How Does the CBC Understand its Mandate?

How they define their flagship product: online news

This post is a by-product of a larger project I’m working on relating to the CBC.

Given that the CBC’s online platforms are now, for all intents and purposes, the only company assets Canadians consume, it’s worth exploring exactly what it is that they offer.

The cbc.ca address takes you to a landing page from which everything on the site is accessed. But besides links to sections like sports, TV, and regional services, the page also typically contains a few dozen news headlines. Visitors to the site are, no doubt, meant to think of the stories linked from the home page as that day’s key news. These, in other words, are the corporation’s primary expression of their mandate as a “unifying force” for Canadians.

How do CBC editors envision “key news?” Trying to manually summarize the recurring topics from just a couple of days’ worth of headlines risks misrepresenting the site’s editorial approach. But gathering headlines from historical versions of the page involves a lot of work - and time travel.

Being professionally averse to hard work – and having no access to time travel – I chose a different approach. I dug into the WayBackMachine from the Internet Archive to download one sample homepage edition from each of the past 12 months. I then used AI tools to automate the process of stripping the text of just the headlines from each busy webpage.

Once I’d collected around 550 headlines, my goal was to feed the data to my default generative AI tool, ChatGPT-4o, and have it categorize each headline. Sadly, GPT was having a difficult day and failed to categorize more than 20% or so of the entries. Anthropic’s Claude-3 AI, on the other hand, was in fine form. Here’s the prompt I gave Claude:

The following text contains headlines from the website of a Canadian news provider. I would like you to assess each headline and assign it one or more of these categories: national news, local news, tabloid-style news, gossip, sports, politics, gender identity, fashion, culture, or sex. Could you then create a new CSV file that identifies the categories you've selected for each headline?

I took the results (which you’re free to view here) and graphed how frequently article headlines from each category appeared on the CBC homepage:

Which is fascinating. From the balance of categories, I’d say that hundreds of millions of taxpayer dollars have purchased us something not too different than Britain’s Daily Mail tabloid. But is that what Canadians really need?

Here are just a few representative samples to give you a taste of what’s there: