How we found a million style and grammar errors in the English Wikipedia

The slug in the url is how_we_found_600000_grammar_errors. Did they not get enough attention at first? How many style and grammar errors did they really find?

> So we manually looked at 200 of the errors, finding that 29 of the 200 errors were real errors. Projected to the whole Wikipedia (currently at 4.3 million articles), that's about 1.1 million real errors

Projections are fine, but maybe the title should be something like "How we found (probably) a million style and grammar errors in the English Wikipedia".

I hope these are not "errors" like splitting infinitives, using conjunctions at the beginning of sentences, or putting prepositions at the end of sentences. Because those aren't real errors in English grammar.

To all the other commentators, this isn't about wikipedia, this is about a tool that can be used by copyeditors to help them enforce house style and grammar rules.

If you work for a magazine publisher, it's likely you have 2 lines, a web line and a published line. The web line might not go through the copyeditors/production/subeditors before it goes online whereas the published line will likely go through those people/departments. By adding this into the web line, it can help enfore the rules of the publication so that web and magazine both match up nicely.

Here are the rules this system checks: https://github.com/languagetool-org/languagetool/tree/master...

Imagine a pre-commit hook for writing quality test ;)

Here are some other scripts (much more basic) for "automated language tests": https://github.com/ivanistheone/writing_scripts

The false positive rate seems high enough to make this effectively useless. I wonder if they have a strategy for reducing it?

Is this even news? Pick any Wikipedia article at random that has more than one author. It isn't hard to find run on sentences and verb tense mismatches.

It would be more interesting to know how that compares, perhaps per hundred words, with other bodies of text, professionaly edited and not.

And, for an article on finding grammatical errors, there were a surprising number of grammatical errors.

How I found a million style and grammar errors in the English Wikipedia: by browsing it.

So what! Grammatical and style errors are only important when it impacts the meaning. The majority of pages I read that contain poor grammar are still perfectly understandable. I correct many that I see, but you don't have to look far to find more.

I'll take a wikipedia that has these kinds of errors over a time before wikipedia. As a boy going to the library and immersing myself in encyclopedias and world maps I am gobsmacked at how wonderful great resources like wikipedia, freebase and google maps are.