Nutshell: periodically, I'll search for my own name in search engines. If you say you've never done this, you're a liar. Everyone has. You've also checked your Images results. I was quite amused a few months ago when I realised that a relatively terrible picture from Pubcon was ranking for my name. Unfortunately, it isn't ranking anymore. I rather enjoyed it.
The strange thing about that picture is not that it's not of me. Barely any of them are. The strange thing is how the picture ended up ranking for my name. I had linked to the image from a blog comment here at SEOmoz and blog comments are all nofollowed. The picture was from Facebook. Even though content is supposedly "behind closed doors" and available only to those people who are authorised to see it, linking to a jpeg file works. Like this.
Nowhere else on the Internet had I linked to the other Facebook image file. As far as I could tell, the file hadn't been linked to with my name (or in the general vicinity of my name) either. I began to see more and more pictures ranking, all of which I'd linked to in comments. A lot of them had been edited in, and "edited by Jane Copland" often showed up underneath the pictures.
Again, SEOmoz blog comments are all nofollowed. Comments are not like profile links, where a higher level of participation will have the nofollow removed.
Most recently, this picture shows up for "edited by jane copland" after someone linked to that picture in a blog post I wrote last week.
There are various other instances of these pictures showing up for my name, presumably because my name appears very close to their links. The images are shown with their original URLs, but the pages that load beneath the result are our blog posts. Let me know what I'm missing here, as I can't work it out for myself.
I thought Google was supposed to treat nofollowed links as though they were not there. When nofollow was born, Search Engine Watch reported that:
Most recently, this picture shows up for "edited by jane copland" after someone linked to that picture in a blog post I wrote last week.
There are various other instances of these pictures showing up for my name, presumably because my name appears very close to their links. The images are shown with their original URLs, but the pages that load beneath the result are our blog posts. Let me know what I'm missing here, as I can't work it out for myself.
I thought Google was supposed to treat nofollowed links as though they were not there. When nofollow was born, Search Engine Watch reported that:
Nofollow effectively will cause Google to ignore the link, to pretend it doesn't exist.In an interview here at SEOmoz from August 2007, Matt Cutts said:
For Google, nofollow'ed links are dropped out of our link graph; we don't even use such links for discovery.Are the rules different for images? If not, why are these pictures being discovered as having relevance to my name? And does that mean I should start linking to flattering pictures, rather than to pictures of scary cats?
If you link to an image Google treats it in the same way as if you have embedded the image instead of linking to it. This has been an exploit for some time.
Whether the link is nofollow or not is irrelevant because google doesn't spider the link or pass PR to it. They just see the link as an embedded image and treat it accordingly.
Nice point you have made Patrick. Thumbsup!
I agree. What happened was probably the image being "read" as embeded into the page and not as link to an outside file that had to be followed to be read.
So the nofollow is inconsequential...
Another example of this happened in the Clueless YouMoz post, and was addressed then as an instance where the Image Search showed its weaknesses.
I figured there must be something about the fact that they're images. This does indeed seem to be the reason. The funny thing is, Ciaran and Danny's images of me probably have more likelihood of ranking for their names instead of mine. Har har har.
I can't cite it right now, but I remember awhile ago Matt Cutts mentioned there used to be a glitch where nofollow was passing anchor text that affected the rankings, but not link juice.
I think that alone says the nofollow system is more complicated than they make it appear. I have trouble believing they made such an elementary mistake if there wasn't a fair amount of complexity to it.
Also, I'll point out that has anyone noticed blog spam hasn't disappeared even though 90%+ of blogs have nofollowed comments? That's because it still gets you indexed quite fast. Although it's hard to track specifically due to scale, I have a tremendous issue believing that the few do-follow links in a blog spam batch are enough to speed indexing the way they appear to. IMO, there's still effect, and at least something is passed. And definitely indexing happens faster.
Pretending the link doesn't exist? Psh.
But anyways, more on track with this, I think if they can botch anchor text passing, they could definitely have missed images.
Here's the link:
https://www.stephanspencer.com/search-engines/matt-cutts-interview
Good job finding that.
God forbid blog spam disappeared. Entertainment sites like Perez Hilton would never receive the 600 or so comments "per picture", that help make them so famous...
