handle COVID-19 information in newsfeeds today, it is unclear how well the challenge of sufficient context and breaking news is being met.
1) What are breaking news reports trying to accomplish, and what different roles do media companies and technology platforms play in surfacing the information?
Should goals around safety, understanding, responsibility, solutions make a difference in what news rises to the top and how?
2) When additional information unfolds about an earlier reported situation:
- Should the format of online stories adapt and how?
- Should the record of evolving context be incorporated into results from news ranking, recommendation, and even search? Is this even possible to do well and how?
3) Finally, as timelines are amended, aspects of past media reporting may be valuable to preserve even if they are no longer up-to-date. These aspects might be related to larger conversations about understanding, safety, responsibility and solutions. How might this affect the presentation of results?
These are some of the questions we’ll ponder this summer, along with other inquiries around algorithmic ranking and recommendation to include opinion, local, and science/health reporting.