Matt talks about links vs. social data, but I've heard here more interesting subjects:

• links vs. user experience or
  
• links vs. machine learning.
  

which he is not addressing in this video. But it's good to hear anything about "signals".

I think it's more about links and user experience, not so much vs. They work together.

It's a great (2-minute) video, and Matt says something about the web overall that is important for SEOs to hear...

...that most of the web does not behave the way the SEO industry imagines it does... that the percentage of nofollowed links on the entire web is only a single-digit percentage, and "in fact it's a pretty small single-digit percentage".

...that most of the web does not behave the way the SEO industry imagines it does... that the percentage of nofollowed links on the entire web is only a single-digit percentage, and "in fact it's a pretty small single-digit percentage".

Even so, it's unlikely a lot of old pages are using nofollow. I'd be interested in knowing the percentage of new pages in the past year or so that are nofollow. My guess is it's much higher than single-digit. I'm sure the majority of links are still do-follow but nofollow usage has been rising considerably from what I've seen.

Wasn't it his partner that went green with envy when Facebook became popular?

that the percentage of nofollowed links on the entire web is only a single-digit percentage, and "in fact it's a pretty small single-digit percentage".

I find this hard to believe. Just look at all the big websites that automatically add nofollow tags to all external links:
-- Wikipedia and most other wikis
-- facebook and most similar sites (I'm not sure about Google+)
-- Ehow and most other content farm sites
-- Most forums
-- Most article sites

Cutts' statement might have been true 5 years ago, but I strongly doubt that it still is.

that the percentage of nofollowed links on the entire web is only a single-digit percentage, and "in fact it's a pretty small single-digit percentage".

I'm not really sure why he would lie about this? What would he and Google have to gain? Especially when we all know how important links are to their algo. If he wanted people to stop gaming the system, you'd think he'd say it the other way around. On this, I think you have to believe the man.

the percentage of nofollowed links on the entire web is only a single-digit percentage

I guess Matt is reffering to all links (external and internal) and not only external links.

I don't see much difference b/w social signals and links, links are in fact social signals of a website ..... the more social a website is the more links it gets.

I find this hard to believe. Just look at all the big websites

Yah, but that's not "the entire web". The entire web is all those billions (literally) of sites that don't even make the top-1000 cut, let alone the first 10.

I detoured here for a simple and hokey experiment. Went to vanilla google in Safari, searched for "cats" (I wanted something that would produce gazillions of hits), opened the first 10 pages and looked at the code. On three of them, the word "nofollow" doesn't occur at all; on a fourth, there's just one. Can you make the Top Ten with no external links?

For comparison purposes I pulled up the last 10 pages in the Top 1000. OK, technically 10 out of the last 12 or 13, skipping the "This site may harm your computer" * pages. Surprisingly little difference: 4 of 10 again had no "nofollow" at all, even the ones with more links than Safari could count.


* With a total hit count in nine digits, couldn't they find even 1000 non-harmful sites to display?

I don't think that social media should be counted at all. But I don't think that anybody should be allowed to vote if under 30 or on social welfare.

social media? does anybody heard about the gaming of twitter by celebrities with thousands of fake accounts? not to mention facebook....that one person with a small budget can easily create 100000 trillions accounts using the unemployed (due to panda update) former SEO cheap labor of the far east .....

Well, any algorithm can be gamed - and any part of any algorithm (which is really what we're talking about here.) However, algorithms can also be protected by constructing appropriate footprints to detect the gaming.

I don't expect social media gaming to be completely bottled up by next year, but neither does Matt, as this video demonstrates. However, I've been using Twitter for a good while, and they're already much more on top of the false accounts than used to be the case. The progress is very clear as they begin to mature.

I'm 100% convinced there is strong value in social media (as one of many ranking signals) and as time goes on I do see that value increasing. And since links can certainly be gamed too, a mix of links plus social plus user data seems to hold promise as I see it.

A competitor of mine is a PURE rss feed mashup. They do NOTHING but show snippets from a million sources. 38 million pages indexed, only 68,000 visitors a month but still...

What do they have? Incoming links, lots of them, many from people who complain about them stealing content even. The front page "looks" like it might be a reputable site but there is not one word of unique content on any other page, and it doesn't matter.

Links matter.