A/B Testing
A/B testing feels jargon-y, but it's actually incredibly simple. Take an interface A and an interface B, and show them to users randomly. Then, assess users' behavior in reaction to each interface given a metric. You have done A/B testing! Businesses typically test users on click-through rates or conversions (money spent) to determine the best interface to suit their goals.
Multiple iterations of this process can be used to isolate specific features of a given interface. This is known as feature flagging, where a feature is only enabled for a subset of users. This allows businesses to test the effectiveness of a feature in isolation, and to roll out the feature to all users if it is successful.
How layout shapes what readers take away
When the same article is shown as full uninterrupted paragraphs, readers often report that it feels thorough or authoritative even before they're even done reading. This is helpful to communicate professionalism, but it also makes it harder to quickly scan for specific pieces of information unless they're highlighted in some way.
Usually if a piece is that authoritative, it's because it's a news article or some other type of piece that is trying to prove its worth. Other characteristics like dense line heights and smaller subheads play into this effect.
In practice, A/B testing this kind of layout means serving two different versions of the same page to randomly assigned users, then comparing how they react to the content. This could be time on the page, how far they scroll, or how much they share it around. Whichever version wins will eventually be rolled out to all users.
Which version fits this article best?
There is no wrong answer. It's all up to preference.