In an earlier post, I mentioned that I'd noticed a Google ripple take place on September 7th, with rankings being heavily affected for some sites, although I expected it was probably a narrow set of sites. A few days later, I think the fact that I called it a "ripple" was a foreshadowing of what really happened.
Barry pointed out an interesting Google Groups thread that caught my eye. Not long before I saw Barry's post, I noticed something odd in my Google Webmaster Tools for the site that was part of the earlier "ripple". I was seeing the error "We can't currently access your home page because of a timeout." and "URL timeout: DNS lookup timeout." Barry's post alerted me to the fact that it wasn't just me.
Reading the Google Groups thread, I saw that the earliest date that anyone had noticed this problem was September 6th - the day before the "ripple" occurred. Ah ha! Ding! Ding! Ding!
I now think the Google "ripple" should be re-named to The Google Butterfly Effect. The butterfly effect was first popularized in 1961 by Edward Lorenz. According to Wikipedia, "the phrase refers to the idea that a butterfly's wings might create tiny changes in the atmosphere that ultimately cause a tornado to appear (or prevent a tornado from appearing). The flapping wing represents a small change in the initial condition of the system, which causes a chain of events leading to large-scale phenomena. Had the butterfly not flapped its wings, the trajectory of the system might have been vastly different." In 1972, the title of a talk given at a meeting of scientists was Does the flap of a butterflys wings in Brazil set off a tornado in Texas?.
So many times over the years, I've heard Matt Cutts or GoogleGuy admit to "minor" changes having been made, but I don't think Google really "gets" that with their vast influence over the search world, even the tiniest things can cause huge ripples.
A simple bug in the crawling of some sites could radically change the SERPs and affect the fortunes (both good and bad) of many. Obviously, changes have to occur, and just as obviously, bugs will occur as well. Those are things we have to expect and live with. I do think, however, that Google needs to take great care, and more importantly, RECOGNIZE the impact that even the tiniest events can cause. I'm talking about an attitude adjustment here.
How many times has it taken Google several lifetimes (or so it seems) to recognize and admit to an error on their part? Remember 302s for example? Hopefully, with each misstep along the way, Google's blinders will gradually fall off as they see just what a wide ripple is cast whenever they make a boo-boo. Google's little "oopsies" can wreak great havoc. I'd like to see Google become more aware of this fact, thereby causing them to take greater care about causing oopsies in the first place, and then responding much, much more quickly when the oopsies are pointed out to them.
Luckily, Susan Moskwa is already looking into this particular "oopsie" and hopefully, a quick resolution will be the consequence. But I can't help but worry about all the future oopsies, and wonder if the next time Google flaps its wings, will it cause a world-wide-web of virtual tsunamis?
Note: It is entirely possible that this particular "oopsie" and the ripple I noticed on the 7th are unrelated, but I have a strong gut feeling that there is actually a direct cause-and-effect relationship at work here. Still, I concede that I could be wrong. 🙂
Donna, I love the comparison to the butterfly flapping its wings and causing a tornado half a world away. That’s always been one of my favority saying (though in my telling, the butterfly is always in China for some reason).
Donna, as usual, I couldn’t agree more.
Does anyone remember the Simpson’s episode where Mr. Burns builds a giant shade to block out the sun? To me that’s similar to what Google does. Now I’m sure they don’t intend to make mistakes and end up blocking out the sun but that’s not much consolation when you’re sitting in the dark.
They need to get down off their “oh we hire PHD’s” high horse and admit they make a ton of mistakes and even the smallest one can have HUGE consequences. I think a large reason they take so long in responding is because they place too much faith in their employees abilities. I mean how in the world could us lowly SEO’s find a mistake made by their amazing engineers?
Jon, I’m sure there’s some kind of “parents telling you a story about something in China” Freudian thing about that. 🙂
BlogEx, that is the way we are made to feel, imo.
One small ripple that might be having an effect is the complete reindexing of the ODP now that they redirect all requests to http://www.dmoz.org where previously both www and non-www for both dmoz.org and dmoz.com were all indexed. That must be having a small PR and backlink effect for some sites. Who knows how that ripple spreads?
Additionally, the Google directory was updated for the fist time in 18 months just a few weeks ago. What effect will that have on sites recently in, or recently out, and how far will that particular ripple spread?
Good points, g1smd, good points!
I also think that the Dmoz redirect thing hasn’t had enough attention, Google hasn’t even recognised the 301 from Dmoz.org/shopping to the www. one, I thought with a site like Dmoz the redirects would be picked up on within days, they should have stuck with dmoz.org and done it the other way around, also all of them 60+ regular directories that have been penalized has probably had an impact.
The problem with a site:dmoz.org search was that it was polluted with many other subdomains in the list.
Moving to http://www.dmoz.org means that only the directory and the various guidlelines pages will now show up in a site:www.dmoz.org search.
Your chaos theory reference is a very good analogy. It’s a well acknowledged fact that search engine engineers can’t reliably predict what any given “minor” algo change will result in.
In this sense, it’s actually entirely inappropriate to use categories such as “minor” and “major” (“big/small”) to describe code or algo changes – while it may reflect the amount of programming effort invested (e.g. “only one line of code changed” as opposed to “the entire module of 13,203 lines of code had to be rewritten), it’s quite meaningless on the cause-and-effect scale.
And yes – it’s fairly obvious, too, that Google hasn’t even remotely grasped the consequences anywhich move of their can (and will) have for major parts of the world’s economy. Talk of a sleeping giant…
Thanks for a great piece, Donna!
I think we’re pretty mindful that any change, even a small one, can affect large numbers of sites. Because of that, we try to do a lot of testing to ensure that changes are a net quality win.
If it makes you feel better, I was all set to ask around about the DNS resolving until I read that Susan was already on it. I think that’s the right idea; getting more people checking on things, reading forums/blogs, etc. means that we hear about stuff sooner and (I hope) respond faster. I was in a bugs meeting today talking about Dan Thies’ post about proxies, for example. Every search engine has to do things like choose the best canonical url, so getting that feedback helps. One of my regrets is that I’m finding myself on the critical path for things like emails, and so we need to find a way to scale that process better.
Thanks for chiming in, Matt. I have no doubt that scaling this kind of thing is difficult. But, well, although I feel sorry for you personally, it’s kinda hard to feel sorry for a company that has more money than…well..you know… hehe. I’m pretty sure y’all can figure something out. Throw enough brains and enough money at anything…
Anyway, I just know that there have been times, historically, that Google has denied things that they later un-denied, so I just hope that it’s all improving. Crossing fingers….