The Medium “Claps” Tool Is Fatally Flawed. How It Can & (I Think) Should Be Fixed.
By David Grace (www.DavidGraceAuthor.com)
What The Clap System Was Supposed To Do
Medium’s “Clap” system was intended to be a tool that would allow readers to apply a metric to Medium articles so that they could be ranked, rated and rewarded.
I think that the reasoning was that if readers could grade the articles so that the cream rose to the top then all kinds of other things would be possible:
- Top writers and stories could be promoted;
- Writers could be financially rewarded in proportion to the article’s approval rating;
- Writers would have feedback both for future article topics as well as actual content.
The Current Clap System Is Seriously Flawed
For several reasons I think that the implementation of the clap system does a fatally bad job of achieving these goals, but it can be fixed.
Allowing 50 Claps Is A Fatal Mistake
Suppose one-hundred people open the article: A New Breed Of Cat.
Twenty of them actually read it all the way to the end. Eighteen of the twenty give it zero claps. One is the best friend, relative, or spouse of the author and the other is the cat breeder in question. Each gives the maximum 50 claps for a total of 100.
A different one-hundred people open a satirical article on Politicians. Sixty of them read it all the way to the end. Twenty-five of them give it one clap and five more of them give it five claps. Fifty claps total.
Under the current clap system the cat article is twice as good as the Politicians satire article. Clearly, this is already a broken system.
The first basic (and fatal) flaw in the current claps tool is the idea that 50 claps by one person are equal to 1 clap by 50 people. Do I even need to explain why that is a fundamentally false idea?
Beyond the false rating results allowed (encouraged?) by the current tool, there’s another reason why you don’t want to allow unlimited claps — People are not capable of making that range of fine distinctions.
The human brain is unable to make fine distinctions over a large scale.
In human terms it’s not practically possible for anyone to effectively or accurately rate one commercial pizza as being 100 times, leastwise 1,000 times, better than another commercial pizza. Or even fifty or twenty-five times better. The human brain’s bandwidth is not close to wide enough to be able to make those kinds of meaningful distinctions.
In photographic terms, the human brain doesn’t have that much dynamic range.
No one who likes the cat article is capable of actually, meaningfully, judging that the cat article is 50 times better than some other person who also likes the cat article thinks it is.
The widest possible meaningful scale that most humans can realistically and effectively use to evaluate and score things is about ten or eleven, that is minus 5 to plus 5. But, you don’t want to use an even number range like 10 because you want “Average, OK, Fair” to be exactly in the middle between Terrible and Terrific.
That means you need an odd-number range, for example 0–4 (scale of 5), 0–8 (scale of 9) or 0–10 (scale of 11).
Is Your Rating Scale “OK To Terrific” Or “Terrible To Terrific”?
Medium needs to decide if the rating scale should continue to be from “Just OK to Terrific” or if it should be widened to “Terrible To Terrific.”
Right now the scale is “Just OK to Terrific” meaning that a score of zero claps is, by default, a score of “Just OK, Fair, Average.”
If Medium Wanted The Range To Be From Terrible To Terrific
On the other hand Medium may want to allow the reader to score the article in the range of Terrible to Terrific. If you’re going to allow the reader to express a negative opinion of the article then a scale of 0–4 would be as follows:
- 0 — Terrible
- 1 — Bad
- 2 — Fair, OK, Average
- 3 — Pretty Good
- 4 — Terrific
If The Range Is “Terrible To Terrific” Starting At Zero Claps Would Be A Fundamental Mistake
If your range is from Terrible to Terrific rather than OK to Terrific, then you need to either (1) force the user to give a rating or, (2) in the face of no user action, by default you must rate the article as “average.”
If I read a book or eat a meal or watch a movie or read an article and it’s just OK I’m almost certain to take no action. I will just ignore it and move on.
The default score that every article must start with is “Fair” “Average” or “OK.”
If your range is Terrible to Terrific then unless your scale includes negative numbers, to get accurate results you don’t want to have a default rating of zero. You should never use a clap system where the default number of claps converts to “Terrible” instead of “Fair, Average, OK.”
If you insist on using zero as your default rating and if your scale is Terrible to Terrific then you need to have a scale with negative numbers, e.g. -2 to +2 (scale of 5) or -5 to +5 (scale of 11).
If you are using a scale of five and you don’t use negative numbers then Average is not 0. On a scale of 0–4, Average is 2 and 0 is Terrible.
Don’t Use A Negative Number Scale
You never want to use a negative number scale. Why?
