The Censorware Project

Blacklisted by Cyber Patrol

From Ada to Yoyo

A report from The Censorware Project

Pre-emptive Comments

General Censorware Topics

All you discuss are a few dozen items. Is that all? Cyber Patrol has 50,000 entries, you complain about around 50, that means you had a problem with one-tenth of one percent of the whole list, so even you admit Cyber Patrol is 99.9% accurate, right?

Wrong. To begin with, just one item could block tens of thousands of pages, as was the case discussed in Gay Sites, or over a million pages, as was discussed in ISPs.

But this report does not claim to be exhaustive. It was prepared in the spare time of a few volunteers, without any funding at all. If so many problems could be found from such a cursory investigation (though still more in-depth than the typical software review or newspaper article), what other revelations remain to be discovered? The important question is not "Is that all?", but "What else is lurking in there?"

But still, there's a lot of porn on those lists, isn't that so?

Sure. These days, with commercial sex sites trying to do anything to stave off more government action, it's no trick at all to gather up a large list of them. In fact, one of the best ways to find a lot of sex vendors is to search for links to censorware manufacturers. (Just searching on sex keywords will result in turning up many pages on gender and women's studies, reports pro or con on pornography debates, pages with those keywords just to attract attention, and much other such related content.) Oh, note most of those sites probably require credit cards or some form of age certification.

But nobody has ever claimed that those programs didn't have any "true" entries. The blacklists of McCarthyism had real Communists on them too. The issue is what else did the blacklisters slip in, and according to what agendas, once there was a panic and hysteria over their topic.

Then these programs at least will block out porn, even if there is collateral damage to feminist newsgroups, gay sites and so on.

Actually, they won't even do that very well. As summarized in a paper presented at the 1997 annual conference of the Canadian Library Association

In a recent study conducted by Consumer Reports of 22 easy-to-find Web sites that had been judged by investigators to be inappropriate for young children, not one of the four most common software blockers - CyberPatrol, CYBERSitter, Net Nanny, and SurfWatch - blocked all of the sites. Net Nanny failed to block any of the 22 sites, while 14 were blocked by CYBERSitter, 16 by CyberPatrol, and 18 by SurfWatch (Consumer Reports, May 1997, 30). These rates are far below the levels that parents and other consumers have been lead to expect.

The sex-seeking user trying to evade the cybercensor only needs to find one site which has escaped the blacklisters, while the list-makers need to try to keep up with the entire web. But while this battle is going on, a great deal of other content can be swept up in the necessarily obsessive quest to find every last nipple on the net.

I checked a few of the sites you list. They're not blocked! Did you just make all this up?

At press time, December 21st, 1997, using the latest version of Cyber Patrol's database, all the sites listed here are indeed blocked in the categories listed. Double-checking them all was the last thing we did.

If you are reading this more than a day or two after this date, Microsystems Software may have updated the product's database to eliminate the errors. The company is naturally concerned about negative publicity, and we would be surprised if unblocking all the sites we mention is not the first step they take.

Update: indeed, two days after this report was issued, most of the sites we revealed have been unblocked. Surprisingly, many were not. Our aftermath update is available.

The list of errors we provide is not exhaustive, far from it. When they are fixed, Cyber Patrol's quality will improve by a small amount. But the real problem is the system which makes certain that new misblocks will continually be introduced, and existing misblocks -- unless exposed by muckraking activists -- will not be noticed and will remain in place. It bears repeating: the important question is not "Is that all?", but "What else is lurking in there?"

But how do we know you didn't make it up?

We expected someone to ask this. David Burt, of Filtering Facts, has prepared rebuttals to earlier reports on Cyber Patrol and other censorware. His chief response is to point out that, of the list of sites given, most of them are "Not Blocked." (Of course they are no longer blocked, months after the reports were made public. What would we expect?)

Mr. Burt has even suggested that earlier reports are mistaken (or indeed outright lies) because the blocks were removed by the time he got around to checking them, and because he finds all anti-filtering organizations to be untrustworthy by definition. He writes:

Filtering Facts has carefully examined many of the more sensational claims about what filters block, and found that almost without exception these claims have three things in common:

1) The claim is made by someone with a strong bias against filters.
2) The evidence is derived under uncontrolled, unscientific conditions.
3) The claim cannot be independently verified.

To prevent this spurious charge from being leveled at us, we have contacted several independent journalists and pre-arranged for them to download and archive a copy of the product's database. They can confirm our data. For details, please contact Jamie McCarthy.

