Monday, 23 April 2012

Internet Porn Filtering

11th February 2011

Dear Ms Perry,

I posted this text on your campaigns blog, as a comment on your internet pornography campaign (hence the references to Darren and Rowen).  I am copying it to you by email because your blog software strips out newlines - I hope the text with paragraphs intact will be easier to read.

Best wishes,

James N Kennett,
Worcester, UK.

Regarding your article on, I am sorry that you feel you have been "ridiculed for raising this issue", or "barraged with information as to why the internet should be treated differently".  I hope my comments do not stoop to ridicule, and are not so long as to constitute a barrage.  However, you are proposing significant changes to a £3 billion sector of the economy that serves millions of people, and so I hope my page of text is not too long to consider.  Although you say objections to your proposals are of three kinds, I am offering a fourth: that the proposals will not work very well, and the "porn filter" will end up being switched off even by responsible parents.  With regard to Rowen's comments, I don't much care for the "human right to access porn" argument either.  However, it would be a pity if this debate falls into the trap of "male versus female", because I am sure most fathers share Rowen's concerns about the online safety of their children.

If it were possible to create a perfect porn filter, it would definitely be a Good Thing; unfortunately, however, no such device exists.

I had the same experience as Darren with a 3G mobile Internet service blocking innocuous Web sites.  The problem with your proposals is that the filtering service will make the Web unusable because of false positive "porn sites"; yet it will not adequately protect children because it will have too many false negatives: even if 99% of porn is successfully blocked, millions of items of pornography will remain unblocked.  With or without the filter, it will never be completely "safe" to leave a young child unsupervised with a Web browser.

Personally, I do not care if the proposals are implemented, because I will immediately opt in to the full Internet; and so will many parents, when they hear for the umpteenth time "Mum, I need to finish my homework in the next 20 minutes and the Web site I need is blocked".

The ISPs will be pleased to cooperate with your proposals: asking them to sell us fig leaves is an irresistible business proposition.  The fact that many of the fig leaves will be in the wrong places will not worry them too much, as long as consumers are willing to pay, and - the biggest fig leaf of all - the ISPs are seen to be "thinking of the children".

I think the basic problem is threefold.  Firstly, billions of Web pages are provided by hundreds of millions of individuals and corporations in more than a hundred jurisdictions, and these pages are changing minute by minute as content is added, modified, moved or deleted; this method of provision and management is quite different from that of television channels, even if we choose to watch both media on the same screen.  The problem of inappropriate content is unfortunately inherent in the Internet itself.  This means that there really is no alternative to adequate parental supervision.  That is the second problem with the proposal: it is trying to provide a technical fix to what is really a social problem.

You may insist that an internet service provider is a kind of content provider, rather than a "pipe"; but in technical terms an ISP does indeed provide a pipe: the apparent contradiction arises because the issues that you raise are not entirely technical, but are also social.

The third problem is that the Internet is not just the World Wide Web.  Other services run on the Internet, most notably peer-to-peer file-sharing networks.  Teenagers are among the most prolific users of file-sharing networks, because they can find free music and films there.  Unfortunately, there is also a huge abundance of porn.  None of the advocates of filtering, in newspapers, discussion boards or even Mumsnet, mention Internet services other than the Web.  Perhaps this is because children know more about the Internet than their parents, and their desire for free music gives them an economic incentive to hide their file-sharing activity from their parents: after all, parents might disapprove of "copyright theft", or be freaked out by the pornographic material listed alongside the music.  If, when the ISPs were showing you their range of highly effective fig leaves, they neglected to mention the significance of file-sharing networks and the nature of much of their content, then shame on them.

It would be possible for ISPs to block file-sharing services completely, but ISPs' customers really do want to use these services: they are one of the most popular types of Internet use.  Therefore ISPs have an economic motive not to block these services, and that motive has nothing to do with porn.  Children too have an economic motive for resisting or bypassing any block, and again that motive (free music and films) has nothing to do with porn.  Asking ISPs to block file-sharing services is therefore unlikely to be effective.

Asking ISPs to filter out the porn on file-sharing services - apart from any technical difficulties - is problematic.  If they filter out pornographic material, copyright owners will be legally entitled to ask why they cannot also filter out copyright material.  But if the ISPs do filter copyright material, they will lose their customers.  So the only solution that does not incur severe economic penalties for the ISPs is to leave the file-sharing services unfiltered and unblocked.