First bitches! :)
Couldn't agree more. Everytime I post, I'm riddled with SPAM comments within 30 minutes. I'd really like to believe "nofollow" is the end all, be all. Yet, from the examples above, and my own personal experience with our blog and our site, I think there's something afoot with the "nofollow" link.
I've also suspected that Nofollow wasn't as absolute as G has always said because of sites of mine with almost no link love other than some that are nofollowed ranking better than would seem reasonable if nofolllowed links really don't pass any juice. I think they do. And not just images.
I'm late and this has more or less been said: There are two issues here.
#1) I believe Google follows nofollow. It just doesn't pass any link juice other than a bare minimum in order for the algo to work. They especially follow it for discovery if it is a brand new link or site. And the anchor text does get indexed.
#2) Google Image Search Results are all messed up and they grasp at straws to return results.
I think we can all confirm that each vertical of Google indexes things differently. I'd think what Google Images are doing could be wildly different than what Google Web, Google News, etc. are doing.
Brent D. Payne
I'm not the SEO in our building (we have a quite a few), but from what I understand, "nofollow" is a bit of a misnomer. Spiders still follow "nofollow" links, they just don't count the nofollow link from page A to page B as a vote by page A for page B.
So in this case, Google still found the image first on our blog in the comments, spider ed it and then saved the surrounding markup. When Jane searched for her name, it found a bunch of GIS results for her name from all those blog comments and pulled them up.
It's just weird because Google reps have stated that they don't even follow the links, but rather pretend they're not there. In the case of images, I suppose we've discovered that this isn't true.
Well I've always been told by Rand nofollow != no spider, it just means don't count this link as a "vote." So they'll still go look at the page you've linked to and spider it, just not count the link you wrote as a vote.
GIS doesn't appear to follow that exactly as you seem to have found.
I would normally say that Google's claims, which Jane quoted above, represent to me a definitive answer and one that I'd trust... However, this thread has me re-thinking it.
But, yeah, Jeff - from at least Pubcon December 2006, when I heard Matt talk more about nofollow to now, my understanding was the same as Jane' - for Google at least, nofollow=do not even use for discovery.
I can't provide any empirical data offhand, but I agree wholeheartedlty with David on this - both for images and content. Something's happening there and this is definitely a topic worth investigating / testing.
SOLVED: We don't no follow links on our comment RSS feed. The Google Image bot must have indirect access to that.
https://feeds.feedburner.com/seomoz-comments
I blame Jeff ;-)
And apparently now we do... *regrets blaming Jeff*
Yeah... there is a nofollow on archskrk's link from above.
From the feed:
<description><p>Oh, something is definitely up.</p><p><a href="https://archshrk.com/2008/04/nofollow-my-arse" rel="nofollow">https://archshrk.com/2008/04/nofollow-my-arse</a> </p></description>
I code even while sleeping.
Interesting! Well done! I wonder why Google still shows the original post when you click through from the image, and not the RSS feed? Rand speculated that they might have some canonicalisation system that connects the pages.
Opinions?
Very interesting, though. Does this mean that blog spam can still be worthwhile if the blogs' comments are available with RSS?
And Jane Copland gives birth to a whole new business model...
Come on ciaran you know that Jane Copland is sensitive about her image SERPs. What you did was mean.
Danny - that's just rude, if not verging on racially insensitive. Jane Copland is going to kill you now.
Guys come on grow up! Why would you do that to Jane Copland from SEOmoz?
because every one just loves Jane Copland so much :)
When these things show up for you guys' names, I'm going to laugh.
Both the Black Sheep and 'Kira' images are showing up page 1 in Google Image search for 'Jane Copland' - at least here in the UK.
Guess some posters are in trouble now? ;)
That's it. Ciaran and I are no longer friends, and Danny's life is going to be a living hell once I get to work ;)
You, and your friends above, will be in massive amounts of trouble ;)
Oh,oh...
I think Google takes as much liberty with nofollow as it does with dofollow links, meaning, some nofollow links are worth much more than others. If the link is editorial, peer reviewed, and from a high trust site it's likely pulling power one way or the other.
I'll also throw in my vote on the scraping angle. That sounds pretty plausible, especially for a blog.
