Keep It Legal To Embarrass Big Companies

Maybe Peacefire's timing is bad. Two courts have recently said that the reverse-engineered DeCSS program is illegal to publish in the United States, and UCITA gets closer every second. Yet Peacefire today released a program that reverse-engineers the encryption on a list of sites blocked by a major censorware product. Maybe T-shirts that say 'X-Stop has a 68% error rate for blocking student homepages' will get classified as munitions next. Bennett Haselton shares his thoughts (below) on corporate crypto.

Bennett Haselton is the founder and head of Peacefire, an activist group to support the free-speech rights of young people. He suggests that you might want to download the X-Stop "smoking gun" evidence (4MB) before the company has a chance to remove it from their server.

The feature below was written by Mr.Haselton.

X-Stop is an Internet censoring program with an encrypted database of 370,000 URL's blocked under various categories: Sex, Drugs, Rock `n' Roll, etc. Their competitors like SurfWatch and Cyber Patrol also do not publish their blocked site lists; the officially given reason is to keep kids from using the lists to find smut on the Internet. This is silly, given how easy it is to find Internet porn without the aid of X-Stop's secret database (although if you still want to, you can download our codebreaker, follow the instructions to get the X-Stop list and decrypt it, and help yourself). But for the next part of our report, after we decoded the URL list, we looked at the first 50 URL's in the .edu domain that were still valid, and found that 34 of them were regular student home pages with nothing offensive (hence the "68% error rate" t-shirt slogan). None of those 34 students who responded to our e-mails could think of why X-Stop would want to block their pages.

X-Stop admits on their Web site that their database is put together by a Web spider called "Mudcrawler" and not by human reviewers, but even for a machine, a 68% error rate is pretty bad. And even though the real reason why these lists are encrypted is obviously to keep competitors from stealing them, this also makes it much harder for third parties to find out what the programs really block. In fact, X-Stop had once claimed that every URL on their list was reviewed by a human before getting blocked, but cyber lawyer Jonathan Wallace called them on it when he published "The X-Stop Files" in 1997, asking why X-Stop blocked several sites like the Quakers home page, the AIDS Quilt, and parts of Jonathan's own e-zine, The Ethical Spectacle. Peacefire also put up a page in 1998 about sites blocked by X-Stop, including an affirmative action site and a blind children's hospital. But these examples were all found through trial and error; today is the first day that the entire list of URL's has been made public. And to determine the 68% figure, it was necessary to have a copy of the entire list, so that the first 50 blocked sites could be used as a random sample.

So far, this is more or less the same story that took place in 1997 with another blocking program, CYBERsitter, right down to Jonathan Wallace posting a page about CYBERsitter and getting his site blocked. First, several people posted articles criticizing CYBERsitter's policies, and slowly CYBERsitter's public image deteriorated as word got out that they were blocking sites which criticized their company (even Time magazine got blocked, and then posted an article about how they found themselves on CYBERsitter's list). Then in April 1997, Peacefire released a program that broke the encryption on CYBERsitter's list of blocked URL's. CYBERsitter sent Peacefire a threatening letter demanding that we take down the program and remove all of our links to CYBERsitter's Web page. Jim Tyre, a volunteer lawyer and future founding member of the Censorware Project, sent CYBERsitter a reply telling them they had no case, and we never heard from them again. But UCITA, the Digital Millennium Copyright Act, and the two court injunctions against the right to post DeCSS, didn't exist in 1997. If we had released the CYBERsitter codebreaker today, would CYBERsitter actually file a lawsuit?

The outcome of the DeCSS court cases could, in fact, determine the rights of a private citizen to embarrass a big software company by reverse engineering their products and catching them in a lie. It's easy to forget the importance of legal protection for reverse engineering, because sometimes public opinion is enough: RealNetworks never sued Richard Smith when he revealed that copies of RealPlayer included a "globally unique identifier" to track user's listening habits, and Microsoft never sued Andrew Schulman when he discovered that Windows 3.1 threw up fake error messages about DR-DOS. These were large companies that would have been crucified if they had tried to sue someone for discovering something that the public thought they had a right to know anyway. But legal protections are still important, because sometimes public opinion isn't enough - when the software company doesn't have much of an online reputation to worry about, or when then they have a reputation but they don't care about it.

