Filtering Worst Practices: Keyword Filtering and Blocking by File Type

The American Association of School Libraries just reported on the use of filters in schools based on the results of the School Libraries Count! survey conducted January-March, 2012. The results are the predictable mix of good and bad. On the good side (per this report), the filters reduce student distractions and decrease the need for direct supervision.  The filters may even result in "more appropriate" search results.

However, on the other side of the coin, the filters are being used to impede collaboration (chat, social networking, and IM functionality is being filtered) and all things video are evidently suspect. 

While some filters do a decent job of filtering out the content they are configured to block, none of them are perfect.  I consider a decent filter one that has an 85% "accuracy rate." The accuracy is the balance of overblocking and underblocking.  So an 85% accurate filter will correctly block about 85% of what you want it to block and also ensure that no more than 15% of the desireable content is mistakenly blocked.

However, if you block Internet content by file type, all bets are off.  Yes, the filter will block Husky Homemaker's Webcam (because you block the category "Sexually Explicit") but it will also block National Geographic videos and TED Talks (which are categorized as "Educational.")  So blocking Internet sites by file type or functionality is not the same as blocking a certain category of content. The filter just disables everything is sees in that format (e.g. videos) or disables that function (e.g. chat) regardless of what you want to do with it.

That's because keyword filtering doesn't work worth a damn.  

While I try to take a somewhat neutral stand on filtering, keyword filtering is not something I'm neutral on. There is just no place for it in libraries or in school. It simply doesn't work. And for the reason I explained above, using filters to block by file types isn't a good idea either.

Also, filters designed for home use shouldn't be used in schools and, most definitely shouldn't be used in libraries.  These filters tend to err on the overblocking side so even if you are careful about selecting only the fewest categories to block, it is likely that a lot of desireable content will be censored.  In other words, the 85% accuracy doesn't apply.  Instead you'll get something more like 70% accuracy meaning everything you want to block will be blocked but as much as 30% of the content you would expect to get through will not.

And that goes double with keyword blocking.  You may think you are blocking a word but you are blocking the characters of that word wherever they appear regardless of the context. Did you mean to block Charles DICKens?  What about blue-footed BOOBies?  Some filters won't let you even type the evil word in the URL which is really not going to get you the results you were hoping for. 

When it comes to keyword blocking, I have one keyword for you:  NO.

The full report of the survey results are out in December. It will be available at https://www.ala.org/aasl/research/slc.  I'll be interested to see what filtering products the schools are using and what exactly they are choosing in terms of categories to block.  Based on this report, I'm guessing some of the schools are using the wrong filters and they are definitely configuring the filters in a way that, IMHO, creates more (but different) problems than it solves.