It is a very visible indicator of the quality of the whole. If the spelling is frequently not correct, which a reader can detect relatively easily, how many more mistakes are hidden in the content, which a reader can not detect easily? Are these completely independent variables? I do not think so. Therefore, I also assess the reliability of an article based on the frequency of careless mistakes.
What is an even larger warning sign, are cliches used to spice up an article. Ars Technica is hardly to blame here, but the Smithsonian magazine is full of it.
My mother[0] was a scientific editor, and she was brutal. She was a stickler for proper English, as well as content accuracy.
She once edited a book I wrote. It was humbling as hell, but it may be the only "perfect" thing that I've ever done (but it did not age well, and has since gone the way of the Dodo).