I had some links into pages such as "privacy policy" etc. I nofollowed those links a couple months ago. Now I can't find them by searching Google and the pages went from PR5 down to PR0 graybar.
Unlike Sean and David, I haven't noticed anything strange with regular links, just these images.
It's quite obvious that Google doesn't treat these links I've left in comments as though they didn't exist... and although I doubt the links pass PageRank / link strength etc, I can only deduce that these results show that Google does use them from discovery and relevancy.
<sits down and waits for someone from Google to clear this up>
Think about it. Google's scheme of link=vote - the intent of nofollow was to keep advertising and other paid links from equaling a vote, but the way that it's being used most of the time is to control the ebb and flow of link juice. Most nofollowed links on SEOmoz for example really should equal a vote. Goggle can't (shouldn't) just ignore that fact.
It's confusing that Matt said, "For Google, nofollow'ed links are dropped out of our link graph; we don't even use such links for discovery" as recently as August 2007 if this was never their intention.
Perhaps they have to say this so people don't continue to link spam with the intention of being discovered. Call me an idealist: I'd just like to hear the real deal :)
"For Google, nofollow'ed links are dropped out of our link graph; we don't even use such links for discovery."
One could speculate as you suggest that a nofollow link to an established page may carry some weight - not in a "link juice" but a "keyword juice" one... In this theory "link juice" is how bright the words may shine and "keyword juice" is what words potentially become illuminated.
There are a couple points that I believe may play a roll in this.
1) What patrickaltoft mentioned about G treating links to images as embedded images is something I've seen over and over again. So when you're linking over, the NOFOLLOW won't do you much good.
2) Not sure about XHTML but with regards to HTML, NOFOLLOW only works in version 5... not 4
3) Jane is just such an amazazing SEO DIVA that Google indexes all iterations of her name, regardless of relevance or instruction. :)
So where is the "Jane Copland Random Fact Generator"? aka JCRFG
I've always divided up the algo into 2 main categories. Relevancy and Authority. It's a simple explanation, but this is from reading countless books on breaking it down and the simple explanation is what I reserve for clients (btw the best book so far is 'google's page rank and beyond - Langville & meyer)
Where a nofollow won't give authority, it still gives relevancy. I've seen this work in a controlled environment (made up keyword combo with sites of all the same rank competing). The one with the most nofollowed blog comments won.
Is this something that only seems to have started happening recently?I know there was the whole Clueless Image experience a few weeks ago, but I'm just wondering whether these new instances of images ranking from associations to nofollow'd links may be the result of Google bringing in this new VisualRank algorithm?
Perhaps it's an unexpected side effect as the new algo tries to make sense of all the images it can find near the words Jane Copland until it can figure out what she actually looks like.
Hey Scott, you should make this whole thread into an overly dramatic documentary.
I think this is a matter of other sites scraping SEOmoz' content and putting it on their own sites/newsgroups with the nofollows removed. I did a test in a YOUmoz post a while ago and there was no impact to rankings with all of the inbound links from YOUmoz to LyricVault.com (note: I have since moved on to another test regarding nofollows on LyricVault.com with deindexing, reindexing, siloing, and other nofollow related tests, so don't take today's code on the site as examples).
I'm trying to rank well for SEO specialist on Google and I have been at #12 for a while now with Google. We'll see if the above link will make a difference.
Brent D. Payne
the pages were not also indexed as I recall, right? so I the nofollowed links didn't help in discovery either...
That is correct SEO Smarty. Google was telling the truth. Maybe I am the only SEO in the world that looks for other ways things could be happening BEFORE I assume that Google is doing some type of F.U.D. or other tactic.
I have yet to blantantly catch them in a lie . . . but I'm not saying it doesn't happen--just haven't been able to find one.
Brent D. Payne
Despite what Google says, I have always had a sneaking suspicion that Google still follows NoFollow links, but that it doesn't pass PR through them.
I could be wrong, though. It would be interesting to make a webpage with no inbound links except for NoFollowed links, and see what happens. Would it get indexed? I suspect that it probably would, but it's PR (or at least it's toolbar PR) would likely be 0.
Of course, since I've never tested this, I could be totally wrong.
Consider it done... or, begun, at least.
I have several test-beds carrying out this test all the time.
So far Google are doing what they say - I see no indexing of my target pages, but there is indexing of target pages linked to only from the same page with a non-nofollow link.
