February 13, 2004
Facts about Filters
Sometimes when people talk about the problems with using filters in libraries they get some of the facts wrong or don't quite tell the whole story. I think it is because they want to make filters sound as perfectly awful as possible.
The fact is, there are very good reasons against the idea that libraries must filter. It's not necessary to exaggerate the negative aspects of filters to make the point that there are significant problems implementing them in a library setting.
That said, there are some scenarios where filters can be put to good use. The real trick is to understand how they work and use them appropriately -- not as the way to protect children from all the dangers they might encounter on the Internet (which CIPA implies can be done), but possibly as one tool.
The real problem with CIPA is that it mandates libraries use filters instead of allowing libraries to address the issues in the way that makes sense for their communities. And the fact that ALL computers must be filtered -- how is filtering a staff computer going to protect children from obscenity again?
Anyway, here are some facts to consider:
- There is no CIPA-compliant filter--by choosing any category to block, the library is blocking more content than would be legally defined as obscenity, child pornography or "harmful to minors."
- Many filter companies allow the administrator to see the URLs in each category -- its only the most well-known ones that don't.
- No filter is going to be 100% effective at preventing access to sexually explicit content, especially when the end user is trying hard to find it and/or is a even a little bit Internet savvy.
- Filters can prevent children from using online email, peer-to-peer file sharing, and downloading -- all of which are ways that pornography is distributed -- but the filters can't distinguish between when the file being shared is a homework assignment or a collection of obscene pictures.
- Some filters can block the visual depictions (images) on a page, and only do so for certain selected categories of content. So, for example, if the filter had a content category called "CIPA" one could choose to block images only in the CIPA category. The problem is that no filters have a CIPA category.
- Filtering staff computers isn't going to help protect children from accessing inappropriate content. And it just makes it more difficult for staff to monitor the accuracy of their filter and it costs the library more in "per-seat" licenses. This is criminal!
- There's no pracical difference between the requirement that an authorized representative can disable a filter for an adult versus that same adult disabling the filter for themselves. Therefore, what's wrong with mandating that each library offer some percentage of filtered computers to their patrons to ensure that adults always have the choice to browse filtered or unfiltered. Children should be able to browse unfiltered too -- if their parents have authorized them to do so. Giving patrons options eliminates the requirement that librarians get involved in turning off filters. First Amendment freedoms would be alive and well at the unfiltered Internet computers.
- No library has to buy an Internet filter that doesn't allow them to view the URLs being blocked. They can use open source filters such as Squidguard and Dan's Guardian -- which are free -- and maintain their own block list. Such lists can be created from lists publically available such as URLblacklist.com or other libraries using these same products.
- Maintaining your own block list of sites your patrons have accessed which you (or a designated representive) think are illegal is no more troublesome than installing an expensive Windows server with an expensive filter on it and worrying about maintaining the server, dealing with the issues associated with disabling the filter on request and correcting the overblocks. By taking responsibility for your own block list, you can virtually eliminate the need to unblock erroneously blocked pages, and ensure that any illegal sites being accessed by your patrons are blocked within a day a two by regularly analyzing the server's logs. Both approaches demand work if they are to be implemented responsibly. It's just a matter of deciding where to put your time and effort.
Posted by Lori at 11:35 AM | Permalink
Comments
Lori, I want to post a number of things regarding this post.
First, it is good to see you dealing directly with the fact there are differences, significant differences, between various filtering options and products.
Second, it is important to realize that the mechanism for CIPA compliance involves, but does not rely upon, having a filter.
Essentially, to continue to recieve E-rate funds CIPA requires that a library or library system self certify its compliance with the provisions of CIPA. The FCC, which is tasked with determining compliance, has stated that it is looking for a "good faith" effort. Thus the essential question a library must answer in the self certification process is whether or not it's internet use policy, in combination with the filtering technology it has implemented, constitutes a good faith attempt to comply with CIPA.
The requirements of "good faith" are flexible, but at a minimum a library should be satisfied its CIPA compliance strategy uses up to date technology, respects the first ammendment rights of its adult patrons and is well communicated to its patrons.
More later.
Posted by: Jay Currie | February 17, 2004 10:35 AM
"There's no pracical difference between the requirement that an authorized representative can disable a filter for an adult versus that same adult disabling the filter for themselves. Therefore, what's wrong with mandating that each library offer some percentage of filtered computers to their patrons to ensure that adults always have the choice to browse filtered or unfiltered."
The first part of this is absolutely true. The Supreme Court was express in stating that the solution for adult users was to turn the filter off, no questions asked. From which it follows that cutting the librarian out of the loop makes sense. (Which is exactly why IF2K has the "click through" feature which lets a patron make their own decision about whether they wish to view the content which has been blocked.)
Unfortunately, the problem with a sensible mix of filtered and unfiltered computers is that it does not comply with CIPA. Yes, that is goofy. But so is much of CIPA. To certify compliance a library has to have filtering on all computers. Otherwise it is not complying.
A reasonable approach might well be to filter all computers and have the click through option disabled on computers in the children's area. That would comply.
