Many people are alarmed that the London Public Library is even considering putting Internet filters on its adult computers! This letter from Gavin Stairs follows those from Dr. Sam Trosow and Dr. Toni Samek. Please refer to our previous post “What You Can Do to Stop LPL Internet Filtering” and read our first post for a brief introduction to the issue. Please send letters and e-mails to the LPL Board to show your support for intellectual freedom! Their meeting is June 20th at 5:30 pm in the Friends of the London Public Libary Board Room at the Central branch (251 Dundas).
-SIO
*****
Dear Board Members of the London Public Library:
I was recently informed by Dr. Samuel Trosow of your impending deliberations on filtering of public access terminals (computers) at the LPL, to counter the availability of offensive material. Dr. Trosow was kind enough to send me a copy of his letter to you regarding this issue. I am in complete agreement with his position as stated in that letter, and immediately sent him a note to that effect, which he suggested that I send to you. Here it is, verbatim.
I am Gavin Stairs, a resident of London North Centre and an active patron of the LPL, especially the Masonville Branch. I also serve as the publisher of Pendas Productions, a poetry press in London. In my work I make heavy use of computers and the Internet, and in my previous careers, I was both a designer of computer systems and a user in a technical context as the Design Engineer of the University of Toronto Physics Department High Energy Physics Group for a decade and a half until 2000. I would classify myself as an expert user of computing equipment.
It is my experience that filters of the type used to restrict access to textual and other data on the Internet are susceptible to both false positive and false negative bias at a high rate, meaning that they tend both to exclude inoffensive material and include offensive material outside their intended functions. Where there is a strong desire to protect users, such as children, this may be acceptable when the filters are tuned to exclude as much as possible without regard to unintended exclusions of inoffensive material. However, it is much less acceptable in the context of adult use, where utility may be defined in terms of inclusivity of all relevant material in a given search, regardless of what unintended material may have to be winnowed by inspection. This is particularly true in the case of artists and sociologists who may be directly concerned with the otherwise offensive material itself, for purposes of legitimate study.
If a personal context is useful, let the reader inspect the reject and inbox files in his or her own email account to observe how inefficient the filter is at catching all the junk, while passing all the intended mail. In an account like mine, where the email address is widely known because it has been available for many years from various sources on the Internet, I receive a large volume of mail, not more than 10% of which is valuable, while about as much junk fails to get filtered, and a percent or so of good messages have to be retrieved from the junk mailbox. This is after a long period during which the filters have been tuned by training with real messages.
The filtering rate on images and audio is even less impressive. Much of the failures in the email message filtering reported above comes from the practice of including large amounts of masking text and the use of images to convey text, where the message is effectively masked from the filter. Such techniques are also available to web sites.
Regardless of the technical efficiency of textual and other filters, they will always exclude a certain number of websites and messages because of their legitimate use of objectionable language, as any survey of literature will show. The same might be said of non-textual material. This is an intractable problem at the current level of sophistication of such tools, and one may speculate that it will remain so for a long time.
Unless filters in use on a system like the LPL are constantly monitored for these failures, and continually tuned for new threats and overly aggressive filtering, they are highly likely to interfere with the use of library facilities by serious researchers. I therefore deplore the blind application of such devices to all access computers in the library. If there is a demonstrated need for such filtered terminals, let them be so labelled and made available separately from unfiltered terminals, which should be made available to any adult library patron desiring an unbiased research capability. Moreover, there must be an informative effort made to distinguish these two, so that patrons are not misled.
It is my opinion that a simplistic approach to such filtering is inappropriate to any library in which any form of research is to be done. While some of the branch libraries may be used predominantly by recreational readers and families including small children, they also have a certain number of patrons whose purpose is likely to suffer to some degree from such bias. It would be an unfortunate restriction on their use of library facilities to force them to resort to the Central facilities, or to exclude them entirely.
I therefore strongly support Dr. Trosow’s position in this matter, and urge the library to take a deeper look at their policies regarding filters on public access terminals.
Sincerely,
Gavin Stairs
Publisher, Design and Production Manager
Pendas Productions