Delivering Libellous Content

The DeliveryDemon had to chuckle at this article http://www.pressgazette.co.uk/content/dont-let-internet-linked-stories-land-you-libel-writ

The law has certainly been working hard to catch up with technology, and the impact of this sort of libel is very real to those who are libelled. But the legal profession is missing a trick here. Behind the scenes, there is technology which looks for keywords and tries to interpret them. By and large this software is still remarkably primitive. It has yet to get to grips with the ability to interpret the context. Basically it lacks ‘intelligence’. It is designed to provide an answer at the expense of providing a sensible answer.

Google predictive text gives some good examples of what can happen http://www.telegraph.co.uk/technology/google/6161567/The-20-funniest-suggestions-from-Google-Suggest.html and various mobile phone predictive text engines can be even funnier. The automated parsers used by recruiters cannot distinguish between Coral the bookmaker and Coral the programming language. Amazon’s ‘you might like’ suggestions suggest you buy an identical item to a recent purchase, with a different brand name.

To some extent, many of these tools are designed to depend on data which is not quality-controlled in any effective way. Certainly an Amazon vendor will enter the keywords likely to maximise search hits. that can mean the entry of keywords with little relation to the product being sold.

Google is one of the more sophisticated players since its product depends on understanding what a searcher is likely to want, but the Telegraph article shows how primitive the logic is. Asking users to log in and relating searches to their search history has the potential to improve search result quality, but people are becoming increasingly sensitive to the amount of their data held by large corporations, and legislators are starting to respond to those concerns, so relying on users logging in may not be the most fruitful development path for this type of tool.

The examples in libel article certainly have merit. Either the tool is not fit for purpose, or it is being used unintelligently. A fairly obvious solution would be for the news website to flag articles as being either positive or adverse, provided the tool refrains from coming up with links to ‘similar’ articles unless they were also flagged as adverse. If the tool can do this, the web publisher is at fault. If the tool can’t do it, then there are two potential breaches. The tool may be inadequate for the purpose for which it is being sold. Or the web publisher may be making inappropriate use of the tool. Of course, when a payment model is based on click throughs, the incentives tend not to favour anything which limits the number of links displayed.

A fruitful approach for legislators would be to look beyond individual libels and examine the capabilities of current tools, and the processes which web publishers use to to mitigate the risks arising from tool limitations.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: