8 September 2002: Add responses.
7 September 2002
Date: Sat, 07 Sep 2002 13:10:59 -0500
Subject: Open email from Google Watch
I run a site at http://www.google-watch.org that is critical of Google's privacy policies, and that also offers an anonymous Google proxy. China's recent blocking of Google and AltaVista has me wondering whether they might be doing it for reasons that have not been mentioned in the press.
Our proxy has not been blocked. We wonder who will block us first -- China or Google, Inc.? The former because we're a back door to Google's index, or the latter because our interface is ad-free and unauthorized by Google?
U.S. intelligence agencies have recently shown a great deal of interest in Internet surveillance. One important aspect of this is determining geolocation from IP number. Currently this is about 80 percent effective in fixing the IP number to a major city, and over 90 percent in fixing it to a country.
Another important aspect is the search terms used to query search engines. These terms are absolute pearls; they are a succinct window into the Internet user's interests and state of mind at a particular point in time. Clustering analysis that used geolocation along with search terms would provide an insight into a society.
If I were a Chinese official, I'd be worried that Google logs all search terms together with the IP number, a time stamp, a unique cookie ID, and browser information. If this information is available to the National Security Agency from Google -- and I believe that current U.S. laws almost require Google to provide this information to the feds, especially when the Internet user is located in a country that's of national security interest to the U.S. -- then China may be well-advised to block the use of U.S. engines to protect their own national security.
The NSA, if it gets this information straight from Google, is operating at a level of efficiency much greater than Chinese officials themselves, who must intercept and collate such information by monitoring the packet stream. This puts the NSA at a tremendous advantage for determining where pro-U.S. sentiment may exist in China.
The privacy policies of search engines generally do not cover things such as IP number storage, and storage of search terms. In the case of portals that use Google results, it is important to know whether the portal forwards the IP number to Google along with the search terms. I've asked this question of several portals, and received a reply only from Netscape, which said that they do not forward the IP number to Google.
An organization such as Cryptome can provide a service by asking engines such as Google to restrict the user data they collect. When engines fail to do this, then they should at least allow proxies and meta-engines to access their index as a matter of policy and convenience to Internet users around the world.
-- Daniel Brandt,
Public Information Research, Inc.
Cc: email@example.com, firstname.lastname@example.org
Subject: Open email from Google Watch on Cryptome.org
Date: Sat, 07 Sep 2002 23:30:52 -0700
Open email from Google Watch on Cryptome.org
FYI, the following statement is inaccurate and needs to be revised as it presents an invalid picture of the situation of tracing the origins of IP connections.
"Currently this is about 80 percent effective in fixing the IP number to a major city, and over 90 percent in fixing it to a country."
Depending on the website, AOL can account for 30% to 80% of the web traffic that is recorded in web logs. AOL uses proxy servers that have varying IP addresses that all trace back to McLean Virginia or other cities in Virginia. Most Webmasters that run web analysis reports by reverse DNS, find that 30% to 80% of the traffic comes from Virginia (AOL). Unless AOL assists in the tracing of the connection to the AOL proxy servers, it is impossible to trace the source. Proxy servers hide the true origination of the connection, using a pool of IP addresses.
So the 80% claim of fixing on a major city is inaccurate. And the 90% claim is not provable since AOL connections come from all over the world. Only AOL could tell you the countries of origin. And if you have ever dealt with AOL, then you would know that they do not provide any information to outside parties, with the exception of law enforcement.
To verify my statements, contact any reputable Webmaster that processes a large amount of web logs and they will verify what I have said about AOL traffic.
My credentials: 25 years in the computer industry, International consultant to Fortune 100 companies. Experience with web log analysis and data mining for several major web sites that logged millions of requests (hits) a day, data warehousing of web log information that exceeded three Terabytes in size.
Date: Sun, 08 Sep 2002 18:47:20 -0500
Subject: Response to a rebuttal
The email from the poster who claimed that 30 to 80 percent of all traffic comes from AOL is not correct. AOL reverse-resolves quite reliably to *.aol.com, proxies or not, over the seven years I've been log watching. Currently aol.com accounts for under six percent of our traffic. Yes, AOL is a problem for geolocation, unless you have inside information from AOL (which the feds most likely have). But it's not a 30 to 80 percent problem by any stretch of the imagination.
There are firms that contract with the feds to map geolocation to IP number. Advanced techniques require a network of servers to run simultaneous traceroutes to help pinpoint the geolocation. Once computed, of course, it gets cached. If you do this long enough, you build up a fairly decent map. MetaCarta is one such firm. They've been sending out spiders to crawl the web, as part of their geolocation effort. Check out their page at www.metacarta.com/jobs.htm to find out the sort of folks they hire. We had to block their spider, it was so intrusive. It was also difficult to trace, because they were trying to disguise it.
Other sites interested in geolocation information are:
I have no evidence that those two sites are interested in federal contracts.
