Web porn: Just how much is there?
Many things in porn are exaggerated, including the statistics regularly quoted to show how much pornography is on the web.
The need to get a good sense of its size and reach has sharpened as British politicians, pundits and the media debate what influence pornography is having on children, teenagers and their understanding of what sex is all about.
A lot of different statistics were quoted in those debates. However, few stand up to scrutiny.
One figure that cropped up again and again is that 37% of the internet is made up of pornographic material. Many of those people who quoted the figure took it from a press release put out in June 2010 by net filtering firm Optenet.
A spokesman for the firm told the BBC: "The statistics are not up-to-date and I would not use them to reflect the reality of the web nowadays."
That might just be because there is a lot more data around.
According to estimates from Scandinavian research centre Sintef, 90% of all the data the human race has ever produced has been generated in the past two years. That explosion is due to the rise of the web, smartphones, social media and the "big data" projects in which businesses, governments and scientists are involved.Market maker
But it is worth asking if 37% was ever accurate, even in 2010. Optenet said the figure came from a "representative sample" of about four million URLs taken from its database of web content.
However, that same year the largest study ever done into human sexuality published a very different figure for how many of the web's most popular sites were devoted to porn.
The academics behind the research based their results on analysis of the million most frequented sites in the world.
Their estimate? Just 4% of those websites were porn.
While the two studies do not measure exactly the same metric - Optenet counted pages, the academics sites - it's worth noting that the number of pages on a site says nothing about its influence or audience. As other studies suggest, porn sites are likely to be disproportionally large as they trade on giving visitors new content and, as a result, create hundreds of new pages each day.
So, suggest these studies, porn sites have huge archives but the number of pages that people actually view is relatively small.
This means that it is better to measure visitor numbers rather than site size, said Dr Ogi Ogas, co-author of the huge 2010 study, A Billion Wicked Thoughts.
So, why do people pick the big numbers instead?
"Big numbers are more sensational and make for good press," said Dr Ogas. "But those big numbers have always been an urban myth."
Three years on, the amount of porn online was likely to make up even less of the total, he said, thanks to consolidation in the porn industry and the rise of sites that duplicate and syndicate content.Zombie stats
Claims about the total amount of porn online are worth questioning because further analysis suggests the same about figures for how much traffic those sites generate.
An article in Extreme Tech is regularly quoted for calculating that 30% of all net traffic is generated by porn sites. It reached this figure by estimating how much traffic one popular porn site generated each day and multiplying that by the "dozens" of other similar sized porn sites it said were out there.
However, there are problems with its maths. To begin with, it underestimated how much data travels across the net every day. Extreme Tech claimed daily data totals were half an exabyte in 2012. Figures from network hardware giant Cisco said the figure was 1.4 exabytes. At the very least, porn traffic is part of a much larger total.
There are also reasons for querying its traffic estimate. Extreme Tech said its example site got 100 million visitors a day.
However, for a different story about porn sites, the BBC was told a much lower daily visitor total for that site and many others, all of which are overseen by porn hosting firm Manwin.
Across all its sites, Manwin said it got about 70 million visitors per day. This figure may also be inflated because, as other studies have shown, porn sites are built to be click generators. Every mouse click on a video or still can kick off many more as a viewer is re-directed to affiliated sites, adverts or pop-ups.
So, visitor numbers for those sites and consequently how much traffic they generate look likely to be a lot lower. A blogger at the Ministry of Truth blog reached a broadly similar conclusion when they analysed the 30% statistic and many others in a series of posts. That blogger concluded that porn accounts for a few percentage points of overall traffic and is dwarfed by YouTube and the bigger social networks.
They described facts and figures about online porn as "zombie statistics" because, like the undead, they refuse to lie down even when effort was made, time and time again, to kill them off.
Few people took the time to look into the source of the statistics, said the blogger, which explained why people continued to quote figures that date from the days before the web existed or come from marketing materials, small unrepresentative surveys or off-the-cuff comments.
The dubious provenance of statistics about porn are well-known inside the tech industry.
"We are aware that a number of statistics are being used in relation to online safety and have concerns over their accuracy," said Nicholas Lansman, secretary general of ISPA, which represents net supply firms. Anyone quoting stats should check their veracity, he said.
"It is vital that any decisions in relation to online safety, like any other policy area, are based on evidence rather than myths and assertions," he added.
What is worth remembering is that even if there is a lot less porn online than many people claim, there is still a lot out there, said Dr Ogas "Fourteen per cent of searches and 4% of websites devoted to sex really are very significant numbers, when you stop to ponder it."