“15% of all Google searches have never been done before.” I have so many questions about this

“Fun fact: 15% of all Google searches have never been done before.” When I saw this statement today, I thought it was an outdated percentage. So much research has been done in the meantime, so many new search terms can’t be added every day, can it? But Google turned out to confirm this statistic: right now, 15% of searches are new. In 2007 it was still 25%, since 2013 Google reported that the percentage of new searches is stable at around 15%.

I have so many questions about this: what kind of new research is this? Are there many searches with a typo? Or are these new substantive issues? Are these new combinations of existing search terms? How many search terms are there per day? And is 15% of searches new to the average user? Or are there some users that generate a lot of new assignments? (Here I take a meaningful look at my daughter who, for example, today searched for ‘skeip belen opa’ and ‘beautiful mailboxes seven years old’)

Over lunch, I asked a student if he could guess how many search terms Google processed per second. It turned out that he was able to do this: he was looking for something himself about eight times a day, there are more than 80,000 seconds in a day, so he was looking for something on average every 10,000 seconds (he had learned that you can be careless with these kinds of estimates). can count – not mine by the way, but I was still a bit proud when he did that). There are about 8 billion people on Earth, of course not all of them have internet and maybe not all of them use Google. He guessed that a quarter of people use Google about as often as he does, so you have 2 billion people searching for something every 10,000 seconds, and that’s 200,000 search terms per second. It turned out he was only wrong by a factor of two – which is perfectly fine for such a quick guess.

According to recent sources Google processes 102,253 searches per second (I myself looked a second ago to see if 102,253 is a prime number). On average, Google therefore receives more than fifteen thousand search terms that have never been queried together before. But Google almost never says “Sorry, I have no idea what you mean”. Almost every word you type has been used by someone somewhere and these sites appear effortless and in case of typos the algorithm conveniently searches for words that are somewhat similar.

I thought wistfully back to Googlewhack: the game where you had to enter two existing words into Google in order to get exactly one webpage containing those words. From Volkskrant wrote in 2002 about big finds like athletics + math teacher (now yields 4,050 results) or semolina pudding + snow white (now yields 910 results).

The great thing about this game was that you couldn’t brag online about a Googlewhack you found, because once you did, there were two websites that listed those words together. This week, someone pointed out to me that comedian Dave Gorman was hosting a show about his big Googlewhack adventure – and there’s a full recording online. It’s enjoying a bit of internet nostalgia, as Google processes 592 million botched search terms in the meantime.