Google and Child Porn-Related Search Engine Traffic

Warning: May contain actual evidence.

Three weeks ago, I published a couple of articles in the space of two days that conclusively debunked Amanda Platell’s claim that it was easy to find child pornography online using nothing more than Google proving, first, that a porn video she described watching in her article was actually a legally produced commercial porn video and, second, that the female performer in the video, whom she described as being ‘in her early teens’ was at least 18, if not 19 years old at the time the video was made, way back in 2001.

In writing those articles, I chose my words carefully in order to quietly run a little experiment.

You see although I’m no SEO expert, I know my own blog and how it interacts with search engines like Google pretty well.

The Ministry has been around for a pretty long time as British blogs go, five and half years on it’s present domain, and that longevity plus the facts it’s updated reasonable regularly, if not as much as it used to be, and the fact that I’m careful to avoid practices that Google’s indexing algorithms treats as evidence of spamming.

The upshot of all this is that unless I write about something that’s stupendously popular or common place then my material will frequently achieve very high search placements on Google, including placements in the all important top ten and that is exactly what happened on this occasion.

The second of my Platell articles, which uses the term ‘child porn video’ in it’s title and URL is ranked sixth globally on Google searches for ‘child porn video’ and of the 213 different child porn search terms that returned an entry for the Ministry in the 19 day period for which I’ve been tracking the SEO data, 110 returned a top ten ranking with a further 30 ranked between 11 and 20.

That’s on the back of just four published articles that refer to child pornography that, between them, have pulled in around 10,000 unique visitors to the Ministry, and the mere fact that personal blog with a modest regular readership is now one of the top ten ‘child porn’ related websites in the world according to Google’s search ranking should immediately tell you just how unlikely it is that you’ll find actual child porn using that search engine.

Nevertheless, we’ve got a few statistics to play with here, courtesy of Google Analytics, which should help to supply  some much needed perspective to the current moral panic surrounding online child pornography but before we do the numbers I’d like to get my caveats in first, so there can be no misunderstandings as to what the figures I’ve got are, and more importantly aren’t telling us.

First, the data I have is for searches carried out in English, so the data doesn’t tell us anything at all about the extent to which people may or may not be searching Google for child porn in other languages.

Second, it doesn’t automatically follow that because someone has carried out a child porn related search on Google that automatically makes them a paedophile, or that they were looking for child porn to obtain a bit of sexual gratification.

Particularly when it comes to the figures for the UK, where there has been a huge amount of publicity about this issue in which Google’s name has been prominent, some of the more generic searches for terms such as ‘child porn video’ may easily amount to nothing more than a bit of ill-advised ‘rubbernecking’ – i.e. people trying to find out whether it’s as easy to find child porn on Google as the media have been claiming. It isn’t, of course, but that won’t stop prevent some people from taking a look just to make sure.

To be absolutely clear, the keywords I set out to trap here are those most likely to be used in searches for child pornography made by people who have no real idea where to find it – that’s why they’re using Google of course – and I deliberately avoided any known signature keywords for child porn in my recent articles in order to ensure that I only picked up casual searches.

Third, when it comes to data for different countries, don’t read too much into the raw figures and certainly don’t take these figures as being any way indicative of the levels of interest in child porn in those countries. In addition to the fact that we only have data for searches in English, the numbers of searches for child porn from individual will depend on the amount of search traffic those countries generate which, in turn, depends on everything from the size of their population to levels of internet access in those countries.

We also only have data from one search engine, Google, so much will also depend on Google’s share of the search engine market in particular countries and this can vary considerably from country to country. In the UK and most of Western Europe Google’s share of search engine market is upwards of 90% but in the US it’s just under 70% and in Japan, where it loses out to Yahoo, it’s just 31%.

And, of course, we also have no way of knowing how many of these searches were made by unique individuals, as this data is not captured in the SEO figures. Data for unique visitors is only captured when people actually click through from Google to the Ministry and only 1 in 8 of those who found any of my articles from child porn related keywords, actually bothered to pay me a visit.

Fourth, and this one is particularly important, despite what you may have read in the media about ‘gateway effects’, the idea that people start out just watching child porn but then go to become actual abusers, the actual research evidence from numerous sex offender studies is equivocal and does not support the existence of such a general gateway effect.

Voyeurs and ‘hands-on’ sex offenders are psychologically very different and where there is evidence of people appearing to make the jump from viewing child porn to committing sex offences against children, what researchers have invariably found is that their sexual interest in children pre-dates either kind of activity and that, in many cases, the hands-on offending began before the offender started to use pornography.

There is no simple pipeline from watching child porn to committing sex offences against child just as there is no such pipeline from watching adult pornography to rape and indeed, on the latter, there is no reliable evidence that hands-on sex offenders are more likely to seek out and watch violent pornography than any other run of the mill porn consumers.

Finally, to put things into their proper perspective, in 2012 Google handled 1.2 trillion Internet searches over the course of the year, an average of 3.28 billion internet searches per day.

If we assume that their figures for this year are going to be at least comparable to those for that year we can estimate that over the same 19 day period that I’ve been tracking child porn related searches, Google will have processed just under 62.3 billion internet searches.

That’s all searches in any language but, of course, we’re only really interested in those made in English and this is where it gets a little trickier because, as far as I can see, there are no published figures for numbers of searches by language.

