How Big Is Online Porn?

I’ve spent quite a bit of time over the last three weeks or so debunking a variety of spurious claims about online porn.

Amanda Platell claimed to have watched a child porn video she found using Google, but that proved to be an old commercial porn video in which the female performer was at least 18 years old at the time it was made.

The Internet Watch Foundation claimed that 1.5 million people in Britain have ‘stumbled across’ child porn online, but it’s ‘study’ turned out to be nothing more than a dubiously constructed opinion poll and the IWF’s own data shows that around three out of every four web pages reported to it by the public as containing child porn turn out to be false alarms.

Then there was The Sunday Times, which put up a rash of ‘statistics’ culled from an entirely dubious source, all of which turned out to be either flat out wrong or so woefully outdated that Facebook and Youtube didn’t even exist when they were first published online.

I’ve also shown, using analytics data from the Ministry, that of the estimated 1 billion Google searches generated by the British public every two weeks only around 1400 or so are looking for child porn and that of those under half use keywords that are specific enough to consider them to be genuine attempts to access child porn as opposed to people responded to recent media coverage by looking to see if its is actually possible for find child porn using Google, which it isn’t.

Still, none of that take us any closer to a viable estimate of just how ‘big’ online porn is in terms of the amount of Internet traffic porn sites generate – so how do we estimate that, and is it even possible to estimate how much porn is out there?

Well, before we try to answer that question, the first thing we need to decide is what exactly we are going to try an measure and, more importantly, how we can measure the scale of online porn in a way that allows to compare it fairly with other online services.

One way of doing this, which has been used by the technical press, is to think in terms of the amount of storage space given to hosting porn and the amount of network traffic used to deliver pornographic content to end users. There is, however, a major problem with this approach when it come to making comparisons with other online sectors because the online porn sector is now overwhelming based on video-streaming and that means that when you try to make comparison with a social media site, like Twitter for example, you are not going to be comparing like with like.

To understand what I mean let’s start with a simple question – how big is a single tweet?

Well, we already know that the maximum length of a single tweet is 140 characters which can be represented in just 140 bytes of data – 1 byte per character. On top of that, there are a few addition piece of information that Twitter need to store and transmit for the system to work.

For starters Twitter needs to know who is sending the tweet and who to send it to.

If there’s an image or a video attached to the tweet then Twitter will need to know where that image/video is stored.

If the tweet is in a language other than English and, in particular, in a language that uses a different character set such as Russian, Arabic or Chinese, the Twitter will need to know which character set should be used to display the tweet.

So, on top of the 140 characters allowed for an actual message, Twitter will use, store and transmit a small amount of additional data with each tweet in order to make its system work, not much additional data admittedly but enough for us to arrive at a conservative estimate that, in terms of data storage and network traffic, Twitter probably gets around four tweets to the kilobyte (1024 bytes).

So, how big is a streamed video?

Well that depends on a lot of things including the running time, the pixel resolution and frame rate of the video, the specific digital format used, whether or not the video is compressed and, if so, what type of compression is used and at what compression ratio – and that’s all before we get into the question of whether the video has a soundtrack, which throws in further questions about audio quality, format and compression ratios.

However, to give a few examples, a typical five minute Youtube video will weigh in at around 5Mb and a bootlegged feature film downloaded via bittorrent at around 1Gb, while a single-sided, single layer DVD image comes in at 4Gb.

So, assuming four tweets to the kilobyte, a five minute Youtube video is the equivalent of around 20,500 tweets, a pirated film comes in at around 4.2 million tweets and a DVD at around 16.8 million tweets, so you can easily see why comparisons based on the measurement of storage space or network traffic will fail to give either a fair or accurate picture of the popularity or scale of a particular online system.

Put simply, if you’re going to go down the measuring traffic route then nothing compares to Youtube which, as of January 2012, was streaming around 4 billion videos a day.

So measuring network traffic is non-starter for making fair comparisons, so what other measures could we use?

Ideally, we’d measure the numbers of people using different online services using Google’s widely accepted unique visitors measure but, unfortunately, that kind of information is no longer available for adult websites from Google’s Display Adverts Network planner, leaving us to took for alternative sources of information – and one such source is the web metrics company Alexa.

So what Alexa got to offer?

Well for starters, a list of the top 500 global websites with a estimate of the number of daily, weekly and monthly page view for each site given as a percentage of total global page views and a estimate for average number of unique pages that users view when they visit the site.We also get graphs showing the traffic figures for each site for the last two years and with that information and the figure given for the number of monthly page views for a porn site in the tech press article I linked to earlier, we can use a bit of simple maths and a spreadsheet to estimate the number of monthly page views and number of individual visits for each of the sites in Alexa’s top 500.