Isn't it just a matter of time before we solve the problems you describe? If we can put a man on the moon, why can't we create a computer program which can automatically distinguish between pornography and art?

Because the space program was basically a tour de force of engineering, as opposed to a fundamental scientific knowledge breakthrough of Nobel-Prize winning proportions. People argue all the time as to the difference between literature and smut, how can a computer program be expected to know it? Look at how much effort and money it took IBM to create a grandmaster chess-playing program, and how far it is from the fixed and fairly simple rules of chess to language comprehension, and then yet another level up to artistic meaning.

It would be so nice if when censorware-makers made extravagant and hyped claims in their press releases, reviewers displayed some natural intelligence and asked "How is it that you've created advanced artificial intelligence orders of magnitude beyond currently known computer science? Who is the great genius who must now be being lionized by the research community? And why is this giant leap for siliconkind being used only in searching for sex?"

Just as the great and powerful OZ turned out to be only a little man behind a curtain, manipulating people's gullibility, the advanced and breakthrough OZ-XXX usually turns out to be just a group of little censors behind a curtain looking at monitor screens (but still manipulating people's gullibility).


Why does it matter what is blacklisted? The librarian could just override the censorware for that particular site.

To begin with, this statement may not even be true. In the Austin Public Library, because of the way the computers are set up, no overrides are done. Just think back to the last time you tried to make a change to your computer, and the trouble you may have had. A busy librarian may not be able to spend the necessary time fiddling with passwords and controls, it's hard enough to get programs to work in the first place.

Further, despite not having a crystal ball, as sure as the sun will rise tomorrow, one can see there will likely be many instances of this sort of conversation:

"Excuse me, librarian, I need the censorware turned off here."
"Sir, there are children around, would you mind not viewing pornography right now?"
"But I just want to see this anti-censorship site, it's not pornography..."
"Since the porn-blocking program is stopping it, then it must be pornography."

We can all have a laugh at such an exchange, but don't discount it. There's a common impression, fueled by uncritical and fawning articles, that being blacklisted by censorware is the same as being pornography or obscene or generally evil.

In this vein, it's important to note that blacklisted sites do not generate a message saying "This page might be a gay-rights resource, or discuss breast cancer, or be an anti-censorship collection; disable our program in order to find out". Cyber Patrol's message, for example, is the company's Internet-police-badge symbol combined with an obscure and forbidding phrase about a "CyberLIST checkpoint!". It certainly gives the impression that the page is guilty of something, even if it's not clear exactly what that is. The reader doesn't know beforehand if the site is a commercial sex vendor or simply ran afoul of some bigot in Cyber Patrol's back room.

Besides, the software only allows 64 overrides. From the help pages of Cyber Patrol 4.0, educational version:

Rules for the lists
You can enter up to 64 sites in the Approved field and 64 sites in the Restrictions field.
Each site address can include up to 64 characters.

This is not even enough to make exceptions for the errors found in this report.

Well, maybe the librarian could go over the blacklist beforehand.

First, almost all censorware makers, including Cyber Patrol, keep their blacklists a secret.

But even if they were open, who would have time to go though all of it? Cyber Patrol claims more than 50,000 entries. At 1 minute per entry, 8 hours a day, a person could check approximately 500 items a day. That means 100 working days of nothing but going through the whole list. And the blacklist changes all the time, so some of the work would have to be redone. This is the other side of the problem of creating the list in the first place, anything involving looking at a significant fraction of the World-Wide-Web is simply a task of enormous size.

Does a library have to subscribe to every book and magazine? These programs represent selection, not censorship.

The basic different between these two cases is that for print, it takes resources to acquire material, but with censorware, resources are devoted to banning material. This is very clear if one thinks of a budget statement, where it will have the costs of all of the print books and magazines for the patrons to read, and then a fee to the blacklister to prevent the reading of a set of electronic books and magazines. It's as if someone was hired to rip sections of books and blot out portions of magazines, with the justification of "Those sections and articles are not being censored, they are simply not being selected."

The effort to cover over this basic difference can become quite elaborate, and this document won't examine every permutation. For example, there have been elaborate arguments as to how a "single work" shouldn't be altered, but it's permissible not to have works as a whole, and that's the library justification for censorware. Briefly, the reply here is then that in this framework the relevant single item is the connection to the Internet. All of this is an attempted distraction from the core point, that it is a blatant violation of library principles to buy into such efforts to prevent people from reading material based on some secret blacklist.

back to top
next: conclusion