Then there is the following quote from the Washington Post, January 4, 2002, p. E1 (copy below):
"Some analysts say the barriers could grow with the development of 'geolocation' technology, which attempts to match a person's location based on a computer's Internet address.
"Silicon Valley's Quova Inc., one of the leading providers of this technology, claims it can correctly identify a computer user's home country 98 percent of the time and the city about 85 percent of the time, but only if it's a large city. Independent studies have pegged the accuracy rate of such programs, which also are sold by companies such as InfoSplit, Digital Envoy, Netgeo and Akami, at 70 to 90 percent."
Finally, I should mention that credit card companies use geolocation for real-time validation, as one of their fraud-detection measures. Frequently they will return this information along with a percent reliability figure, so that the merchant can decide whether he should delay shipment of the product pending further processing and validation.
Between the feds and the banks, there's plenty of interest and money behind geolocation techniques. To claim otherwise is irresponsible and uninformed.
-- Daniel Brandt
Date: Sun, 8 Sep 2002 13:50:09 -0500 (CDT)
Subject: Re: Open email from Google Watch
The importance of concentrated search sources for spy agencies has little to do with pinpointing individuals.
Chinese rulers could easily eliminate cookies and IP data by using proxies, and reap publicity rewards for protecting its citizen's privacy.
The most crucial aspect is having a hand on the pulse of subjects, not individuals but trends. In today's world of 100% manufactured news the real mood among subjects becomes harder and harder to find an diagnose on time. How do you find the real prevailing public opinion on this fatherland hysteria ? The one you use to prepare your enforcement and propaganda trops for. In the west coast town where I live, most people that I know positively cannot stand flag displays and consider Bush a criminal. This is valuable information.
Search engine queries are much easier to harvest and process than e-mails or even web traffic. Even with full taps at MAE nodes and fiber feeds it is a huge amount of diversely formatted information, from which inferences must be made. Search engine strings tell you in a very straightforward way what people are looking for. People from my town, for example. Imagine just collecting ONE string, not the query but Google category URL that is often included in response. Everything has been done for you, all that is needed is a simple program to tally and print. Even FBI can do that.
I wonder if there are federal troops discreetly protecting google facilities.
Of course, it was not only American rulers that benefit - Chinese rulers benefited as well. But I think that they decided to make their own search node. It would be interesting to find out if there is a correlation between blocking of google and a major launch effort of a chinese search engine.
By Ariana Eunjung Cha
Washington Post Staff Writer
Friday, January 4, 2002; Page E01
It is the modern-day equivalent of a border sentry. When visitors try to enter UKBetting.com, a computer program checks their identification to determine where they're dialing in from. Most people are waved on through. Those from the United States, China, Italy and other countries where gambling laws are muddy, however, are flashed a sign in red letters that says "ACCESS DENIED" and are locked out of the Web site.
For much of its life, the Internet has been seen as a great democratizing force, a place where nobody needs know who or where you are. But that notion has begun to shift in recent months, as governments and private businesses increasingly try to draw boundaries around what used to be a borderless Internet to deal with legal, commercial and terrorism concerns.
"It used to be that a person sitting in one place could get or send information anywhere in the world," said Jack Goldsmith, a professor of international law at the University of Chicago. "But now the Internet is starting to act more like real space with all its limitations."
These new barriers take many forms. One method is to simply restrict who has access to computers and gateways to the Internet. Another is to make all communications pass through filters that seek to weed out objectionable content, such as pornography or information deemed to endanger national security. Growing in popularity is software that attempts to match a computer's unique Internet address with a general geographic location, a technology that is becoming more precise every day.
The debate is no longer about if we can create these barriers, but rather whether we should. Even those who support the idea in theory disagree on who should erect and maintain the electronic fences, whether it should be done by nation-states or by the Web site operators.
The new borders provide what some call a neat solution to the vexing problem of how to resolve the often-conflicting policies of the roughly 200 independent states of the world on matters such as gambling, commerce, copyright and speech.
But critics fear that the barriers will create an Internet that's balkanized. And civil rights groups warn that freedom of speech will suffer, that the technology will make it easier for oppressive governments to stifle nonconformist viewpoints, and that people's privacy will be eroded, especially because some technologies can pinpoint one's location.
"It's likely that the Internet of tomorrow will look radically different from different parts of the world," said Lee Tien, a lawyer with the Electronic Frontier Foundation in San Francisco.
Already legislatures and court systems around the world have been attempting to assert their country's authority over the World Wide Web. Hong Kong's government, for instance, has been debating whether to pass a law that would make it a crime for any overseas gambling site to offer services to its residents. A court in Genoa, Italy, recently found the operator of a Web site in another country guilty of libel. A French judge has ordered Yahoo to stop selling Nazi paraphernalia because a law there bans such practices.
Without an international treaty or mediation organization, such rulings have so far been largely unenforceable on parties residing outside a country's borders. But that has not stopped countries from drafting rules for what is and is not permissible online.
