Introduction – John Felton, NLC. One of the opportunities available to assist libraries in supporting their broadband connectivity costs is applying for E-rate discounts on internet service. This necessitates complying with the requirements of the Children’s Internet Protection Act (CIPA) by filtering internet access. In this session we will hear about what CIPA actually requires, about the issues a library director must deal with when deciding to filter internet access, about how filtering can be accomplished with low cost, and about the results of a survey completed by Nebraska libraries that have installed a filter.
CIPA: Myths vs. Reality – Christa Burns, NLC, will clear up some of the misconceptions about CIPA and how it relates to E-rate.
Filtering Internet Content – Pam Soreide, Holdrege Area Public Library, & George Matzen, Webermeier Memorial Library. Participants in this session will leave with a sense of what factors should be considered in the decision whether or not to filter Internet content at public workstations. Discussion will include patron perceptions and management issues.
DevoxxFR 2024 Reproducible Builds with Apache Maven
Fear of Filtering: The Reality of Internet Content Management (Soreide DOC)
1. Internet Filtering In Libraries
Pam Soreide
Technology Planning Summer Camp
8/22-23/2011
History of Internet Filtering in Libraries
Libraries started filtering Internet access in the late 90s under pressure from
local community groups
With the passage of the Children’s Internet Protection Act (CIPA) in 2004, all
libraries accepting federal funds from E-Rate, ESEA or LSTA for computers
that access the Internet or funds used to pay for Internet access were required
to adopt an Internet Safety Policy, hold at least one public meeting on the
policy, and certify that they have adopted and implemented a Internet Safety
Policy, which must include blocking or filtering software.
CIPA states that the software must protect against visual depictions described
as obscene, child pornography, or any content harmful to minors in any way.
It does not require filtering or blocking on text, though since most of the
products were developed before CIPA passage, almost all do block on text,
and unfortunately, do not do a very good job at blocking images. CIPA does
not speak to any need to filter content delivered to laptops brought into the
library by patrons or staff, but in practice, it has come to be seen as optional.
CIPA also states that libraries may disable filtering for adult patrons 17 years
or older upon request, for purposes of lawful research.
Another act, NCIPA, specifies elements which must be present in the library’s
Internet Safety Policy, which is also required.
Why Filter?
In public libraries that filter, librarians talked about being responsive to
community demands:
o "Our patrons are happy that we have this type of filtering. As one
said: 'We discourage indecent exposure. Why allow it on the screen in
a public place?' Our job is to give the public what it wants, not what it
doesn't want."
o "Politically it was the right move for our community, which is
conservative. It makes us responsive to the requests/attitudes of our
users."
School and public librarians who are against filters based their opinions on a
belief in intellectual freedom and the inefficiency of filters. Those librarians
who oppose filters but used them nonetheless talked about their desire to
remain employed. Here are some of their remarks:
o "Filtering tries to enforce morality externally. We should be teaching
internalized morality."
o "I do lots of research on books and curriculum areas for teachers. I
am only 50 percent as effective with this insane censorship [caused by
2. filters]. If kids cannot be responsible now, when will we expect them to
be responsible? When they retire?"
o "Internet filtering lets parents and others 'think' that the inappropriate
sites are blocked. In reality, not all of them are--it's impossible. It's
mainly symbolic!"
o "Intellectually I object to it; conversely, I like my job.“
“Filtered or Unfiltered” by Ann Curry, Ken Haycock, School Library Journal
1/1/2001
What are the Options?
Client side filters
o Software on each workstation. Can be customized, but only disabled
by a person with the password
Content-limited ISPs
o Primarily used to provide Internet access suitable for children, this
option provides access to only a portion of the content of the Internet
judged to be safe and appropriate for children
Server-side filters
o Very popular for institutional settings such as schools or libraries.