The comparison with the filtering of child pornography by the IWF is misguided in at least two ways.  Firstly, the number of child pornography sites on the web is comparatively small, at least in part because most jurisdictions share our views on the subject and will take such material down and prosecute offenders.  Adult and extreme pornography is far more widespread and is therefore a much more difficult target.  Secondly, I suspect but cannot confirm that child pornography is widespread on file-sharing networks, because these appear to be totally unregulated.  I cannot confirm this, not only because I do not wish to go looking for such material, but also because the law makes it clear that private "research" of this type is a crime.  Therefore one must rely on the responsible authorities, i.e. the IWF and CEOP, to keep us informed on the matter; but on the contrary, they would much rather keep quiet about the subject while telling us what a very good job they are doing.

The strong economic incentives in favour of file-sharing are the reason that the legal measures taken to suppress it have not been particularly successful.  As one service is closed down, another arises that uses slightly different methods so that it evades the restrictions imposed by statute or case law.  The Digital Economy Act 2010, yet to be fully implemented, will be no exception.

The only way to prevent file sharing would be for ISPs to stop connecting consumers to the Internet, and instead give them a connection to a "Walled Garden" with proxy servers for the web, email, and a few other services.  Besides being highly unpopular, this measure would stifle innovation, because anyone who invents a new service would need to seek permission from the ISPs, and probably pay them, before consumers would be allowed to connect to the service.  I hope such drastic action will not be taken; but that means we will have to live with file sharing for the foreseeable future.

Our first step in shielding children from internet pornography must be to make sure that we understand the nature of the technology, and its social dimensions - how children are using it, their motives, and those of the other parties involved.

With that in mind, may I suggest a different solution that might satisfy everyone - except those who will have to pay for it, because it will add perhaps £50 to the retail cost of an Internet connection.  While you might not particularly like my suggestion, and I expect there are problems with it that I have not thought of, I hope it will at least illustrate the kind of reasoning that I think is helpful in producing a more workable regime than the one you propose.

Modems could include a small built-in unit, with a screen, smartcard reader, keypad, and a few gigabytes of flash storage, that would be hardwired between the modem connection and the downstream wireless or wired LAN ports, with a mutual lock to the upstream (ISP) connection.  This unit would serve as a parentally controlled filter.  The screen, smartcard reader and keypad would be not unlike those of the Amstrad "E-m@iler" telephone, but the keypad would have fewer buttons, and the unit would need neither an alphabetic keyboard nor telephone functionality.  When a child uses the web on their computer and encounters a page that has been blocked by the ISP, the parent would be able to go to the modem unit, insert their smartcard and type their PIN number, view the page that has been blocked by the ISP, and choose whether to allow their child to view the page on their own computer.  The unit would also function as a proxy server for file sharing networks, and use its flash memory to store any files downloaded by the child, for later inspection and possible release by the parent.  Amstrad has several years' experience at preventing users from unlocking their E-m@ilers (their motive being that the E-m@iler, unlike the unit proposed here, calls a premium rate number every day to pay Amstrad for their service).  This experience is essential to prevent children from cracking the protection mechanisms.  It is essential that the unit is an integral part of the user's internet gateway, and is cryptographically locked to the ISP connection (and vice versa). In contrast, software running on the family PC would be harder for parents to operate, and much easier for children to bypass.  For the avoidance of doubt I have no connection to Amstrad, nor any financial interest in my proposal.

The chief problem with my suggestion is the howls of protest that it will bring down from the film and recording industries: equipment manufacturers would be aiding and abetting file sharing of copyright material by providing the proxy software on their modems, and the government would be condoning it.  This is where the government would have to show leadership, and tell the industry that child protection is more important than their revenue streams.  It might need to amend the Digital Economy Act to exempt the equipment manufacturers from criminal or civil liability for aiding and abetting consumers' file sharing.

Like your proposal, mine requires ISPs to classify web resources, and still suffers from the problem of false classification; but it gives parents, even technologically challenged ones, the power to unlock false-positive content immediately: it will not be necessary to switch off filtering completely, or argue with an obstinate ISP, in order to view a false-positive page.  Unfortunately, no system can avoid the problem of false negatives: some web porn will always get past any classification system.  So we're still in fig-leaf territory, it's just that parents will be able to remove the fig leaves that have been put in places where they should not be.  Children will therefore not have a valid reason to ask their parents to switch off the filtering completely.  Another advantage of my proposal is that it offers effective parental filtering of all content from file-sharing networks: something that is conspicuously absent both from your proposals and from the status quo.  However, it does raise a number of challenging questions.  The first, obviously, is who will pay the extra cost of the modem.  The second is how many parents will be sufficiently concerned to use the system - bearing in mind that many parents equip their young children's bedrooms with a TV set, a computer, and a games console with 18-rated games.

I think this proposal will be more effective than purely ISP-based filtering because it attempts to address the social as well as the technological dimensions of the problem; it tries to be honest about internet content that is outside the web; and it puts parents firmly in charge.  If we really want to protect children by regulating ISPs, then whichever scheme we adopt must have these ingredients, if we want it to succeed.