All of my recent test-beds confirm that, for web searches at least, nofollow is precisely that.
It is hard, outside of a test-bed environment, to be sure that there is not someone out there who is linking to your target page, and I obviously have a limited number of test beds which I can control and monitor, but our research is fairly comprehensive and I am confident that I would stand in front of a client and say that Google do not use nofollow links for anchor text contextual marking, for link-juice or for even discovery of pages.
I do not have any data for images - I simply do not have any test-beds set up for this, but I would suggest that the image search algorithm, being much less of a target for spam, probably behaves differently and still uses nofollow links for both detection and for anchor text contextual marking.
My conclusion? That is an ugly cat.
Good analysis. It does work differently for images, although I'm sure we could invent awesome ways to spam as a result ;)
Cat = totally horrifying. I'm glad it doesn't rank anymore.
Hi Jane. I noticed this, too, first at Yahoo. I blogged about it here: https://www.seo-writer.com/blog/2008/04/09/yahoo-violating-nofollow-attribute/ . Then I started noticing at Google, too, there were backlinks reported that had nofollow attributes attached. I am much less concerned now about nofollow attributes; as long as the website appears to have trust value with the search engines (good rankings, reasonable PageRank, significant backlinks) I am pretty sure the links will be followed.
Please feel free to add a comment to this on my blog, too...which, by the way, is a dofollow blog. :-)
Yahoo! Site Explorer has always reported nofollow links as inlinks. This has not historically meant that these have necessarily been included in weightings by their ranking algorithms.
The recent Yahoo! update has vastly altered results from Yahoo! Site Explorer across almost all of my clients. In most cases there has been a marked drop in the number of reported inlinks, but in one or two instances I have seen a several thousand percent rise. I am also seeing situations where reported inlinks from 'all pages' are lower than those reported from 'Except this domain'.
For the moment at least, I am treating Yahoo! Site Explorer as being somewhat less accurate than it previously was and more like one would treat Google's link: results.
I posted a comment on a blog with a link to my website.The link is nofollowed but it is showing up in Google Webmaster Tool. Even if Google seems to take it in consideration, my site is nowhere in the SERPs for the anchor text of that link. I aggree with Sean and David, there might be some investigation to do about this subject.
The problem is that Google con't treat nofollows as absolutes. They don't even "noindex" noindexes! ...
Just an update: Danny and Ciaran's little experiment above is showing results. Their pictures, charming as they are, now rank for my name. This is not an invitation to spam SEOmoz comments in order to have images rank. Although you all probably will anyway.
Hey, Google? Loophole alert. Please fix immediately, as the Black Sheep poster isn't flattering. Thanks :)
Ooh Jane Copland why so upset?
Just letting you all know that I can delete these comments whenever I like :)
This proves there is more to Google than meets the eye?
I have proved it a long time ago :)
But since you need a BIG VOICE in your industry so it doesn’t get much response but anyway I am somehow happy that I was right :)
Here it is, I used to write this blog but then due to unavailability of time I quit. :)
https://seomization.blogspot.com/2007/10/nofollow-tag.html
IMHO I don't believe Matt Cutts when he says that nofollowed links are not considered at all.
For example Google Webmaster Tool shows also nofollowed incoming links, why store this data if you don't want to use it?
Maybe that the only way to block Googlebot is using Javascript? :-)
Great job Jane!
Un saluto dall'Italia!
Do we even have confirmation that the Google Imagebot treats the nofollow the same as Googlebot?
I would think the spiders work much differently.
@mbarr - good point. I know for example that there is a separate META tag that you can use to tel the Google Imagebot to "noindex or nofollow" a page, so perhaps it's the same with images.
I also side with those who believe Google will follow but not pass value links that are no-followed.
If the question is why is the content indexed, then perhaps a no-follow or robots.txt is in order.
I guess this explains why Rand's profile page ranks for "orgzhetwarhyu" even though the link on his post about only the first anchor text counting is nofollowed.
Oh, something is definitely up.
https://archshrk.com/2008/04/nofollow-my-arse
As far as I'm aware, no follow'd links do show up in webmaster tools area. How is that possible if G isn't at least following them?
Is that possible? I think G will not follow the nofollow links as far as i tested and tracked.
