August 13, 2004
Last Updated: August 13, 2004
http://www.opennetinitiative.net/bulletins/004/
Contents:
- Background
- Methodology & Results
- Conclusions
- Methodology & Limitations
- About the OpenNet Initiative
Access to the Internet in Iran is presently subject to official censorship, although the precise scope and scale of the filtering is unclear. (1) For many years, Iranian authorities allowed unencumbered access to the Internet, offering a departure from its own practices towards traditional media, such as newspapers, television, and radio, which are subject to government control. (2) In early 2003, however, news and other reports indicated that Internet censorship would be introduced in Iran, with some reports indicating up to 15,000 websites to be filtered. (3) Shortly afterwards, Iranian users of the Internet began reporting blocked websites, including non-pornographic and increasingly popular blogging sites. (4) It was also reported that access to Google's cache function was filtered in late 2003, although that appeared to be a temporary measure. (5) Recent news coverage indicated a tightening of content controls had occurred leading up to the February 2004 parliamentary elections. (6) One report claimed that 100 billion websites had been censored by Iran in the past year. (7)
Typically reports such as these lack precision, referring instead to general trends and rounded-off numbers. Two recent reports have offered more specific details about what websites in Iran are filtered. One prominent Iranian activist site, "stop.censoring.us" reports that Iranian authorities issue official "blacklists" distributed to ISP operators, who are then responsible for putting in place the content filters. According to the report, the blacklists of banned sites are updated regularly, stored on CDs, and then distributed to each of the ISPs, some of whom do not always comply. Although the lists are said to include only pornographic sites, a recent blacklist acquired by stop.censoring.us and posted to their website is unique in that it contains a list of political, dissident, religious, and blogging sites. (8) The second report is authored by the advocacy group Reporters without Borders (RSF). In their annual Internet Report, "The Internet Under Surveillance, 2004," the section on Iran contains a list of censored websites. (9)
As part of its ongoing research into Internet censorship and surveillance worldwide, including in Iran, the ONI connected to remote computers in Iran and tested the RSF list and the blacklist obtained by stop.censoring.us. (10) We attempted to connect to each of the URLs in the two lists through 10 remote computers in Iran and our control location in Toronto, Canada. Each of the remote computers in Iran that we employed in this probe is connected to the Internet via state-owned Telecommunication Company of Iran (TCI), Iran's largest ISP and telecommunications company. Although there are hundreds of smaller ISPs in Iran, TCI is by far the largest and the one from which many other smaller ISPs acquire their connectivity. Results of these tests can be found here and here.
Almost all of the websites contained on the blacklist and RSF list are Iranian or Iran-related (and mostly Farsi) in their content. In order to further understand the scope of Iranian content filtering, we compiled a categorized list of general - what might be called "global" - websites and tested them through the remote computers in Iran and our control location in Toronto. The results of those tests reveal a much different picture. With respect to mostly non-Iranian websites, such as general human rights, news, and government websites, Iran appears not to be engaging in much content filtering at all. There are exceptions, to be sure. Pornography, sex, gay, and some proxy/anonymous surfing-related websites from around the world are all subject to censorship. But at the present time, Iran seems mostly concerned with filtering Iranian, and mostly Farsi-language, content.
In total, we tested 1,076 unique web pages and were blocked at least once from accessing 439 of these pages (41%), with 425 pages (39%) inaccessible in the majority of tests. The chart below shows the number of pages tested from each list and the number blocked in the majority of our tests (pages inaccessible from our control location in Toronto have been excluded). (11)
| List Name | Pages Tested | Pages Blocked | Percent |
| Blacklist | 546 | 378 | 69% |
| RSF List | 22 | 11 | 50% |
| Global List | 519 | 54 | 10% |
Blacklist Analysis
Over 80% of the pages categorized as either sexual or political/religious/social within the blacklist were blocked during our testing, suggesting few situations where previously blocked pages were later made accessible in those particular categories. The percentage of pages categorized as blogs that remain blocked, however, was only 34%. This may represent a refinement in the criteria for blocking blog pages in response to the attention called to Iran's Internet filtering by CNN in July of 2003 (12). For example, the blacklist contains 88 pages from the blogspot.com domain, only 37 of which were blocked at the time we conducted our tests. We also discovered 6 pages on the persianblog.com domain that were blocked in a minority of our tests.
Global List Analysis
On the global list, the majority of the blocked pages (78%) were pornographic in nature, with every such site we tested being blocked. As our list deliberately includes both high profile sites and others not so well known, the inference can be made that Iran is both very interested and very successful at preventing pornography from entering the country via the Internet. Another 15% of the pages related to gay and lesbian topics, including such sites as www.gayhealth.com and www.iraniangaydoctors.com. (13) We were, however, able to access 86% of the pages categorized as "Gay and Lesbian Rights", suggesting either a less concerted effort to prevent the viewing of this content or that these blocks were a result of an over-aggressive attempt to filter pornography (given that these sites frequently discuss sexual topics). The remaining 7% of blocked pages from this list concerned methods to circumvent Internet filtering, such as www.anonymizer.com.(14)
A number of limited conclusions can be drawn from this preliminary probe. First, Iran is indeed engaged in extensive Internet content filtering beyond just pornography, including many political, religious, social, and blogging websites. Most of these censored websites are Iran-specific; very little non-pornographic, "global" content is filtered from Iranian users. A better understanding of why Iranian authorities have decided this content must be subject to censorship will have to await contextual research that follows upon our initial technical interrogation probes outlined here.
