Since the Internet appeared in Uzbekistan in the 1990s content has been filtered consistent with national legislation. Testing carried out by Black Watch focusing on news sites operated by BBC and Radio Free Europe/Radio Liberty reveal that Uzbekistani authorities block news websites that are critical of the government and provide this information in widely-understood languages in Uzbekistan.
From January to February 2015, Black Watch testing revealed that Uzbekistani authorities block specific websites operated by BBC and Radio Liberty/Radio Free Europe that present content that is critical of the Government of Uzbekistan in a widely-understood language. SecDev analysts confirmed the test results between May 5th — 7th, 2015.
This is not entirely surprising. Uzbekistan is among the first countries in the Eurasian region to apply content controls on the Internet. Since independence in 1991 Uzbekistan has experienced significant political violence. On February 16th, 1999, car bombs in Tashkent killed 16 people and wounded over 120 more. Clashes in the mountains between the military and Islamic Movement of Uzbekistan (IMU) insurgents occurred in 2000. Simultaneous bombings at the American and Israeli embassies in 2004 killed at least two people. In the most extreme example of political violence, brutal government suppression of protests in Andijon killed approximately 750 civilians in 2005.
External violence and insecurity also play a role in the general instability of Uzbekistan. The civil wars in Tajikistan and Afghanistan in the 1990s created unstable countries on Uzbekistan’s borders, and the decade-long war in Afghanistan after the 2001 American invasion did not help. Furthermore, unrest in Kyrgyzstan flows into Uzbekistan; the 2010 ethnic cleansings in Osh led to approximately 100,000 refugees seeking shelter in Uzbekistan’s Fergana Valley. Kazakh and Uzbek communities have long viewed each other with distrust, both on state and communal levels. Water politics hurt governmental relations between Uzbekistan, Tajikistan, and Turkmenistan, and the borders with Kazakhstan and Kyrgyzstan remain disputed. Authorities also fear, legitimately or not, interregional threats such as the Arab Spring and militant Islam in the form of the Islamic State and the IMU. Needless to say, Uzbekistan is not a bastion of stability.
In recent years, internal and external pressures have heightened. Politically, the country faces a significant succession challenge. In March 2015, President Islam Karimov won his third term in office under the current constitution, a position he has held since independence. He is 77 years old, and in bad health. While he soundly beat his largely symbolic opponents with over 93% of the vote, the elections coincided with the release of details of corruption against government officials and, significantly, the President’s family.
The country’s economy has plateaued and opportunities for youth are stagnant, or decreasing. The population has soared by full 50% over the past two decades — from 20.9 million citizens in 1991 to 30.2 million in 2013 — contributing to pent-up political frustrations, and fuelling rivalries between competing political and tribal factions.
The Internet has also grown more significant in Uzbekistan. Internet access and usage increased from 7.5% of the population in 2007, to 38% by 2013 — a fivefold increase. All indications are that this number will continue to increase. The growing availability of smart phones, which offer access to the Internet, as well as the steady growth of wireless access throughout the country will continue to contribute to a steady pace of growth.
Uzbekistani authorities are clearly concerned about the Internet’s potential to form colour revolutions, as occurred in neighbouring Kyrgyzstan, and the kind of political turmoil that occurred from November 2013 until February 2014 in Ukraine. Criticism of the government is carefully monitored. Domestic mass media is tightly controlled as are local Internet sites and resources, evidenced by crackdowns over the better part of the 2000s. International news represents a challenge as it falls outside of national jurisdiction. At the same time, the government of Uzbekistan does not want to present an image of a closed authoritarian state. Blocking of international media sites is therefore selective, and targeted.
What did Black Watch testing uncover?
In the case of Radio Free Europe/Radio Liberty (RFE/RL), authorities limit all content in languages widely understood in Uzbekistan: Russian, Kazakh, Kyrgyz, Tajik, Uzbek, and English. RFE/RL’s content across all platforms in Eurasia is relatively similar, sharing information critical of governments in Eurasian countries. Uzbekistan’s decision to limit access to RFE/RL stands in stark contrast to the rest of the former Soviet Union where it is largely accessible. (Note: Testing is not conducted in Turkmenistan.)
Uzbek authorities apply this same model to blocking BBC content, but the context is different. Only Uzbek-language outlet is blocked while the English-language version remains accessible. One reason for this may be that English and Russian language BBC sites provide very little coverage of Uzbekistan and are different than those covered by the BBC Uzbek Service. For example, the day after the Organized Crime and Corruption Reporting Project’s report on Gulnara Karimova’s corruption in the ICT sector broke on March 21st 2015, the BBC Uzbek-language site’s ‘must read’ section had two articles on the upcoming Presidential elections, an article about Ms. Karimova’s involvement in the ICT sector, and an article about football. The BBC Russian-language version was very different. It featured articles on: corruption in Ukraine, Ukrainian oligarch Ihor Kolomoysky’s firing, a television ad in poor taste, news about the ‘political reality show’ in Ukraine, and the 10 ‘most outstanding’ things Jeremy Clarkson has said. The difference in coverage between the two services likely contributed to the decision to filter BBC Uzbek services, while allowing BBC Russian and English to remain available.
