I got a lot of pushback from readers regarding my post yesterday supporting Netflix’s switch from a 5-star rating system to a simple thumbs up/down system. The gist of the complaints is that some people do carefully consider their star ratings, and do value the granularity of being able to say that you like/dislike something a little or a lot. But of course some people take that care. The problem is that most people don’t, and collectively, 5-star rating systems are garbage.
This post from YouTube back in 2009 shows it with data: when they had a 5-star rating system, the overwhelmingly most common rating was 5-stars. The next most common was the lowest, 1-star. 2-, 3-, and 4-star ratings were effectively never used.
For a personally curated collection, 5-star ratings can be meaningful. But for a recommendation service that averages ratings among all users, they are not. It’s the difference between designing for the ideal case of how people should behave versus designing for the practical case of how people actually behave.
John Gruber makes a good point about the difference between binary ratings (thumbs up or thumbs down) and granular ratings (stars). Binary ratings make a lot more sense in certain contexts, and with YouTube, it’s a natural fit. You don’t rate a movie on YouTube; you generally rate a cat video, a TED Talk, or something short.
I disagree that this type of rating will work on Netflix. I sometimes look at ratings when I’m browsing Netflix or Amazon Prime Video. If a movie has a 5-star rating, I’m more likely to check it out. If a movie looks interesting but only has, say, a 2- or 3-star rating, I’ll give it a pass.
On YouTube, you see the number of thumbs up and down, and you have to do the math. (I know, it’s not hard.) I think it would be better if Netflix shows a percentage rating based on the thumbs up and down rather than just the totals for each rating; it’s a more logical way to consider movies, since many people are used to Rotten Tomatoes ratings, which are percentages, or IMDB ratings (which are presented as a number out of 10, such as 8.5/10, which is easy to see as a percentage).
There is a corollary with the ratings available in iTunes. The app long had star ratings, which, as John Gruber says, are good for a personally curated collection. But Apple added Love ratings, so you can help Apple Music’s algorithms. Curiously, the company didn’t add the opposite – Dislike ratings – for some time; I think they truly didn’t see the need. You can use Loves and Dislikes in your own library, if that works for you, but that’s not what they’re really for. Having two different rating systems in an iTunes library makes it all very confusing, especially since Apple Music does not use your star ratings when deciding what to present to you.
Source: Daring Fireball: Why YouTube Switched From 5-Star Ratings to Thumbs Up/Down in 2009