Can AI turn off toxic comments?

For an industry that’s putting all its eggs in the basket of ‘engagement’ and developing communities, it seems like a self-defeating strategy. Why take away one of the oldest and most direct ways in which an audience can communicate with journalists and each other? After all, many print titles still include a letters section for just that purpose.

The answer, ultimately, is that the balance of cost vs. return for maintaining a comments section is off. If you’ve ever been online you’ll know they can often degenerate into toxic argument in seconds flat, so comments sections require curation.

Often the cost of having someone on staff to do just that gets weighed against the revenue-generation potential of the section, and the decision is made that it’s simply not worth it.

Another reason, the one that is more palatable to audiences than ‘we don’t want to pay to wade through your arguments’, is that the rise of social media means that discussions take place on platforms like Facebook, Twitter and LinkedIn now, so old-school comments sites are redundant.

In a post a year after NPR closed its comments section, its interim managing editor for digital news Sara Kehaulani said that comments on its Facebook page had increased year-on-year, and that there are no plans to reintroduce comments on the site as a result.

Despite that, there are a few publishers who are doubling down on the original idea of fostering community through comments. The New York Times, for instance, is expanding the number of stories on which comments are active, and the Washington Post is rolling out a new commenting system called Talk.

Both projects are counting on Artificial Intelligence having progressed to the point that comment moderation can be automated, with Talk in particular being heralded as a bulwark against toxicity. Emilio Garcia-Ruiz, managing editor at the Post, says Talk gives publishers ‘the technical capability to connect with commenters in a deeper, more meaningful way at scale’.

Developed by The Coral Project in collaboration with both the Post and the NYT, Talk (in theory) removes the cost and toxicity barriers that have led so many publishers to close their own comments sections.

Because it also – in theory – flags up interesting and on-topic discussions among the commenters, it also appeals to publishers’ desire for more user-generated content.

Garcia-Ruiz explains: ‘Many of our most loyal readers are commenters. The combination of Talk and ModBot will allow us get to know them better, more easily interact with them and quickly find and highlight thoughtful and insightful comments for all readers to see.’

And crucially, given that the decision to close comments section is as often as much about financial return as it is a response to toxicity, keeping users on site rather than shunting them onto social to comment will increase those users’ loyalty to that brand.

At a time where publishers are increasingly reliant on a smaller number of highly-engaged readers forming a community around the brand, the revival of the comments section is a vital rung in the ladder back to profitability.

———————————————————–

Just when you thought it was safe to advertise online, there’s a new scandal about how many people Facebook can actually serve ads to. A study by Pivotal Research Group senior analyst Brian Wieser found that Facebook’s Adverts Manager tool promised to reach 60 million 25 to 34 year olds in the US, which sounds great until you realise that’s 15 million more 18 to 24 year olds than were counted in last year’s US census.

Similar discrepancies were found for other age groups and countries, including the UK, raising further questions about whether third party advertising data requires more oversight from industry bodies.