I have checked a few popular LW posts using the online Readability Calculator and they all came up in the 60-70 range, meaning “easily understandable by 13- to 15-year-old students”. This seems like an exaggeration, but still a vast improvement over the score of 23 for your post (“best understood by university graduates”).
I wonder if the LW post editor could use a button “Estimate Readability”.
Using a different calculator I found that the ten highest scoring articles on LessWrong averaged a score of 37, range 27-46. That suggests that there’s a fair bit of variance between scoring methods, but if we could find a consistent method, a “Estimate Readability’ button in the post editor could be interesting.
That seems to illustrate a potential shortcoming of the Readability Estimator, though. The Simple Truth doesn’t use as much sophisticated vocabulary as many posts on Less Wrong (it seems that posts are penalized heavily for multisyllabic words) but it is a fair bit harder to understand then to read.
I didn’t really get it (if by ‘get it’ you mean ‘see why Eliezer wrote it, and what questions it was intended to answer’) until I’d read most of the rest of the site.
In short, it seems like a decent measure of writing clarity, but it’s not a measure of inferential distance at all.
In short, it seems like a decent measure of writing clarity, but it’s not a measure of inferential distance at all.
Very true. The reason I picked The Simple Truth for an example is that I thought it did a good job of explaining a hard idea in simple language. The idea was still hard to get, but the writing made it much easier than it could have been.
Interesting. I wonder if there’s a relatively easy way to derive the score of the average LW article.
I have checked a few popular LW posts using the online Readability Calculator and they all came up in the 60-70 range, meaning “easily understandable by 13- to 15-year-old students”. This seems like an exaggeration, but still a vast improvement over the score of 23 for your post (“best understood by university graduates”).
I wonder if the LW post editor could use a button “Estimate Readability”.
Using a different calculator I found that the ten highest scoring articles on LessWrong averaged a score of 37, range 27-46. That suggests that there’s a fair bit of variance between scoring methods, but if we could find a consistent method, a “Estimate Readability’ button in the post editor could be interesting.
I’m contemplating using some wget trickery to get a larger sampling-size.
Don’t contemplate, just do it!
I second (third?) the suggestion of a readability estimator; I need it. I have a tendency toward excessively long sentences.
Another comparison: The Simple Truth Flesch Reading Ease of 69.51, and supposedly needs only 8.51 years of education to read.
That seems to illustrate a potential shortcoming of the Readability Estimator, though. The Simple Truth doesn’t use as much sophisticated vocabulary as many posts on Less Wrong (it seems that posts are penalized heavily for multisyllabic words) but it is a fair bit harder to understand then to read.
I didn’t really get it (if by ‘get it’ you mean ‘see why Eliezer wrote it, and what questions it was intended to answer’) until I’d read most of the rest of the site.
In short, it seems like a decent measure of writing clarity, but it’s not a measure of inferential distance at all.
Very true. The reason I picked The Simple Truth for an example is that I thought it did a good job of explaining a hard idea in simple language. The idea was still hard to get, but the writing made it much easier than it could have been.
Yeah, polysyllabicity gets a bad rap ’round some parts.