Nevertheless we can make a ballpark estimate as Alexa’s traffic statistics indicate that Google’s US/Global search portal (Google.com) handles around 12 times as much traffic as the company’s biggest country-specific portal (India) and 25 times more traffic that it next largest set of local portals, which include the UK, Germany, France and Brazil. Throw in the fact that five of Google’s top 15 local portals default to English (India, UK, Hong Kong, Canada and Australia) and it would not be unreasonable to assume that about two thirds of Google’s global search traffic is in English.

So, apply that estimate to our figures we get a baseline of 41.5 billion searches in the 19 days for which I’ve been tracking the traffic to my blog.

So how many of those were looking for child porn?

In total, a touch over 13,500, which is just  0.000033% of the estimated amount of search traffic in English over those 19 days.

That’s the figure for searches from anywhere on the global, and in total Google logged child porn related searches from 164 different countries and territories around the world on every continent, including countries that are known to exercise strict controls on access to pornography, and the internet generally, such as Saudi Arabia, Iran, Burma, and China.

Of our global punters just over 23% used search terms that indicating that there were interest in either prepubescent children (age 10 or under) or porn involving a child of particular age. Of those, 72% were looking for a 12-year-old child, 4% for an 11-year-old and 21% for prepubescent children with rest (3%) looking for porn featuring a teenager under the age of 18.

7.7% of searches specified a particular gender and, unsurprisingly, 97.5% of those were looking for females.

Finally, 67.3% of these searches were looking specifically for videos, with just 0.48% looking specifically for still images – the remaining searches did not specify a particular medium or format.

As far as where these searches were coming from, this graph shows the top 50 search locations as a percentage of the total number of searches – click to embiggen and do please remember all those caveats from earlier about language, population, internet access and search engine market share:

cpstatsT50

So The UK comes out on top followed by the US, India and Turkey with ‘Unknown Region’ in fifth place – that’s the 4.69% of punters who are at least smart enough to hide behind an anonymous proxy and block any web-based trackers when searching for child porn.

These top 50 countries accounted for 82% of searches of which 40.5% came from Europe, 35.6% from Asia (including 1.1% from the Middle East), 15.8% from North America and a little over 2.6% each from South America, Africa and Australasia/Oceania.

Looking specifically at the UK, which is the one were most interested in, Google logged a little over 1800 searches in 19 days.

Currently it’s estimated that the UK generates around 1 billion internet searches every 2 weeks, so that’s around 1.36 billion searches in 19 days of which Google’s current market share in estimated at 90.36% giving us a baseline of 1.23 billion searches.

Put those numbers together and what we get is a figure of 0.00015% for the proportion of UK Google searches using keywords which suggest some degree of interest in child pornography, give or take any rubbernecking that might have stemmed from the current media hype surrounding child porn and Google.

That’s an average of around 675 child porn searches a week for the entire population of the UK, which is around 63 million people of which around 50 million are aged 18 or over.

Based on figures from the Office for National Statistics, 79% of UK adults use a computer on at least a weekly basis (67% on a daily basis) and 80% or so of UK households have internet access in the home, so it wouldn’t a stretch to estimate that in any given week around 30 million adults in the UK go online at least once.

So even if we were to assume that every one of those searches was carried by a unique individual over the age of 18, which is highly unlikely, then we’re looking at just 1  in 44,500 adults typing phrases like ‘child porn video’ into Google each week – and let’s not forget that we have no way of knowing exactly why these people chose to use those keywords or how many were actually looking for child porn as opposed to looking to see simply if it was possible to find child porn using Google in the wake of everything they’ve been reading in the press or seeing on the TV in the last few weeks.

We can, however, narrow things down a little by making the assumption that any searches that specified a particular age, age group or gender are most likely to represent genuine attempts to access child porn rather than just curiosity searches or rubbernecking and 43.9% of UK searches did indeed include such specific information.

39.8% of searches contained some sort of age-related keyword with 12-year-olds again being by some distance the most popular search, accounting for 79% of age-based searches for child pornography. Prepubescent children came in next at 16% with searches for 11-year-olds accounting 4.4% of age-based searches.

Only 8.3% of all UK searches specified a gender – all female.

Finally, searches for videos massively outweighed those for still images, although video searches accounted for only 36.5% of all UK searches, compared to 67% in the global data, with just 0.66% of searches looking specifically for still images.

Give or take the fact that our own home-grown voyeurs are much less fussy about their preferred medium these figures are broadly consistent with the global picture in terms of of both scale and in terms of expressing a sizeable preference for pornography featuring 12 year old girls – exactly why that should be the case is a question that requires further research.

It has to be remembered that the kind of activity we’re looking at here is not that of the kind of hardcore paedophiles who manage to amass huge caches of child pornography by tapping into private peer-to-peer networks and the like. The people whose search activity we’ve got a snapshot of are, for the most part, the kind of people who, if they are ever arrested and convicted, are likely to merit half a column of coverage in their local paper unless they hold some sort of position of trust, such as teacher, vicar or a local councillor, to make their arrest/conviction particularly newsworthy.

They are, to borrow a phrase from Douglas Adams, ‘mostly harmless’ in the sense that very few of them are ever likely to go on to present any kind of direct risk to children (see Endrass et al., 2009).

The numbers involved in this kind of activity and the amount of such traffic they generate seems to be relatively small. If we consider that the people making age/gender specific searches are to one who are most serious in their efforts to obtain child pornography this it’s not reasonable to assume that they are also likely to make repeated searches for child porn on different days and multiple searches using different keywords, behaviours that will very quickly eat into any estimates of prevalence based on an assumption that one search equals one unique individual, leaving us with an active population of voyeurs, in the UK, in the low thousands of individuals.

Not exactly Paedogeddon then, in fact not even close.