Forum Moderators: Robert Charlton & goodroi
One key development that Matt shared with the audience was that underscores in URLs are now (or at least very soon to be) treated as word separators by Google.
Underscores are now word separators, proclaims Google
[news.com.com...]
Beware of making any changes based on this post, dancing to the tune of the piper may lead you astray.
I don't think anyone is making any changes. The whole purpose of this topic is to let Webmasters know that they "don't need to make the change" after all. But, up until that announcement was made, Google had treated underscores and hyphens differently.
That has been proved on many occasions to simply be a last-minute styling factor applied to the HTML page that is returned, and not in any way a reflection of any internal indexing or ranking factors.
Underscores sometimes become obscured in hyperlinks.
Yes, I've noticed a few of those (requests with %20s where there should be _s) in my logs - not a major problem, but still a problem.
My use of mixed-case is a worse problem here, though the requests for all lower case files all seem to come from automated software, not web browsers. (My hosting service doesn't have mod_speling enable, presumably because of the load.)
And no, I'm not Oprah, thankfully!
The only exception I can think of is certain standard programming language variables and functions. And even then, while the underscore does have significance, it's also a word break; e.g. "HTTP_USER_AGENT".
In fact, I wish they had an interface that DID do this properly, and that allowed us to search actual HTML source code and not just the text that is rendered by the browser. AFAIK no search engine currently supports this, although I seem to recall that Alta Vista used to, back in the olden days.
Can Google's code search be made to search for regexes in .html files?
Results 1 - 10 of about 11,500 for "[+-]?[0-9]*\\.[0-9]*[eE]?[+-]?[0-9]*". (0.11 seconds)
hXXp://www.google.com/search?hl=en&safe=off
&q=%22%5B%2B-%5D%3F%5B0-9%5D*%5C%5C.%5B0-9%5D*%5BeE%5D%3F%5B%2B-%5D%3F%5B0-9%5D*%22
&btnG=Search
Domain Names as supported in the Domain Name System must be less than 63 characters in total length, begin and end with a printable character, and can contain only letters, numbers and the hyphen character (the hyphen '-' must be in the middle somewhere). Underscores are not valid.
But is it good having only hyphens as allowed words separator for domain name and underscore as a word separator for remaining part of the url? Inconsistency.
But, fact is, in SEO terms, it really doesn't matter. As one of a couple of hundred factors, it counts for too little to matter in most cases.
This was neatly demonstrated by tedster's trial - he saw no appreciable benefit for either form, even when the underscore was deprecated. That doesn't mean it was secretly working, it means it really doesn't matter. In most cases, at least.
As well as working as well as anything for SEO purposes, the hyphen is still better for avoiding visitor confusion - thus helping accurate write-in URLs.
If you read Stephan Spencer’s write-up, he says that underscores are the same as dashes to Google now, and I didn’t quite say that in the talk. I said that we had someone looking at that now. So I wouldn’t consider it a completely done deal at this point. But note that I also said if you’d already made your site with underscores, it probably wasn’t worth trying to migrate all your urls over to dashes. If you’re starting fresh, I’d still pick dashes.[mattcutts.com...]
< discussion continues here: [webmasterworld.com...] >