Because most people don’t want to hurt someone’s feelings and they instinctively view assigning a negative number as a slap in the face. People are far more likely to give someone a rating of 1 on a scale of 0–4 than they are to give that same person a rating of -1 on a scale of -2 to +2.
That’s just how people work.
Examples Of Terrible To Terrific Scoring Scales
Here’s what the numbers would mean on a scale of 0–4.
- 0 — Terrible
- 1 — Bad
- 2 — Fair, OK, Average
- 3 — Pretty Good
- 4 — Terrific
Here’s what the numbers would mean on a scale of 0–8.
- 0 — Terrible
- 1 — Very Bad
- 2 — Bad
- 3 — Not Good
- 4 — Fair, OK Average
- 5 — Above Average
- 6 — Pretty Good
- 7 — Very Good
- 8 — Terrific
By default you give every article that’s rated on a 0–4 scale a rating of 2 (Average) and if rated on a 0–8 scale a rating of 4 (Average).
Each of those ratings is exactly in the middle, which is exactly where it should be unless the reader specifically tells you that he/she thinks the article is better or worse than just OK.
Some Example Numbers When Scoring From Terrible To Terrific
If rated on a 0–4 scale, the new-cat article read by twenty people will have a default rating of 40 and the Politicians article read by sixty people will have a default rating of 120. By default, the ratio of the ratings reflects the number of readers for each article.
The new-cat article had two readers that gave it a very high score. On a 0–4 scale those scores each would have been a 4 and the new-cat article’s rating would have jumped to 44.
If 25 of the people who read the Politicians article and gave it a rating of 3-Pretty Good (instead of one clap) and the other five readers who clapped for it give it a rating of 4 (instead of five claps) its rating would be 30 X 2 = 60 + 25 X 3 = 75 + 5 X 4 = 20 for a total of 155 vs. 44, or about 3.5 times the rating for the cat article.
Clearly, rating on a 0–4 scale with the default being 2 is a much more meaningful measurement of the two articles’ relative approval than the current score of 100 claps for the cat article versus 50 claps for the Politician satire article.
Calculating The Terrible To Terrific Score
As indicated in the above examples, give every read article a numeric score equal to the total of all of the individual scores with each article getting a default score of Fair for each read unless explicitly rated differently by the reader.
Scoring Only From OK To Terrific
If you only want to score articles in the range of “Fair, Average, OK” on the bottom to Terrific on the top then you can have a range of 0 to 4 or 0 to 9 with the default score being 0, meaning it was just OK.
In that case a 0–4 scale would be as follows:
- 0 — Fair, OK, Average
- 1 — Above Average
- 2 — Pretty Good
- 3 — Very Good
- 4 — Terrific
The Article Must Be Read To The End
In order to apply the default “Fair, Average, OK” score, Medium must know that the article was read to the end. Since Medium already differentiates between “views” and “reads” I’m assuming that it has some way to tell if an article was fully read.
If so, then only articles that fall into the “reads” category and not the “views” category would get the default “Fair, Average, OK” score. Mere views would get no score at all.
The User Interface
The end of each article would need to have text along the following lines:
If Scoring From Terrible To Terrific
By default, Medium has rated this article as Fair, Average, OK. If you like, you can change your personal rating for this article by clicking on the appropriate rating choice below.
- 0 Terrible — — — — — o
- 1 Bad — — — — — — — o
- 2 Fair, OK, Average — - •
- 3 Pretty Good — — —- o
- 4 Terrific — — — — — - o
If Medium wanted a wider scale it could use the following 0–8 scale:
- 0 Terrible — — — - o
- 1 Very Bad — — — o
- 2 Bad — — — — — o
- 3 Not Good — — — o
- 4 Fair, OK Average- •
- 5 Above Average — o
- 6 Pretty Good — —-o
- 7 Very Good — — — o
- 8 Terrific — — — —- o
If Scoring From OK To Terrific
By default, Medium has rated this article as Fair, Average, OK. If you like, you can change your personal rating for this article by clicking on the appropriate rating choice below.
- 0 Fair, OK Average — - •
- 1 Above Average — —- o
- 2 Pretty Good — — —- o
- 3 Very Good — — — — o
- 4 Terrific — — — — — - o
Summary
A max of 50 claps makes the results unreliable to the point of being almost meaningless.
Medium must decide if it wants to keep the current ratings range of “Fair to Terrific” or change to a ratings range of “Terrible to Terrific.”
Whichever rating range Medium uses, the default score for every article must convert to a rating of “average.”
Limit the scoring range to no more than 0–10 but a narrower range is probably more effective and more meaningful.
— David Grace (www.DavidGraceAuthor.com)
