What is a “Like” worth ? What does it really mean ?
The “like” mechanism was a great innovation when it was launched. It was a simplistic, yet powerful, way for a consumer to mark his (dis)interest. It was a binary version of the 5-star scale that had been used on other platforms before (Group Lens, Movie Lens) and that enabled the first recommendation algorithms to be developed in the 1990’s.
The “thumb up” / “thumb down” principle, although very easy to apply, has also some serious weaknesses. What does a “thumb down” really mean ? Does it mean that you disliked the video itself, its content, one of the person appearing in that video ? This remains very vague. Think about a recommended video of a Donald Trump’s speech on YouTube. You’ll notice that such videos get a lot of thumbs up and down. People vote massively. But for what ? What do these votes mean ? Does it mean that the user liked the recommendation, liked the content of the video, liked Donald Trump or just supports Trump ? This is all very confusing.
The introduction of emoticons in Facebook was a smart move that fixed the meaning of a like. It enabled Facebook to better associate a publication to an emotion in the brain of the user. In a statement first published on Mashable a Facebook’s spokeperson said :
“Over the past year we’ve found that if people leave a Reaction on a post, it is an even stronger signal that they’d want to see that type of post than if they left a Like on the post”
Obviously this stronger signal is used to feed the algorithm that decides what you’ll see or not in your Newsfeed (most people aren’t aware of this algorithm as a study revealed). Interestingly more than half of the reactions shared on facebook are “love” emoticons, which shows how much of the likes were in fact meaning something else.
Conclusion
This use case should lead anyone designing algorithmic systems to reflect on the value of explicit feedbacks implemented. Although I’m a very big proponent of more explicit feedback in online content, I understand that these feedback mechanisms must be carefully chosen to avoid biases. If there is any doubt on how users will use the feedback mechanism, my recommendation is to simply not implement any rather than a bad or biased one.
Posted in Marketing.
