Statistics
Best Metric to Judge a Medium Article's Popularity
How to quantify the quality of a Medium Article
Judging the quality of a Medium Article is a subjective matter, much like any other artwork in the world - Its value lies in the eyes of the beholder. But we, as programmers, cannot devote ourselves to such kind of thinking, can we? We like logic and structure - And we look for a method to the madness.
But why would "Medium" wants to do this?
It's simple - To improve their platform and enhance their user experience. It's quintessential to be able to quantify (at least to some extent) the quality of an Article.
Quantifying the quality of a piece of content is one of the biggest challenges that a blogging platform, like Medium, faces while coding. It helps improve its recommendation page, distribution system, curation mechanism, and its ranking algorithm.
There are a bunch of criteria, through which Medium can achieve this. Like -
- the number of people who have voted on the article,
- the average number of claps per voter or total number of claps,
- number of comments/responses on the article,
- number of social media shares,
- average reading time,
- a number of backlinks, etc ...
But for us, mere mortal humans, only a few publicly available indicators are there, like the number of voters and the number of claps per voter.
Factors to Consider -
You must have noticed that at the end of each Medium story, there's a clap icon, which tells us how many people have already appreciated that article, and how much, on the scale of 1 to 50.
The medium allows every user to clap almost 50 times, which theoretically signifies how much he/she "liked" that particular article.
The total number of claps on an article (Total clap count) and the total number of people who did clap (Voters count) are 2 excellent criteria that indicate the article's popularity to a certain degree.
But none of the two can be used alone.
Why?
Because Total clap count alone can be misleading since a single person can clap 50 times and people may use that as a loophole to generate false stats.
For example -
Having 500 claps from 500 different people is far better than having 500 claps from 10 people.
And if we consider only the Voter's count, then again, we won't be able to calculate its popularity if the numbers are identical. So we can't ignore the claps as well.
Let's say -
There are 2 articles with 4 voters each. If first one receives 100 claps and other receives 50 claps, then obviously, the first article is of superior quality.
Combining both factors -
We know that taking both criteria - Total clap count & Voters Count - into consideration solves our problem.
But how are we supposed to combine these?
Here is my reasoning ...
Since the psychological resistance to press the clap button is more than the number of times it's getting pressed, we can give more weightage to the voter's count and simply add the total clap count to it.
Resulting in: -
Popularity = 50*(Voters Count) + (Total clap count)
This, slightly-more-advanced metric can give us a better and unbiased idea of how much the article is liked by people.
Conclusion -
Obviously, we should've considered other criteria, as mentioned above in the beginning of the article, but those ain't publicly available. So we've to settle for this now :(
In the future, I'll try to include the engagement factor (number of comments or responses) in this metric as well. But before that, I must figure out how to subtract the loopholes from it.
Till then, I'm gonna work with this. It's a pretty good indicator already!
Hope you learned something from this.
Thanks for reading & have a nice day!
Other Articles that you might enjoy -