I learned English at school as a second language, and we would always start with the written form, and then learn how to pronounce. So my "hash table" is primarily organized based on the written form, and it would be impossible for me to mix "two" and "too", "they're" or "their" or "there", or "its" and "it's" [2]. I hadn't even realized that "too" and "two" are pronounced the same, before a native speaker pointed it out to me, as my hash table doesn't support that kind of searches.
Also my native language (Finnish) uses a (nearly) phonetic writing, so phonetic misspelling of words is mainly restricted to people with no high school level of education, and who didn't do that well in primary school either.
I do, like the writer of the liked article, get that feeling of sloppiness, when I see those spelling or grammar mistakes in English text, but I don't really know how much signal it would be appropriate for me to infer from them?
I probably should not use as harsh standards as I do with Finnish, since those misspellings seem relatively common in English in the web.
[1] Well, typical of people who learned spoken English before written English, but in the modern world this pretty much coincides with native speakers.
[2] Well, "it's" and "its" is maybe a border case, maybe not totally impossible to mix those two, just very unlikely.