There's a 95 per cent chance that the average length of all Web pages in the world is somewhere between 6.2 and 6.8 printed pages. Photo / Thinkstock
We all understand, courtesy of our daily contributions to social media and our search struggles with Google, that the Internet is huge - and only growing more huge. More than 3 billion people are now online. By the end of 2016, Internet traffic could eclipse one zettabyte a year.
But who the heck knows what a zettabyte is?! I want to size the Internet in concrete, physical terms: like, if I printed the whole thing out, I'd have .... how many pieces of paper?
This isn't a novel question, incidentally: artists and mathematicians alike have tried to answer it before. Two years ago, the poet Kenneth Goldsmith invited interested strangers to print Web pages out and mail them to his "Printing the Internet" show in Mexico. In March, meanwhile, two students at the University of Leicester published a journal paper that sought to estimate how much of the Amazon we'd have to fell in order to print the Web.
That paper was largely hypothetical - more a thought experiment than anything else. (Among other things, it makes use of a lot of "assumptions" in the course of its math.) I emailed co-author George Harwood, however, to ask how the page-length of the Internet could be computed more concretely.
It would involve lots of menial labour. And, of course, a spreadsheet.
How many Web pages are there?
In 2007, as part of his master's thesis at Tilburg University, the Dutch Web consultant Maurice de Kunder developed a statistical method for tracking the number of pages indexed by major search engines. The math and technology behind this tool are pretty complicated - his thesis ran 68 pages, in Dutch - but eight years later, it's still constantly updating the number of pages in Google and Bing search.
A couple caveats here: Even this figure is an estimate, and it doesn't capture anything outside the reach of search engines. But for our purposes, de Kunder said, we can assume there are roughly 47 billion pages on the indexed, searchable Web.
How many printed pages is the average Web page?
Meanwhile, to find how many printed pages each of those 47 billion would be, I needed to visit a representative sample of websites and try to print out each. To arrive at a fair average - one with a 5 per cent margin of error, and a 95 per cent certainty - I'd have to test 385 random sites, as chosen by the so-called "Random Website Machine".
This is a fun game you can play at home (!), if you have absolutely nothing else to do: click the "random website" button; hit CTRL+A and CTRL+P; and record the resulting "pages" number in a separate spreadsheet.
There was the site for a Taiwanese Little League (two pages); online stores selling Orthodox icons and horses' hoof grease (1); the homepage for the Czech national lacrosse league (4). I learned that West Virginia only has one business newspaper, and that Bulgaria's constitution was adopted in 1991. Also that, in 2000, the BBC called Britney Spears a "teenage pop phenomenon."
A few of these sites would, of course, take up many printed pages: say, Wikipedia articles or the homepages of chatty personal blogs. But the vast, mundane majority of the websites I visited - the pages of Polish municipal governments, say, or a recipe for "cattle drive chili" - only took up one or two. (We forget that our modern Internet sits atop strata of Geocities pages and long-forgotten forums and sites for dentists offices in, who knows, Kalamazoo.)
The average site came out to 6.5 printed pages - slightly longer than the Wikipedia page on Lithuania's performance at the '92 Olympics. In other words, there's a 95 per cent chance that the average length of all Web pages in the world is somewhere between 6.2 and 6.8 printed pages.
The number of pages it would take to print the Internet = 305.5 billion
From here, of course, the experiment gets pretty easy. Multiplying 47 billion by 6.5 gives you 305,500,000,000 pages, approximately. This is, to be clear, just an estimate: There's some room for error in my page-length average, and the Internet's changing all the time. But it's still difficult to even contemplate how much material that comes out to: It's like 212 million copies of Leo Tolstoy's War and Peace.
"We're dealing with abstraction, and we have no idea what this is," that Internet-printing poet, Kenneth Goldsmith, told my colleague Dan Zak in 2013. "We need new metrics for infinity."
Even Goldsmith's metrics failed, however - and failed rather spectacularly. While he and his gallery would later brag that the show collected 10 tons (three whales!!) of printed papers, that's just a speck in the huge, abstract Internet sea.
Assuming we're not getting fancy, and just printing on normal copy paper, those pages add up to 122 million tons, or ... a lot of whales. Seriously.