The RIAA, with their campaigns against MP3 technology and reverse-engineering SDMI, is an example of an organization that doesn't care about their online image - and why should they, since we all download our music for free anyway. CYBERsitter is another good example - they do care about their reputation, but in 1997 their image was that of a children's guardian angel and an ally in fighting government censorship, almost immune to criticism. It took an enormous amount of bad press - letters from CYBERsitter's CEO threatening ISP's and flaming people in general, and at one point actually mail-bombing a lady who sent them a complaint - before even advocates of blocking software started distancing themselves from the company. Even today, CYBERsitter's public image is fairly rosy, and their campaigns of legal harassment hardly affected their reputation at all. (What had you heard about CYBERsitter before you read this article?) It's hard to imagine Microsoft, for example, filing a similar lawsuit without embarrassing themselves and turning their intended target into a martyr. The real threat to "reverse engineering for the public good" is from medium-sized companies, small enough that not everything they do will get in the news, but still big enough to afford lots of lawyers.

This threat affects not just programmers, but even journalists who get anonymous tip-offs - like Brock Meeks and Declan McCullagh, who were threatened with an FBI investigation by CYBERsitter in 1996, after they published their "Keys to the Kingdom" article about sites that CYBERsitter and other "censorware" programs blocked. The part of the article that got them in so much trouble was this excerpt from CYBERsitter's bad- word file:

[up][the,his,her,your,my][ass,cunt,twat][,hole]
[wild,wet,net,cyber,have,making,having,getting,giving,phone][sex...]
[,up][the,his,her,your,my][butt,cunt,pussy,asshole,rectum,anus]
[,suck,lick][the,his,her,your,my][cock,dong,dick,penis,hard on...]
[gay,queer,bisexual][male,men,boy,group,rights,community,activities...]
[gay,queer,homosexual,lesbian,bisexual][society,culture]
[you][are][,a,an,too,to][stupid,dumb,ugly,fat,idiot,ass,fag,dolt,dummy]

If this now counts as a "trade secret" under the Digital Millennium Copyright Act, then our list of the 50 .edu sites blocked by X-Stop - and the study that found the 68% error rate - could be declared illegal. And under UCITA, CYBERsitter could even claim the enforceability of these excerpts from their license agreement:

Reverse Engineering Prohibited
Unauthorized reverse engineering of the Software, whether for edcucational, fair use, or other reason is expressly forbidden. For the purposes of this license the term "reverse engineering" shall apply to any and all information obtained by such methods as decompiling, decrypting, trial and error, or activity logging.

Non-Disclosure
Unauthorized disclosure of CYBERsitter operational details, hacks, work around methods, blocked sites, and blocked words or phrases are expressly prohibited.

So any CYBERsitter user who even discusses what the program blocks, would be in violation. Not that CYBERsitter would enforce this against everybody, but they probably would have liked to enforce it against Brock and Declan.

At this point, we don't know how X-Stop will respond to our report. But we do know that for all of their bluster, CYBERsitter never actually sued Brock, Declan or Peacefire. Given that CYBERsitter pursued the matter for months (and the fact that Brock and Declan had actual money), if CYBERsitter gave up, it's because they had no case. If the Digital Millennium Copyright Act, UCITA, or the DVD court rulings change that situation, then it will become much harder to criticize blocking software - or any kind of software - except for the user interface and other things that users can "see" without looking under the hood.

Comments

Maybe sell smut files?

[www.sorehands.com]

When I was at MSI (the original developer of Cyber Patrol), we talked (more like joked) about selling the CyberNot list as a seperate product. This would be for users who wanted an easy way to find porn.

Of course, it did not get far beyond the joking stage. I do suspect that a few people at the company kept copies of the porn lists for their own use. ;). And no, I was not one of them.

It's a little hard for a company to keep a straight face when selling a list of porn site (or publishing porn themselfs) and selling an internet filter or blocker. It's along the line of a virus protection software company also selling a virus writing kit.

CyberPatrol does have a page to check their list to see if a site is on it. It does not do much good for seeing how good their list is, but at least you can check if your own site is on it.

Re:User-agent "Mudcrawler"

[arivanov]

That is not enough, but the checklist to make sure that your site never hits the morons checklist is:

• Start with a fresh new domain
• Make sure your robots.txt file says not to scan the entire site. Quite a lot of them ignore the file so this is not enough.
• Disable any unknown user agents. If your site is 100% PHP or PERL it is an elementary function in the beginning of the script.
• Rewrite all texts replacing a dictionary of smut-words (the posted ones are a good example as their quoted or escaped equivalents. The smut checkers are where virus checkers were 10 years ago. They do only elementary pattern matching. This will also help against stream scanning sofwtare.
• Keep an eye on your logs. If you see that someone is crawling check who they are. Firewall or ban the IP range if necessary.
• Firewall all IP ranges belonging to known anti-smut software vendors.

Have fun if you have brains. If you do not you shall be filtered. Resistance is futile. Isn't technology wonderful?

Almost forgot - the described technologies do not consititute reverse engineering and as such do not fall under the provisions of the UCITA and the DMCA.