To date, Internet filtering in Uzbekistan has operated according to a careful calculus; just enough to ensure stability, not enough to incite wide-scale protest or international condemnation on the scale of China’s “Great Firewall.” The patterns detected by Black Watch testing appear to validate that this approach remains in force. Uzbek authorities are highly sensitive to domestic criticism and the potential for internal instability. At the same time, they wish to project outward image of a progressive and stable country. For the moment, that may work. But as more citizens get online, and as economic and political pressures mount, Uzbekistan’s selective Internet filtering may prove insufficient to contain social or political unrest.
Testing methodology and results: Examining DNS results by country
Black Watch testing provides analysts with a large amount of technical feedback, including the requested URL, the final URL, the status code, the geographic path the data took to and from the website, the ISP the test was conducted on, the response header, and a summary of the response data, the country, the size of the page, and a snapshot of the data. For the purposes of this specific set of tests, SecDev analysts investigated the DNS A requests to determine if additional suspicious activities occurred. Black Watch analysis revealed two important indicators of potential intentional blocking. First, DNS A records were inaccessible inside Uzbekistan, but accessible inside Tajikistan. This suggests that content is being restricted within Uzbekistan. Second, SecDev analysts also discovered multiple instances of DNS A records alternating between accessible and inaccessible, depending on the website. This may indicate that blocking is IP based, not DNS based.
Content Delivery Networks (CDNs) send users the content they have requested from the closest available server. This reduces the distance that data travels, increases the speed with which they receive this data, and helps to mitigate the effects of DDoS attacks. Users are served with the same content, but from different servers (and IP addresses) based on their location.
When SecDev analysts conducted their analysis of Black Watch testing results, they found that the testing devices in Uzbekistan and Tajikistan were receiving the same DNS A records in response to their queries for RFE/RL’s language-specific domains. However, tests conducted in Tajikistan were able to retrieve the web content from the associated web servers whereas those from Uzbekistan were not.
Due to the accessibility of these web sites globally and the ability of testing from Tajikistan to successfully retrieve content for the domains in question, analysts concluded that an intentional censorship of RFE/RL is being applied within Uzbekistan along linguistic lines. RFE/RL websites in languages not widely spoken in Uzbekistan are accessible from the same IP addresses that fail to deliver content for blocked sites. This suggests that the blocking is occurring at the application layer, which is supported by the HTTP timeouts that analysts observed.
The following chart shows the DNS A results on February 20th and 21st 2015 for six different sites in Uzbekistan and Tajikistan. Analysts initially identified the pattern in January 2015 and are using the data from February 20th and 21st as a snapshot.
The following chart describes two pieces of evidence:
- Despite identical DNS A responses observed in both Uzbekistan and Tajikistan, the accessibility of the content from the associated IP addresses varies based on the country.
- The DNS A record 18.104.22.168 may demonstrate an example of IP blocking, rather than broader DNS blocking. In this case, the DNS A record produces a response for Azatliq.org (Tatar language) while not for Ozodi.org (Tajik language) or Ozodlik.org (Uzbek language). It is possible that 22.214.171.124 is inaccessible but that Azatliq.org was retrieved from 126.96.36.199. Analysts do not have the necessary information to conclusively confirm IP blocking. This also occurred with the DNS A records 188.8.131.52, 184.108.40.206, and 220.127.116.11.
|Site||Country||CDN||DNS A Record||Accessible|
|Azattyk.org, Svoboda.org||Tajikistan||TransTelecom||18.104.22.168 / 22.214.171.124||Yes|
|RFERL.org||Tajikistan||TransTelecom||126.96.36.199 / 188.8.131.52||Yes|
|Uzbekistan||TransTelecom||184.108.40.206 / 220.127.116.11||No|
|Ozodlik.org, Ozodi.org||Tajikistan||TransTelecom||18.104.22.168 / 22.214.171.124||Yes|
|Uzbekistan||TransTelecom||126.96.36.199 / 188.8.131.52||No|
|Azatliq.org||Tajikistan||TransTelecom||184.108.40.206 / 220.127.116.11||Yes|
|Uzbekistan||TransTelecom||18.104.22.168 / 22.214.171.124||Yes|
Black Watch currently tests Internet accessibility in approximately 45 countries. In our testing, there is nothing to suggest that any of these international sites have widespread inaccessibility problems outside of Uzbekistan.
The above information about accessibility led SecDev analysts to conclude the problem with accessibility is largely within Uzbekistan.
Annex 1 — Testing list
Red denotes the website is inaccessible in Uzbekistan.