Data analysis requirements for BD research report

[agrabeli]

As part of an upcoming research report on internet censorship in Bangladesh, we would like to examine the following questions based on OONI data:

1. Which news media websites are blocked in Bangladesh? Are there cases of collateral damage that we can detect?
2. Which techniques do ISPs in Bangladesh use to implement the blocks? How does the blocking of websites vary across ISPs in Bangladesh?

Note: There were cases where Cloudflare IPs were reportedly blocked, resulting in collateral damage.

Data analysis requirements

Below I share the data analysis requirements in support of this research. The deadline for this data analysis is 30th November 2024.

• Country code: BD
• Date range of analysis: 1st November 2023 - 1st November 2024 (1 year)
• Analyzed OONI measurements: Web Connectivity
• Citizen Lab category codes (as part of Web Connectivity analysis): NEWS

Please analyze OONI measurements collected from Bangladesh between 1st November 2023 - 1st November 2024 (1 year), while limiting the analysis to the NEWS category code of Web Connectivity measurements. Please also check for potential cases of collateral damage, such as websites getting unintentionally blocked as a result of IP blocking.

Based on this analysis, generate charts which display:

• Domains that presented signs of blocking during the analysis period (while excluding domains that received very limited measurement coverage and a low volume of anomalies);
• Measurement results (ok, tls.timeout, tls.connection_reset, dns.nxdomain, etc.).

Such charts would be similar to this: https://ooni.org/post/2024-tanzania-lgbtiq-censorship-and-other-targeted-blocks/images/image6.png (enabling us to gain a bids-eye-view of failure types and censorship techniques)

If you observe variance in censorship techniques adopted by different ISPs, please also generate a chart for each category which displays:

• ASNs that received the largest measurement coverage (based on the NEWS category);
• Measurement results (ok, tls.timeout, tls.connection_reset, dns.nxdomain, etc.).

Such charts would be similar to this: https://ooni.org/post/2024-tanzania-lgbtiq-censorship-and-other-targeted-blocks/images/image13.png (but applied to a category code, as opposed to a domain)

Note: We have previously documented TLS-based interference in Bangladesh: https://explorer.ooni.org/findings/11686385001. However, many sites tested on HTTP are automatically being annotated as confirmed blocked based on OONI fingerprints (https://explorer.ooni.org/search?since=2024-03-30&until=2024-04-30&failure=false&probe_cc=BD&only=confirmed).