Perhaps one mathematical solution to the problem is to score pages based on the average time spent for each article. My hypothesis is that clickbait articles will score very low because the article will attract a lot of clicks from people who are not interested in the actual content but only the headline.

In theory, you could do this by deploying a tracker across the web similar to a Google analytics that performs the measurement. But in practice, of course, this isn't possible.

You might be able to approximate the score by building a browser extension that makes the measurement. A browser extension might also a good way to output results. It could, for instance, colour-code links (on mouseover) to help you spot click-bait before you click it!