More later.
Posted by: Jay Currie | February 18, 2004 11:40 AM
Hi Lori!
You write "So, for example, if the filter had a content category called "CIPA" one could choose to block images only in the CIPA category. The problem is that no filters have a CIPA category."
Actually, you'd need CIPA-ADULT and CIPA-CHILD
categories so as not to overly restrict Adult access
You write "Filtering staff computers isn't going to help protect children from accessing inappropriate content. And it just makes it more difficult for staff to monitor the accuracy of their filter and it costs the library more in "per-seat" licenses. This is criminal!"
I think its reasonable to use the Internet Explorer CONTENT ADVISOR as a TPM to get around that problem
"Therefore, what's wrong with mandating that each library offer some percentage of filtered computers to their patrons to ensure that adults always have the choice to browse filtered or unfiltered."
I agree with Jay - your logic is sound, but the law requires TPMs on all computers
"Children should be able to browse unfiltered too -- if their parents have authorized them to do so"
But of course they can't in an e-rate library ... unless the parent is WITH them and on gets the filter disabled on the parent's own behalf. LSTA-CIPA allows disabling for any age.
"No library has to buy an Internet filter that doesn't allow them to view the URLs being blocked. They can use open source filters such as Squidguard and Dan's Guardian -- which are free -- "
Good point
a question:
"Filters can prevent children from using online email"
is that true for web-based email too? Even if you block hotmail.com, aren't there zillions of other alternatives that you'd have to identify to block?
-Mary
Posted by: Mary Minow | February 28, 2004 1:36 PM
Hi Mary
I wish to address 3 points you raise
1.Squid is free and available
2.There are no CIPA lists
3.Staff computers must be filtered.
1. Squid is not free. Squid licensing is free but there are some problems.
It only runs on Linux. Probably not the operating system of choice for most libraries.
However it might be do-able for a large system such as Kanguard. However, there are two problems with this.
a. They had three engineers to set up the system and three librarians to go over the lists.
Obviously this was not simple. Too complex and expensive for an average library.
b. A ?good? modern filter has 3 components.
i a blacklist (sites that must always be blocked)
ii a whitelist (sites which must never be blocked and will over ride the blacklist
iii a phrase filter
Squid has only one of these features ...the ability to process a blacklist.
2.We are beta testing our new list called ?CIPA f?or the reasons you and Lori point out.
Strictly sexual in content designed to conform to the ?requirements? Also it is a
cosmetic title. Probably, such a list can only be provided by the legislators of the act.
The problem is computer cannot determine what a ?good list is and people do not
agree upon what constitutes ? the best list? ... so I guesses you have to accept the
best of what is ?out there? or create your own. Or have the CIPA mandators supply
one.
3.I also can't understand why staff computers are in this mix. I recommend you ask any filter company you might be doing business to exempt staff computers from licensing charges if they want to be unfiltered. We certainly would make such an exemption upon request. Presumably staff computers are only used by adults, so if you have the right filter you could just leave the ?click through option permanently enabled.?.
Consequently, the staff computers being filtered is not a problem it is just a cost issue.
Posted by: Bob Turner | February 29, 2004 12:26 PM
Mary,
In terms of blocking web-based mail, some of the filter companies include that as a category so you can, theoretically, block web-based email just as effectively as you can block pornography! Well, actually, more effectively because it is less subjective.
Thanks for the clarification on the other issues. Just to be clear, part of my entry is about what makes sense not what CIPA mandates. I'm trying out my lobbying legs...and putting out there what makes more sense to me. Just in case the CIPA gods are listening.
And Bob, I've heard you say before that a "phrase filter" was essential to any filter. And I'm not sure I buy this argument. The danger of overblocking when using a phrase filter makes me more comfortable using a URL list of sites I personally can verify should be blocked. In Lori's Library, we wouldn't be using a phrase filter. We'd be using our own list of blocked sites and we'd be monitoring our logs to see if we should be adding any new URLs to our list based on the activity in our library. I'm more worried about overblocking than underblocking. I think a library can be very effective with this approach without having to spend too much time on it. The time I spend analyzing the logs (perhaps that's where a phrase-based analysis might assist me) will be saved in the number of times I WON'T be asked to unblock a site.
Lori
Posted by: Lori | March 3, 2004 8:24 AM
Hi Lori
I may have mentioned a phrase filter before. :)
I arrive at the conclusion by working with this for a long time.
Here is the situation as I have experienced having physically done what you are considering..
ALL attempts to filter will produce over blocking and under blocking. That is a given. That will NOT change no matter how you approach it.
The key blocking cultural elements of a filter are
a blacklist...these must always be blocked
a whitelist....these must never be blocked
and a phrase filter....
also the adults must have a click through capability ...this can be achieved by leaving the "click through" on or having the librarian jump up and down...in the end it is the same the difference is how much labor intensive do you want it to be...the adults have a right to unfettered access.
You can use all of these elements or some of these elements or none of these elements....
that would depend upon your goal...
those are the elements....
