X-rated talent

Recently, while researching a different story, I asked a telecoms industry executive about the problems of building high performing networks at scale. Where, I asked, did the cloud providers and SaaS developers recruit their techies?

Recently, while researching a different story, I asked a telecoms industry executive about the problems of building high performing networks at scale. Where, I asked, did the cloud providers and SaaS developers recruit their techies?

“Look at the types of talent they need to hire, and ask who can do things like that. Often they’re coming from the porn sites,” he said, “we see a lot of them. These companies need to hire the guys who have built at scale.”

So, if you want to hire staff with a history of adult entertainment on the CV, what sort of company are they coming from?

First, what sort of scale are we talking about? The proportion of the internet that is pornographic is hotly disputed. Some statistics are endlessly repeated, but not really very useful. An example is the research by net-filtering company Optenet which concluded that  “Pornography makes up 37% of the total content on the internet”.

But, read below the headline: “The data contained within the Optenet report is accumulated and compiled from a database of hundreds of millions of URLs”. The 37% means that almost two in five URLs were related to porn, which is somewhat different and, of course, a statistic that useful to a company that was selling a way to filter the content.

The online magazine Extremetech gave us some potentially more useful figures in its oft-quoted article “Just how big are porn sites?”. This is notable because YouPorn (the number two site at the time) helped with research by volunteering details of the volume of data it was serving - which, in 2012, was already 28Pb per month.

YouPorn “opened the kimono” in the phrase that technology companies use, and which seems queasily apposite in this case. In 2012 it claimed to host 100Tb of files and serve more than 100 million page views per day (the author could not independently verify these numbers): this was burst traffic of 100Gb per second. It logged between 8-15Gb of user data every hour. The author concluded that this site alone accounted for about 2% of the internet traffic at any moment, and that  “it’s probably not unrealistic to say that porn makes up 30% of the total data transferred across the internet.”

A later analysis that blog site Ministry of Truth used Alexa data to measure individual visits and page views across all high-traffic web sites, rather than volumes of data. Its calculations based on visits are an order of magnitude lower, claiming that pornographic sites represent about 2-3% of global internet traffic. “If you add together the monthly visitor figures for the 10 busiest porn tube sites on the internet, you’ll still come up 25 million visitors short of the traffic figures for Wikipedia,” it concluded.

However you measure the scale of the business, size isn't everything. It’s what you do with it, and pornography sites have often been forced to solve specific technical problems at scale (video streaming being the most obvious) ahead of the mass market.

The business models are often less impressive. For a more academic insight into the revenue structure of pornography online, a paper called “Is the Internet for Porn? An Insight Into the Online Adult Industry”, was presented to the 9th Workshop on the Economics of Information Security in 2010. Authors Gilbert Wondracek, Thorsten Holz, Christian Platzer, Engin Kirda, and Christopher Kruegel analysed thousands of web sites, and became webmasters themselves (for the purposes of research) so they could characterise “the industry’s economic and security related structure”. As they pointed out, at the time they did their analysis, the internet porn industry yielded more revenue than Microsoft, Google, Amazon, eBay, Yahoo!, and Apple combined.

On one hand, the counter-parties they dealt with showed a knack for automated trade: for example, the authors found many porn sites were openly “traffic brokers”, directly trading traffic by automatically redirecting users to your site for a fee. The users could be segmented by interest (don’t ask) or location. When the authors experimented with this, they found that brokers instantaneously delivered the correct profiles, in exactly the numbers they requested, at the correct rate, for every order. On the other hand this efficient and lucrative trade is a useful way to distribute malware. “We discovered that a malicious operator could infect more than 20,000 users with a minimal investment of about $160,” they concluded.

The ability of the industry to build high-traffic sites is balanced against what the authors call “questionable methods and techniques that can best be described as ‘shady’.” If more mainstream businesses use the technical, if not business, innovations that are common in this world, it would be surprising if they didn't adopt some of its developers and network engineers too.