All this talk about having the best, most authoritative, most professional, most correct and accurate information on your web site - throw it out the window. Okay, maybe I am being a bit sarcastic but stick with me here.
MDN might be higher quality, better information, but that doesn't necessarily mean it's more useful to everyone.
To bring this into the context. He was defending why Google ranks w3schools results so highly, despite the content not being all that accurate. He said, while MDN might be much more accurate, it is often way too over the top for most newbies to understand and thus not as useful.
Ryan adds, "there's a balance between popularity and quality that we try to be very careful with. Ranking isn't entirely one or the other. It doesn't help to give people a better page if they aren't going to click on it anyways."
He then gives an example:
Suppose you search for something like [pinched nerve ibuprofen]. The top two results currently are mayoclinic.com and answers.yahoo.com.
Almost anyone would agree that the mayoclinic result is higher quality. It's written by professional physicians at a world renowned institution. However, getting the answer to your question requires reading a lot of text. You have to be comfortable with words like "Nonsteroidal anti-inflammatory drugs," which a lot of people aren't. Half of people aren't literate enough to read their prescription drug labels: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1831578/
The answer on yahoo answers is provided by "auntcookie84." I have no idea who she is, whether she's qualified to provide this information, or whether the information is correct. However, I have no trouble whatsoever reading what she wrote, regardless of how literate I am.
That's the balance we have to strike. You could imagine that the most accurate and up to date information would be in the midst of a recent academic paper, but ranking that at 1 wouldn't actually help many people. This is likely what's going on between w3schools and MDN. MDN might be higher quality, better information, but that doesn't necessarily mean it's more useful to everyone.
But is that what Google has been feeding us with Panda? Shouldn't we strive to be perfect and detailed, not just partially accurate and useful?
Forum discussion at Hacker News.
Image credit to BigStockPhoto for books