How did Google glitch globally?
The entire SEO community watched the online universe implode during the afternoon of August 10, 2020, when an unidentified bug wreaked havoc on Google search results worldwide. Lasting close to 24 hours, this search engine outage caused widespread dismay.
Major mistake or minor misstep?
A favorite argument-ender, “Just google it!” Reflects the trust that most of us place in the world’s biggest (and presumably best) search engine. And with a whopping 92%-plus of the search engine market, Google has surely built up massive databases during its twenty years in operation.
Although savvy researchers know how to cherry-pick the sources of their information (usually renowned experts and universities, research centers and other respected institutions), its millions of search results extend over dozens of pages to include even the most amateur (and often inaccurate) blogs.
Chaotic search results
But all that trust was blindsided around lunchtime on Monday, when posts and tweets spread the question worldwide instantly: “Is Google glitching?” Nothing this serious had happened within living memory (well, since it’s website was launched in September 1998).
Searchers went nuts when their searches returned crazy results: people looking for HVAC repairs in Florida got lists of caterers in California. Things were so bad that even Pokémon went off the air…
Initially, people in the know assumed it was just a clumsy algorithm update (recalling the bad old days), or even a return to the times when PageRank had a stronger influence. However, there had been no prior warning from Danny Sullivan, who typically announces Google updates.
Just after midday on August 11, a somewhat reassuring tweet from Senior Webmaster Trends Analyst John Mueller said: “I don’t have all the details yet, but it seems like this was a glitch on our side, and has been fixed in the meantime.”
Subsequently, he added: “On Monday, we detected an issue with our indexing systems that affected Google search results. Once the issue was identified, it was promptly fixed by our Site Reliability Engineers and by now it has been mitigated. Thank you for your patience!”
Through a quick analysis, the ‘issue’ in question does not seem to be yet another update in Google’s endless stream of almost invisible algorithm tweaks. Instead, it looks like a mistake of some sort, severely disrupting search results worldwide. In fact, many searches were completely useless, presenting only irrelevant data, while others refused to display websites, even when specific addresses were keyed in.
Reports from France, Italy, and Norway reported that only major news sites and marketplaces (like Amazon, eBay, AliExpress, and similar local platforms) were appearing in search results. Everything else vanished into the murky depths of the tail-end search engine results pages (SERPs).
Some users – particularly e-commerce websites – reported dizzying variations in page positions, jumping from a usual page 1 down to page 7 or more, and then back up again. Apparently affecting all languages and countries, it seemed to spread into niche markets as well, like local HVAC services.
Google Webmaster Trends Analyst Gary Illyes suggested (although without confirming the exact cause) that the problem might be rooted in the Caffeine web crawling and indexing system. Launched in August 2009, Caffeine was one of the most important updates in Google history.
Far more than yet another regular update (like Penguin or Hummingbird), Caffeine was a ground shift in the global infrastructure that handles Google searches and result lists. Scaled for the future, this far-sighted indexing infrastructure still today provides a solid foundation for faster and more comprehensive Internet searches.
Prompted by the diversity of the Internet and its multimedia content (videos, pictures, podcasts, newsfeeds, webinars, livestreams, and more), with thousands of websites flocking in every day, Caffeine was designed to keep pace with this exponential expansion, focused on shortening times between keyword trawling and search results delivery.
Complex, interconnecting structures
By constantly indexing the entire web in real-time, Google can present better search results, with more accurate and up-to-the-minute information. However, this is no simple task, and the Caffeine indexing system works on many different fronts that include:
- ingesting fetch logs
- rendering and converting fetched data
- extracting links, structured and meta-data
- extracting and computing signals
- scheduling new crawls
- building indexes that are pushed to serving.
Even a minor slip in any of the gears meshed into this incredibly complex system can have disastrous effects. As Gary Illyes tweeted on August 11: “If something goes wrong with most of the things that it’s supposed to do, that will show downstream in some way. If scheduling goes awry, crawling may slow down. If rendering goes wrong, we may misunderstand the pages. If index building goes bad, ranking & serving may be affected.”
He then stressed that: “Don’t oversimplify search, for it is not simple at all; thousands of interconnected systems are working together to provide users with high quality and relevant results. Throw a grain of sand in the machinery, and we have an outage like yesterday.”
Although the bug seems to have been fixed by lunchtime on Tuesday, businesses must still keep a sharp eye on their website analytics during the coming months. Sweeping (and distressing) changes in organic search rankings should be annotated and ignored for these two days, of course, as everything is now – apparently – back to normal.
This doesn’t mean that your rankings might not have changed, as they are always fluid. But whatever the global glitch was yesterday (further details are expected hourly), it has been fixed and reverted, according to Google, which is now “in the process of fully diagnosing” the problem.
What lies ahead?
Some companies may have lost visitors, customers, sales, and revenues because of this search engine outage, while others may have benefited. Rankings and traffic levels should now be back to normal.
However, one aspect to bear in mind is that content-lean landing pages using internal link dynamics and off-page factors to keep up the rankings may find it harder to cling to their positions in competitive SERPs. As a result, now would be a good time to flesh out landing pages with keyword-rich content that attracts searchbots.
Longer-term repercussions may be more subtle and appear only over time. The problem is that once Artificial Intelligence (AI) understands something through Machine Learning, it can’t be unlearned when wrong – like these muddled search results.
‘Healing’ Artificial Intelligence
These gross errors can be fixed only by additional learning overlays. Just like human beings, everything learned after these false facts is invalid but remains imprinted in the AI ‘brain’.
A critical segment of probability-based AI decisions depends on confidence: z-scores, standard deviations, variances, and standard normal distributions are all key concepts for fast and accurate search engine results.
In an analogy, when human beings suffer head injuries, neuroplasticity allows our brains to reorganize their functions by forming new neural connections; nerve cells (neurons) adjust their activities in response to damage, thus helping compensate for injured or diseased brain tissue.
Mindful meditation and cognitive behavioral therapy are two widely-accepted ways of rewiring human brains. So, let’s hope that the Google software engineers are finding equally effective ways of ‘healing’ their search engine, and its possibly damaged algorithms.
Watching out for pitfalls
The really scary part is that Google now knows that it can’t trust its own algorithms: they might be accurate – or they might be wrong. This uncertainty will undermine its understanding of its own analyses of its crawler findings, with uncertain outcomes.
There is no predicting what this global tangle of errors will mean for Google over the long term, nor how much damage has been done to its systems by what might be a rogue algorithm.
Meanwhile, the Search Engine Optimization (SEO) experts at CI Web Group are keeping a sensitive finger on the pulse of this giant AI ‘patient’. Constantly alert, we are monitoring any negative effects on listing positions and clicks while ushering prospective customers smoothly through sales funnels to those all-important Buy Now! buttons. During this sensitive period, why not check out what they can do for you by clicking here.