Filtering can be customized and different profiles used for children or
adults
Search engine filters
o Some search engines such as Google offer an option to turn on a safety
filter to limit results. It does not preclude a user from typing in a direct
url to a problematic site. Others, such as Yahoo! offer a child-oriented
version of their product that searches only child-friendly sites
Filtering vs. Blocking
Filtering software denies access to a website based on its content
Blocking software denies access to a website url based on the offending site’s
url
A weakness of blocking software is that each offending url has to be coded
into the software and it may be hard to stay abreast of new sites coming online
every day
A weakness of filtering software is that it bases its judgments of the presence
of particular keywords. It is obvious that many useful resources may be
blocked because of a reference to a particular term.
Another problematic aspect of filtering software is the visual nature of
pornography. If offending keywords are also included, the site may be
blocked, but if the keywords are in a different language (or the search is
conducted in another language), it may not be.
Filtering software is still quite undeveloped in its coverage of race hatred,
violence or recreational drug use
3. How does Filtering Work?
Two basic types, network-based and stand-alone.
Filtering by URL
o Products that filter based on urls will use a search engine such as
Google to run searches for trigger words (live sex chat room).
o The list of results is then reduced by taking out those representing
educational and government sites (those with .edu and .gov
extensions).
o Of the remaining sites, the top 100-500 are blacklisted, sometimes
with spotchecks by humans and sometimes not.
o When the filtering program is in use, each set of search results is
scanned against the list before items are displayed.
Filtering by keyword
o Products that filter based on content analyze web pages as they are
requested by the searcher, looking for keywords and phrases, and
sometimes other factors such as banner ads, number of links and
number of images. An AI program then uses a substantive formula
with a set of criteria to classify the pages as either allowed or blocked.
Filtering by File Type
o A minority of products allow you to block particular file types, such as
video files (.avi), audio (.mp3) or still images (.jpg). This is less than
useful since many porn sites will embed their files in flash or pdf
wrappers, thereby getting around the block, and it will block all .jpg
images not just prurient ones.
Blocking What the User Sees
o Using one or both of these approaches, companies build up lists of urls
and/or keywords that are then blocked. Depending on the product,
patrons will see messages about what was blocked and why. Some
products filter out only the triggering content but letting the rest of the
content display while others will go further and block the whole page.
Another differentiation is that some will allow you to see other
content, but hiding triggering content on the display page (such as ads)
whereas others let you see the triggering content on the search results
page, but will not allow you to click through.
One of the challenges is to accurately understand your products classification
methodology, but they consider it a ferociously guarded trade secret and will
not even give examples of the effect of their blocking algorithms on particular
categories.
Libraries and others have conducted tests of different products, with the
results summarized below. In broad terms, most researchers agree that content
is overfiltered by about 15%, underfiltered by about 15% with images and
foreign materials being the worst filtering only about 40% of the explicit
materials.
4. Date Title Source Summarized Conclusions
2008 Expert Report Dr. Paul Resnick (for North Central • 93.1% accuracy in
Regional Library District) blocking websites
• 48% accuracy in blocking
images
2007 Report on the Bennet Haselton (for the ACLU) • 88.1 % overall accuracy
Accuracy on .com sites
Rate of • 76.4% overall accuracy
FortiGuard on .org site
2006 Expert Report Philip B. Stark (for the DOJ) • 87.2%-98.6% accuracy
blocking "sexually explicit
materials"
• 67.2%-87.1% accuracy
allowing "nonsexually
explicit materials”
2006 Websense: Veritest (for Websense) • WebSense: 85% overall
Filtering accuracy
Effectiveness • SmartFilter: 68% overall
Study accuracy
• SurfControl: 74% overall
accuracy
Pros
Fewer reported incidents of publicly viewed sites which make others in
adjacent spaces uncomfortable or feel threatened
Fewer reported incidents of prohibited behaviors
Reduced risk of having minors accidentally encounter content that they may
be ill-equipped to deal with
Using a filter places the library squarely within the expectations of the
community for a safe, unthreatening, family-friendly public space
Can protect the library against suits by staff such as the one in Minneapolis,
Minnesota brought by 12 library workers claiming that unwanted exposure to
patrons viewing pornography constituted a “hostile work environment,” a
position agreed with by the EEOC in 2001. Cost for settlement was $435K.
