The problem is that this is transparent bias. They may claim to adhere to that, yet it's clear in other cases that they've picked far more obscure Tweets instead of the viral ones.

This is intentional manipulation, they already know that most people read headlines (which are frequently the least accurate part of any story) and ratings fill a similar role.

Worse, they'll later compile the ratings into statistics and spread the misrepresentations further, all while claiming that technically in paragraph 7 they did admit the thing, which is why that's called a motte and bailey strategy.