At least 59 nations limit freedom of expression, according to Leonard R. Sussman, author of "Censor.gov." Singapore, for instance, works with Internet access providers to block any material that undermines public security, national defense, racial and religious harmony, and morals. That includes pornography and hate speech.
Some analysts say the barriers could grow with the development of "geolocation" technology, which attempts to match a person's location based on a computer's Internet address.
Silicon Valley's Quova Inc., one of the leading providers of this technology, claims it can correctly identify a computer user's home country 98 percent of the time and the city about 85 percent of the time, but only if it's a large city. Independent studies have pegged the accuracy rate of such programs, which also are sold by companies such as InfoSplit, Digital Envoy, Netgeo and Akami, at 70 to 90 percent.
The system is not foolproof; people can easily get past by using special software programs to cloak their identities. But experts such as Goldsmith, the Chicago law professor, say the technology need not work perfectly to have an impact. These barriers act like checkpoints on a nation's physical border: They can be evaded, but most people probably won't want to go to all the trouble.
Gambling sites were among the first to roll out the technology, last year. When users from countries where online gambling is not allowed try to get on, they are either not given the option to place bets or they are kicked out when they try to register for an account.
"There are a number of sites out there that just don't care about the laws. They are perfectly happy to let U.S. gamblers in even though they know it's illegal," said Jeremy Thompson-Hill, an account manager for OrbisUK, which provides the sentry technology used by Sports.com, Ladbrokes.com and other betting sites. "But most reputable companies want to be able to say to the United States, 'We're taking every reasonable precaution to prevent the use of our gambling software in your country.' "
The technology also is being embraced by Web broadcasters, whose nascent industry had been growing slowly because of concerns about copyright. JumpTV is betting its future on this technology. The Montreal-based venture retransmits television broadcasts from around the world and is trying to avoid being sued by broadcasters who claim it violates their broadcasting licenses. In early 2000, a U.S. judge effectively shut down another Canadian company called iCraveTV by prohibiting it from broadcasting its signals into the United States for 90 days.
Farrel Miller, JumpTV's chief executive, said the company hopes to begin retransmitting ABC, CBS and NBC only to Canadian viewers early next year but was much more modest about his company's aspirations than some heads of other webcasting companies during the dot-com boom.
"We don't see the Internet as a revolutionary medium that will change the TV business," Miller said. "It'll be just another alternative vehicle for disseminating channels."
The difficulty in recognizing nation-state borders on the Internet became such a concern during the 2000 Sydney Games that the International Olympic Committee effectively banned most Web video of the events.
Television stations had paid enormous fees for the rights to broadcast the games on a country-by-country basis -- NBC, for instance, shelled out $3.5 billion for the United States -- and they were worried that piracy or even legitimate online transmissions that were accessible to anyone, anywhere might devalue the worth of those contracts. The IOC and many of the owners of broadcast rights say the accuracy rate for geolocation technology is still not good enough and they won't allow any webcasts for the Salt Lake City games this February.
"The technology just doesn't pass muster yet. There's no way to guarantee that your broadcast would be confined to your territory and would not run in to someone else's," said Kevin Monaghan, a vice president for NBC Sports.
Even if geolocation technology worked perfectly, some legal experts said it would not be feasible because it would require Web site operators to know the applicable laws in every country.
"Geographical location technology is a red herring," said Alan Davidson, a lawyer with the Center for Technology and Democracy, a Washington think tank. "It would be incredibly burdensome to tailor content to meet all of the different laws in all of the different countries everywhere the world."
That's the heart of the question being addressed by a court case that pits Yahoo Inc. against France.
Last year, two French groups -- League Against Racism and Anti-Semitism and the Union of Jewish Students -- sued Yahoo for allowing Nazi collectibles to be sold on its auction pages. The sale of such hate material is illegal in France. Almost 1,000 such items were on the block at the time, including Adolf Hitler's "Mein Kampf," stamps and coins, as well as hate paraphernalia.
Jean-Jacques Gomez, a judge in Paris, ordered Yahoo to prevent French users from seeing the material by using the geolocation technology.
Yahoo declined on principle and sued in U.S. District Court in San Jose to make the order unenforceable because a foreign judge could not impose such conditions on a U.S.-based company. U.S. Judge Jeremy Fogel ruled Nov. 7 that the First Amendment trumps overseas laws when they pertain to content produced by U.S. companies. An appeals court upheld the decision but the French groups have appealed again and have vowed to take the case to the U.S. Supreme Court if necessary.
The attorney for the French groups, Ronald Katz, argues that the issue is not about free speech but about national sovereignty.
"Yahoo wants to use the court decision as a sort of megaphone to say the U.S. controls the Internet," he said.
Indeed, the U.S. dominance of the Internet is one major thing that observers say will change with the new electronic borders, slowing the dissemination of ideologies and culture across countries.
"Is geographical tracking a panacea that solves international jurisdiction issues? Probably not. But is it a technology that's significantly changing the social, economic and political aspects of how we communicate on the Internet?" Davidson said. "Absolutely."
c 2002 The Washington Post Company