call +44 20 7096 1079
July 03, 2014 | Jim Killock

Getting filter categories right

Yesterday's launch of blocked.org.uk has had an excellent response. Over 10,000 sites have been tested for blocks, and no doubt many problems have been found and reported. However, we will need to change the way we calculate results for talkTalk.


When we set up our filtered Internet lines, we did our best to choose the 'normal' set of filters for each line. These are what we believe an 'average' customer who sets up filtering will choose. In TalkTalk's case, it seems we made a mistake, and three categories got added that should not have been. These are "File sharing, Games and social networking". It seems many blogs are categorised as social networking. Additionally, we've been informed by BT that they use light, not moderate, as their default.

To clarify how the 20% figure was produced, we've outlined below the statistics for each network. No individual network blocked more than 13%, so while the overall total representing a 'default' filter block will reduce, it may not reduce so dramatically. We should know in around a day.

We want to show what our experience of their defaults is in the results on blocked.org.uk. We thought that aim would be reasonably representative of what people in general would experience. We recognise that we didn't make that very clear at launch - but are happy to clarify now. When TalkTalk's and BT's results are recalibrated over the next day or so, we will have lower percentages for their 'default' settings. However, all the sites that are included in our current 20% figure are affected by some kind of blocking, even if only a small number of people switch on filters at this level. If and when we test the stricter filter options for all ISPs, the number of sites blocked for customers using stricter options is likely to be even higher. We will need to test for strict blocking, because people have a right to know if a site is blocked and if mistakes are being made.

We have been transparent from the start about the settings with which our lines are configured: they have always been listed in detail in our FAQs. To make this easier to find, we've also added an explanation directly below the results. If we've made any other errors about the 'default' position at an ISP, we are happy to be corrected.

Ideally we would monitor each filtering level for each ISP, and our system is capable of doing so, but our funds don't stretch to paying for all the connections so we had to make a choice. We are working to expand our system so that people can run probes of their own. That way we can cover many more networks, with different filtering settings, to get a clearer picture of the way these filters work in practice. You can help with this.

It's also important to remember that TalkTalk and other providers should be providing an authoritative tool for people to check whether and why a site is blocked, which would also have helped avoid error. After all, we are just testing for blocks, we are not trying to reverse engineer the precise categorisations that are made.

Finally, there must be a question about whether ISPs should be offering to block a category called "social media" that includes content such as blogs. This is censorship by form rather than content. If Tesco run a blog, should that be blocked? Or a blog about Minecraft, or school meals? In fact which blogs are selected as social media at TalkTalk seems to be highly unpredictable. Furthermore, many  communities such as the LGBT community use blogging as a way to talk about issues on their own terms: blocking such information could be disproportionately harmful for young people.

An argument may be made that comments sections could be unpoliced, but plenty of newspapers have comment sections. If a danger is posed by comments, then the target of filtering probably needs a whitelist solution. We think there is some danger in these categories being available for people to apply: they are both too broad for most young users and not sufficiently narrow for young children.

The broader point needs to be remembered: that these filters are arbitrary, capricious, and even the people who sell them don't fully understand what they do.

For the perspective of a Blocked user and website owner on this issue, journalist Jane Fae blogged her story of the status of her blog as blocked or not by TalkTalk.
http://faeinterrupted.wordpress.com/2014/07/03/is-it-because-i-is-trans/

Network

blocked

Total Result

blocked

AAISP

 

100237

0.00%

BT (moderate, not default)

5229

73199

7.14%

O2

455

10059

4.52%

Plusnet

2

41819

0.00%

Sky

4345

66837

6.50%

T-Mobile

1312

23690

5.54%

TalkTalk (with optional categories)

13126

100237

13.09%

Three

597

10017

5.96%

VirginMedia

2953

69137

4.27%

VirginMobile

2430

64202

3.78%

Vodafone

481

9968

4.83%

All networks (current totals)

22837

114179

20.00%

Edits: added Jane Fae link, information about BT and the results table for clarity  

google plusdeliciousdiggfacebookgooglelinkedinstumbleupontwitteremail


Comments (0)



This thread has been closed from taking new comments.