Welll... how would you explain the fact that these images rank?
I read it somewhere not to long ago, but have no idea where, that because of the amount of links and data that Google crawls sometimes it doen't fully process these tags. I find that a little hard to believe because it's just a machine and they do what they are told, but I also can't even imagine the amount of data that they crunch in a day. This could be part of the reason these sort of things sneak through. Google is crawling as fast as it can and doesn't pay attention to every little detail in this case that being the nofollow attribute.
Just throwing that out there...
The problem is that it is VERY hard to run a truly scientific test to gauge results. If someone here can devise a truly emprical approach I'd be glad to run with it.
Create two pages targetting the same unique keyphrase, give one no link love, give the other nofollowed link love. Repeat many times.
Well you cant just put them on the same domain because how would you get each cached without biasing results? I don't have a hookup at Google for that. So you would need to have them on brand new, never before registered domains that dont have any of the letter combinations in the url (preferably numeric domains). Then you would need to have each domain with a unique ip address. Again preferably never before used. You would need content on that page to be identical - which kind of negates the experiment right there as there would assumedly be a bias in favor of whichever site is cached first. You would need no inbound links. No pagerank, which leaves no way to get your site indexed quickly. You would need the whois data to be identical on each site. You would need the time of registration identical. Running approximations many times would provide us with a correlation, which would probably be useful for our purposes but I guess I just don't care enough to put that much time in.
When talking about nofollow, what do you think about sites that put nofollow to all external links (even if quoting)? I am aware that some people want to avoid link juice leakages, but... does it make sense?
I don't know if this question should be filed under "Stupid Questions" but is it of any importance at all if you put the nofollow before or after the URL?
The reason I'm asking is I've seen people that I know are good SEOs put the nofollow after the URL, but if I put a nofollow after the URL on a link on my Worpress blog, publish it and then look at the source, WP moved the nofollow to before the URL.
Just a thought...
It was not a stupid question. I think it is a bit unfair to put nofollow on everything. Or is it only me?
@Mighty workshop not a stupid question - but its up to you. your blog, your rules. No follow and hoard link juice, but then people are less likely to follow links to you, or follow links, and loose some juice but gain credibility. Just for the record, esseokillen wasnt referring to your question as stupid, but his own ;)
@esseokillen normal practice is to put it before, and WP must have been coded to correct to change position.
Thank you rishil for clearing things up. You're right, I was referring to my own question. And thanks for your answer :)
I got confused :) Sorry. Thanks for your clarification, rishil.
No wonder, why the comment spammers are still not out of business yet. If they read Jane's post, they will have yet another reason to continue to spam our blogs.
I believe that 'nofollow' doesn't exactly mean 'we don't follow this at all, ever', but I don't have any real idea of what exactly it does mean.
It may be possible that nofollowed links end up in google's index not through googlebot per se, but after being indirectly presented to googlebot through a third-party that pays no attention to nofollow tags. Perhaps some sort of scraper bots might grab such items, put them on another site and yet leave the original image URL which G then indexes...
@Jane - dont you remember the clueless photo fiasco? That was no followed but ranked for anything and everything for seomoz plus clueless, including a number of SEOmoz profiles.
Not only is the image being indexed, the nofollow isnt stopping association with anchor text or surrounding text.
Which kind of makes you think that spamming comments for image optimisation (and maybe video) is a MAJOR possibility.
patrickaltoft explained it quite well in a comment to this post, as in the "clueless" thing - when you link to an image, Google sees it as an embeded image and not an exterior image, so the crawl happens no matter what the link is labeled as...
But the rss are not in the cache of Google I believe that Google can not (or does not want Google) rss feedburner see, what you think?
https://64.233.183.104/search?sourceid=navclient-ff&ie=UTF-8&q=cache%3Ahttp%3A%2F%2Ffeeds.feedburner.com%2Fseomoz-comments
Google probably does not value the same way with many search results that search with few results
I suspect that images are treated differently when it comes to ranking. I believe they have a ranking factor that weighs slightly on the alt text of an image compared to a links anchor text on a page. If the alt text has common factors such as name or vertical with the anchor text of prominent links on a page it will associate those images with those links and therefore return those images for an image search that is based off the alt text in the links.