crobb305 - 12:53 am on Mar 23, 2011 (gmt 0)
If you actually read it, it says spelling/grammar/typos and quality are not well correlated. In other words, many high-quality pages, as judged by humans, have errors.
Yeah, I actually read it... I have it sitting right here in front of me on my desk. And like you said, that correlation is based on users’ credibility assessments. The findings do not eliminate the possibility that spelling may count in an algorithm. A human can make subjective judgments of quality/credibility based on experience. An algorithm has to be programmed to look for signals. That's what this discussion is all about -- what might those signals be?
Let's face it, very smart people make spelling errors. I had my fair share of typos when I was writing my Master's thesis and when I submitted my first couple of papers for publication; but, they didn't mitigate the overall value of my work. Peers identified the errors, suggested changes, and I simply made the corrections. We don't have that luxury when it comes to search engines. I'm not sure an algorithm would have held my works in high regard if they were riddled with spelling errors.
Again, this is all supposition, but every idea/clue to what could be an objective "quality" indicator may help us improve our sites. Like you said, it's wise to fix any and all errors. I found a significant spelling error in an H3 tag (which carries weight) and I corrected it. I then moved on to another page.