Well, actually I ran the figures for 99 of the top 100 sites because one site, an adult webcam site currently ranked around 93rd-95th, refuses  to allow Alexa to publish its data and because manually extracting the data from Alexa’s site is pretty time consuming and, frankly, boring task; but as those 99 sites collective account for just over 42% of global Internet traffic, as measured in page views, and because even a video-heavy website, like Youtube or a porn site, still has to embed each of its videos in a web page to send to a user’s web browser, that still gives us a very reasonable* and, importantly, fair means of estimating how much internet traffic is generated by people surfing for porn by comparison to the traffic figures for other types of website, e.g. search engines, e-commerce, etc. across 18 different categories of website.

* To check that estimates I have are in the right ballpark, I used the data for Youtube which is known to serve around 4 billion videos as day. At 1 page view per video, that gives a baseline for Youtube’s monthly figures of at least 120 billion page views per month plus what ever additional traffic is generated by people visiting the Youtube homepage, searching the site and uploading videos or tinkering with account settings. The figures for Youtube from Alexa put the its monthly page view traffic at 131.5 billion page views, so we’re certainly in the right area with our calculations.

From all that I generated a couple of graphs showing the monthly number of visits and page views for each of 18 different website categories, which you can see below (click for larger image)

AlexaTop100

So, as you might well expect, when it comes to generating online traffic, search engines and social media sites rule the roost and the numbers are staggering.

Twenty-two of the top 100 websites globally are search engines and a further eight are social media sites, including Facebook, Twitter and LinkedIn. Collectively they account for just over 58% of the monthly page views generated by the top 100 sites and 45% of all visits; that’s 43.2 billion site visits per month and almost 580 billion page views.

Web portals, such as Yahoo, which combine a search/directory service with news and other content and services, rank third in terms of visits but only fifth in terms of page views behind video sharing sites (e.g. Youtube) and E-commerce sites (e.g. Amazon) which are ranked fourth and fifth, respectively, for visits  and third and fourth for page views.

So what about porn sites?

Well, there are currently seven porn site in Alexa’s top 100, but we only have traffic figures for six of the those sites, all of which are free porn tubes, putting porn in tenth place in terms of monthly visits but as high as seventh place when it comes to monthly page views. Obviously the punters like to browse a little before settling down to their evening’s entertainment, which bumps up the page views count compared to blogging, which tends to generate fewer page view per visit. Traffic-wise, porn sites get just 2.4% of the total number of monthly visits to the top 99 sites and just 1.8% of page views, putting it behind not only search engines, portals, social media sites, general video sharing sites and e-commerce but also, in terms of visits, behind blogs, information services (e.g. Wikipedia, About.com, etc.), the big technology websites (e.g. Microsoft, Apple, etc. – although much of this traffic will be automatic software updates) and email services, although based on the traffic figures for sites around the adult cam site for which we don’t have any data, were its figures they might very well push porn up above email into ninth place for visits, but no higher.

Okay, so you might think that that’s all very well and good, but by taking the top 100 sites I’m including the absolute leviathans of the Internet (Google, Facebook and Youtube) which between them hoover up 21% of global visits to websites and around 45% of all monthly page views. Surely that creates a rather skewed picture of the relative popularity of porn, forcing down its numbers?

Well, no – you see although I didn’t hack my way through every one of the remaining 400 sites in Alexa’s top 500, I did take a 10% sample and use it to estimate how much traffic overall those sites add to the overall picture and it works out to be around another 6%.

Of the 400 sites ranked from 101-500 on Alexa’s list just ten are in the business of delivering adult content; there are eight more free porn tubes to add to the six in the top 100 plus one more adult webcam site and an adult dating site, and collective these sites generate just 0.158% of global monthly page views, which is 2.65% of the total amount of page view traffic generated by those 400 sites.

Overall, the figures here seem to present a relatively consistent picture, one which suggests that overall, adult websites account for no more than 2-3% of global Internet traffic, measured in terms of both individual visits to websites and page views, and to put that in perspective, if you add together the monthly visitor figures for the ten busiest porn tube sites on the Internet, you’ll still come up 25 million visitor short of the traffic figures for Wikipedia.

So, to sum up, online porn is less popular than search engines, web portals, social media, video sharing, blogging, online shopping, email, keeping your computer’s software up to date and virus-free and Wikipedia.