Heh, I thought it was going to be because my blog is back-dated to the 11th century.

Interesting analysis. I am INTJ rather than INTP but who knows what my writing style conveys. I would like to know how they perform this deduction, though. What was their training data?

Gender-wise I am not all that surprised to learn I write like a man. I've worked and studied in male-dominated sciences and industries all my life. My friends have mostly all been men. Also the two characters I identify most with in this story are Egelric and Alred. I am not the girliest of girls. But if the success rate of their guesser is to be believed (currently at only 58% success rate) then it's not exactly a meaningful result.

Reading level Junior High! I like that. If I came out at a college-level I would definitely be writing in the Proustian stratosphere of illegibility. However, it may be that my story rates so low on a reading level because I'm writing fiction and not a technical blog.

Interestingly (or perhaps not) my other defunct and non-fictional blog came up with the exact same results.

These things are probably relatively dumb (in the artificial intelligence sense) classifiers that just look for commonalities between a given text and a corpus of training texts, without really understanding what they're looking at.

Looking at the gender and the personality type ones, in fact, they admit they are using the uClassify service, which is just that.

Mind you, the more & better training data you give such systems, the more proficient they are at classification, so it's a very interesting and potentially useful research area in computer science.

I am sure they don't read the entire blog to do their analysis, so I do wonder what page or pages they read to get their data. All the posts on the first page? The post summaries only?

I just ran the "what reading level am I?" text on the "Aelfden gives an unexpected answer" page directly, and it came out as Elementary School. I think there's something wrong there... maybe it was the comments that lowered the reading level down to that of 8-year-olds!

Anyway, if they're using other blogs as their training data then you would expect results for writing of another genre to be fairly useless.

You might play with oFaust.com too, though it's more annoying since you have to copy & paste a text. But if you do then it tries to compare your writing with other authors. I ran that same chapter and got a 22% similarity to LEWIS CARROLL of all people. Along with the whining message: "There is a slight similarity with the works of Lewis Carroll . I am bored, give me more to read, and please higher quality To which I say, oFaust, oFuck You!

At least OFaust didn't tell you that you write like Sir Arthur Conan Doyle!