A second conclusion relates specifically to the blacklist and the RSF list of banned sites, and the inferences that can be drawn from findings of content filtering within a country. As our results show, while many of the sites contained on both lists are filtered, some are not. There could be many reasons for this variation. Some of the websites included in the "blacklist" and RSF list may have been included by mistake and were never the subject of content filtering in Iran. Alternatively, the Iranian government may have filtered the websites in the past but discontinued doing so while our tests were being performed. Or the websites included in the two lists may indeed be filtered in Iran, but not by Iran's main ISP through which we ran our tests, namely TCI.
Finally, the case presented here provides an instructive lesson concerning the type of limited conclusions that can and should be drawn about both the extent and character of Internet censorship in any country. Determining the exact number of websites filtered in a particular country is a difficult exercise, particularly when such practices are not transparent, as is the case with Iran. Content filtering can be the result of ad hoc local procedures put in place by one or more ISPs acting independently, or a more centralized directive delivered from the state authorities. Even if it is the latter, it is important to remember that compliance with such directives is not always complete. In addition, websites that are filtered one day might not be filtered the next, as policies and/or content may change, suggesting that reports of total numbers of websites blocked in any report need to be treated with the caution.
In order to enumerate Internet filtering worldwide, we use remote computers located in countries that employ content filtering and blocking practices.
However, there are limitations in content filtering and blocking research. The network connection errors indicative of Internet filtering and blocking are identical to normal errors that can occur during the course of regular Internet traffic routing. Furthermore, the remote computers may return results that are not indicative of overall Internet filtering within a given country -- the results might only indicate filtering taking place within a given ISP.
The OpenNet Initiative is a partnership of the Citizen Lab at the Munk Centre for International Studies, University of Toronto, the Berkman Center for Internet & Society at Harvard Law School, and the Advanced Network Research Group at the Cambridge Security Programme, University of Cambridge.
The OpenNet Initiative releases occassional bulletins based on our ongoing research. These bulletins are meant to be limited responses to current events, policy debates, and/or issues raised by our ongoing research that we feel justify immediate wider circulation. Our more detailed analyses can be found in our major reports.
URL submissions for testing
If you have specific URL's you think may be blocked in Iran, please submit them to [email protected] and we will add them to our next list for testing.
Notes
1. At the WSIS summit in Geneva, President Mohammed Khatami defended Iran's policies on censorship of Internet pornography and "immoral websites" but the government has not provided an official list of blocked sites or how filtering is imposed. http://news.bbc.co.uk/1/hi/technology/3312841.stm
2. http://news.bbc.co.uk/1/hi/not_in_website/syndication/monitoring/media_reports/2183573.stm
3. http://news.bbc.co.uk/1/hi/technology/3019695.stm
4. http://www.rsf.org/article.php3?id_article=10733&Valider=OK
5. http://news.bbc.co.uk/1/hi/technology/3310493.stm
6. http://news.bbc.co.uk/1/hi/world/middle_east/3376907.stm
7. http://stop.censoring.us/archives/011478.php
8. http://stop.censoring.us/archives/011038.php
9. http://www.rsf.org/article.php3?id_article=10733
10. The "blacklist" obtained by stop.censoring.us was uncategorized and contained some errors likely due to OCR scanning of a hard copy of the list. We corrected those errors and categorized the list.
11. Blocking Behavior Analysis
97% of the pages we found blocked by Iran (on the TCI network) were inaccessible in all of our tests, across all ten remote computers. This high rate of consistency is not surprising, given that all of these computers access the Internet via the same ISP. The slight variations in behavior, however, may provide insight into the way in which the TCI network modifies and distributes the list of pages to be blocked. For example, the page www.iransexy.web1000.com was checked via all remote computers on three separate occasions. One particular computer was able to access the page every time while the others were consistently blocked from accessing the page. Similarly, of seven pages that were inaccessible to only two of the ten remote computers, all were inaccessible via the same two computers. While it is impossible to know conclusively why certain pages are blocked from some locations and not others, it may indicate that some of TCI's filtering servers might be using differing copies of a list of blocked pages.
We tested multiple pages from 43 domains (for example, www.geocities.com/farstoons and www.geocities.com/fisgo2003 are both on the domain www.geocities.com) and found only three domains that contained pages that were both blocked and unblocked. Two of the three were commercial/free hosting domains (www.geocities.com and www.ultimatetopsites.com), where pages containing sexual content or material questioning Islam were blocked but the rest were accessible. The third was the Voice of America news site, where the front page in Persian was blocked, while the front page in English was not. Our tests found http://www.voanews.com/persian/index.cfm blocked, but not http://www.voanews.com. �These results suggest deliberate action on the part of the authorities to engage in fine-grained content filtering, where specific pages within domains -- as opposed to the domain as a whole -- are blocked.
12. http://edition.cnn.com/2003/WORLD/meast/07/16/iran.blogs/index.html
13. The other sites were www.gaychile.com, www.ggb.org.br, www.khanaye-doost.com, sappho.shoe.org, www.gay.ru, and www.lesbians-against-violence.com.
14. The other sites blocked from this category were download.com.com, www.peek-a-booty.org, and anon.inf.tu-dresden.de.