It strikes me that sifting through the logs each night to update the blacklist is beyond the capacity of an individual library..it could only be achieved by a governing body...and there does not appear to be such a governing body...and if there was such a body I bet it wouldn't take them to long to conclude "what we need is a phrase filter" in this chain..
I am aware that this last paragraph is a matter of personal opinion on a hypothetical non-existent situation.
Posted by: Bob Turner | March 9, 2004 10:09 AM
Actually, since no filters offer CIPA-Adult and CIPA-Minor, the black-list only would serve a useful purpose.
Use a black list based solely on child porn and obscenity (though of course Lori's right, in an earlier posting, that you risk going to jail by researching child porn to create such a filter). Install that as the option for adults. No phrases, keywords etc., and majorly minimizes overblocking without affecting underblocking significantly.
Mary Minow
Posted by: Mary Minow | March 14, 2004 6:44 AM
Hi Mary
Yes ,you are correct you could use only one of the three elements of a filter. This would be a simple and poor filter because it is missing two elements but you could do that if you wanted to.
However, you are totally missing the point regarding adult filtering. You cannot filter an
adult ...the filtering is simply a "warning"...the adult has a right to request a "click through". If a library does not want to leave the "click through option" on then the librarian can "jump up and down" and manually allow the access ...the result is the same...bearing this in mind then nonblocking becomes less of a problem than under blocking ...
And this would be a "good thing" if protecting intelectual freedom was a goal.
If you think Lori is right about "possibility of going to jail for researching a child porn list
"though of course Lori's right, in an earlier posting, that you risk going to jail by researching child porn to create such a filter"
quotes mine
Given that you think "now of course" Lori is right now you might understand why a phrase filter is a usefully element...it would keep you out of jail (for list compilation)...the phrase filter can be configured ...it can be turned off...or it could be implemented...those are the options.
fyi...in nine years I have never heard of a filter company having an issue with list compilation and jail ...we have an understanding with the police that any child porn is submitted directly to then and presumably through them it goes to the global police network that addresses child porn issues (in fact I do not know what they do with the info)... This fear you speak about I think I know who started this myth and it probably is a comment on the political times we live in more than a reality.
Such is the power of myth.
Btw...at the time of this writing sandwiched between your comment and mine are two hard core porn spams porn advertisements. This will have the effect of triggering probbaly every filter on the market if not the whole blog at least aspects of the blog depending on the filter.
....hmmnnnn
Now there is a subject for Lori to blog about. What do you do about that?
Posted by: Bob Turner | March 16, 2004 12:17 PM
Hi Mary
corrction
"bearing this in mind then nonblocking becomes less of a problem than under blocking ... "
should read
bearing this in mind then over blocking becomes less of a problem than under blocking ...
Posted by: Bob Turner | March 16, 2004 1:32 PM
I wish it were a myth - and I'm glad you have an understanding with the police. I mean common sense dictates that to screen out child porn, you're gonna see some.
The law does not actually give that as an exception, however. A number of law review articles focus on the lack of any exceptions and how this hampers research that could be done to PREVENT and TREAT child abuse/pedophilia.
For example, see this footnote in my forthcoming article for First Monday:
Lawrence Matthews [convicted of possessing child porn] had previously produced a radio series on the availability of child pornography via the internet, and said he was investigating pornography for purposes of creating another report. A federal appellate court upheld his 18 month sentence. "The reporter admits that he traded in the pornography but maintains that he did so only to research a news story. He contends that when such acts are committed solely for a valid journalistic purpose, the First Amendment provides a defense to criminal conviction, and he appeals the districts court's refusal to permit him to present this defense to a jury. Because we conclude that the First Amendment provides no defense in these circumstances, and because we reject the reporter's other arguments, we affirm." United States v. Matthews, 209 F.3d 338, 339 (4th Cir. 2000), cert. denied, 531 U.S. 910 (2000). For arguments that researchers should be able to view child pornography, within limitations, see Clay Calvert and Kelly Lyon, Reporting on Child Pornography: A First Amendment Defense for Viewing Illegal Images? Kentucky Law Journal Fall 2000 / 2001 89 Ky. L.J. 13 and Amy Adler, The Perverse Law of Child Pornography, The Columbia Law Review March, 2001 at http://www.ipce.info/library_3/files/adler.htm
I'm the first to support a law carving out an exception for filter designers...Researchers are tougher, but they should at least have their day to defend themselves in court. The court doesn't allow a research purpose. It's just not there.
On the cusp, there is at least one case that considers a literary purpose as a remote possibility - but the defendant still lost
US v Bunnell (Dist Ct - Maine) Crim. No. 02-13-B-S
It's good you've checked in with the police - and again, it sounds reasonable that they would believe you and not make any arrests.
I should point out that a Milpitas Calif couple checked out their own porn video business with local police and got a green light. They were then prosecuted in Tennessee and sentenced to a long prison term there.
I do see a distinction in what filter screeners are doing, of course. It's just no slam-dunk and we're talking about very treacherous waters.
Mary
Mary
Posted by: Mary Minow | March 17, 2004 2:01 AM