Filters with the ability to block against peer-to-peer file sharing can protect
themselves against being found complicit in illegal behavior
Use of a filter arguably serves the same purpose as your collection
development policy, only in the context of digital content
Especially in rural areas with small libraries and budgets to match, the
discount on Internet access and telephone service may be worth going after.
5. Cons
As articulated by the ALA, libraries are fierce defenders of the right of all
citizens to information of their choice without censorship. Use of filtering or
blocking software runs counter to the core values – intellectual freedom and
equity of access – promoted by libraries.
Filtering products need to be actively understood and managed so as not to
over-filter content or assume that filtering is 100% accurate and reliable.
Additionally, it is widely agreed that it is difficult if not impossible to
adequately filter for images as required by the law.
Underblocking can be an issue, either due to the rapid availability of new
content or if licenses are allowed to expire
Use of filtering software weakens the librarian’s role of selecting content
appropriate to the community, forcing reliance on the service provider who
may make arbitrary decisions on keywords to filter without consideration of
context.
Even though patrons over 17 may request unfiltered access, there is no
consideration given to their possible reluctance to do so, fearing the
disapproval of the librarian or requirement to cite reasons for that access.
While a content filter should never be considered a replacement for anti-virus
software, most of them do keep lists of sites known to install malware and
allow you to filter them.
The Supreme Court ruled in 2003 that filtering is more akin to a selection
decision than it is a de-selection decision which requires more scrutiny.
Justice Souter dissented, arguing that CIPA was burdensome to the
intellectual freedom and First Amendment rights of adults, and amounted to
censorship in selection.
Blocking software is by nature reactive and therefore perpetually out of date.
Filtering software is insufficiently developed to be able to distinguish between
offensive content and sensitive content, or between mature and obscene.
The equitable access question is illustrated by the need for local configuration
of filtering software. If one library filters only pornography and another
additionally filters on nudism or violence or weapons, there can be significant
discrepancies in the content provided, thereby infringing on rights of
intellectual freedom for those who only have access through the public library.
As an example of the sometimes heavy-handed effect of filters, I had to
unblock a large number of domains in the process of researching this project.
Perceptions
Many people are deeply suspicious of government tactics such as passage of
CIPA, or may willfully misunderstand the intent
Use of filtering software in library environments may promote a false sense of
security in some parents and be viewed as an opportunity to disengage from
their children’s online experience, thereby abdicating their responsibility for
supervision to library workers
6. While we as librarians may know our limits, the public perceives libraries as a
“safe” place that they can send their children with confidence. We do not want
to betray that trust, either with the parents or with the children.
Personal Experience
Coming into a public library setting from a corporate environment, I had no
experience with filtering or blocking software.
Cyberpatrol had been loaded on the four public access workstations. I found it
difficult to manage properly, even with that small number of installations.
Staff workstations were not controlled.
When I started working with a local ESU for tech support, they suggested
using a filter that also functions as a firewall called Clark Connect (now
known as ClearOS). This is a Linux-based server solution used in the school
districts.
Lessons learned: even though this application is running in the background
and might seem to be able to run on an extra desktop, it is actually a critical
application and needs reliable hardware. Unless you can disarm it before
taking the computer offline, there will be NO Internet access in the building
until you bring it back online again. A second workstation holding the
software that you can apply the backup profile to in case of need is a wise
precaution. Also, make sure that whatever system you are using is registered
after each renewal.
If you use a server-based filter, be sure to back up your profile.
It is quite flexible, allowing you to set the level of enforcement of filtering as
well as the specific content you want to filter and/or whitelist. Managing those
lists is easy to do.
The server solution allows filtering of all workstations in the facility or within
reach of its signal, including laptops brought into the library.
Conclusion
While I have had many requests to whitelist particular domains, I have never had a
request to make an unfiltered workstation available. People understand that the public
access computers are locked down in a number of ways, but we do our best to be sure
that patrons go away with the information they need.
Resources
www.holdregelibrary/delicious_links.html
See internet.safety