‘This content isn’t available in Canada’: How Meta’s news ban is disrupting discourse about the Israel–Hamas war

4 June 2024

By: Nathan Doctor and Guy Fiennes

As the Israel-Hamas War enters its eighth month, ISD analysis reveals that Meta’s media ban is amplifying misinformation and polarising discourse among Canadian Facebook users. This article is part of a research series looking at the impacts of the Israel-Hamas conflict on extremism, hate and disinformation in Canada. This project has been made possible in part by the Government of Canada. All views are ISD’s own.

Summary

In August 2023— just weeks prior to Hamas’ attack on Israel— Meta began blocking access to news links, both local and international, and restricting media outlets’ accounts on Facebook and Instagram within Canada. This followed the passage of a bill requiring Meta to pay news outlets for content appearing on its platforms within Canada. While there is no evidence that either Facebook or large media companies have been financially hurt by Meta’s decision, smaller media outlets are reportedly suffering. Additionally, there are further impacts on users in Canada of these platforms to consider. In the context of the ongoing Israel-Hamas conflict, which has fuelled misinformation and hate speech around the world, the ban has driven a drastic shift in how information is spread online within Canada. While posts by outlets are automatically made unavailable to users, Canadians attempting to share links or receive news on Facebook and Instagram (see Figure 3) now see an error message: “This content isn’t available in Canada.”

ISD collected approximately 80k posts about the Israel-Hamas and Russia-Ukraine conflicts from Canadian and Australian Facebook groups (see Methodology – Data Collection) to compare the impact of the ban in Canada with that in a similarly sized English-speaking country, as well as to analyse another conflict prior to the ban. Evaluating and comparing the content types and external links users shared, we found that users have largely responded to the ban by sharing photos and content from other platforms at higher levels. Such content tends to feature more bite-sized, out-of-context, information (e.g. screenshots of headlines), opinion, propaganda, and misinformation, compared to links from traditional media outlets or re-shares of posts from outlets.

Importantly, we must note that the ban does not restrict the sharing of news-related content; rather, it is a ban on external links to the websites of media organisations. By disproportionately targeting mainstream news publishers, less reputable sources fill the gap, amplifying misinformation and hindering access to authoritative information. Analysis by ISD revealed that 80% of sources categorised as “high quality” were blocked following the news ban, whereas only 36% of the sites rated as “low quality” faced similar restrictions (see Methodology for details on scoring). Just ~6% of the outbound links users shared directed to high-quality news.

This policy appears to be creating incentives for, and expanding the reach of, partisan analysis (“influencers”/commentators, lower quality outlets, etc.) at the expense of authoritative information. With research suggesting Facebook may be facilitating the narrowing of discourse into echo chambers, the policy risks exacerbating polarisation of opinion on the conflict by steering users away from factual reporting towards partisan interpretations.

Key Findings

The news ban disproportionately affected high quality news sources, with 80% of high-quality sources blocked compared with 36% of sites rated as low quality.
Only 6% of outbound links shared by users directed to high-quality news.
Users responded to Meta’s news ban by sharing photos and content from other platforms in greater volumes than in previous conflicts or similarly sized countries where news links are not banned.
This content tended to feature more out-of-context, bite-sized information, such as screenshots of headlines.
Content shared from other platforms also contained a greater proportion of opinion, propaganda, and misinformation compared to links from traditional media outlets or re-shares of posts from others.

Methodology

Data Collection

ISD collected over 16k posts from publicly available Canadian Facebook groups using English and French keywords related to the Israel-Hamas conflict (see Appendix). While we did not and cannot collect information from private individuals, Facebook groups were chosen as the closest approximation for how Canadian users on Facebook engage with news about the Israel-Hamas conflict. Canadians are among the most active Facebook users in the world with 29.4M users in 2022– 77% of its population. Analysts used CrowdTangle’s Local Relevance feature, which showed content relevant to communities in Canada, by looking at the geographic distribution and density of group followers.

Data was collected for the period immediately following Hamas’ terrorist attack on Israel on 7 October 2023 until 7 March 2024, to capture the window in which users were likely to be actively seeking news about the conflict and its development.

To include data from another country for comparison, as well as another event prior to the August news ban, ISD collected data from Australia using the same keywords and posts about the Russia-Ukraine conflict. Like Canada, Australians are active Facebook users, with 16.9M users in 2022– 65% of its population. Posts were collected from 24 February 2022 – 24 July 2022, covering a five-month period immediately following the initial Russian invasion in which users were likely to be actively seeking news about the conflict, similar to the Israel-Hamas data collection timeframe. Roughly 80k posts in total were collected and analysed about both conflicts from each country.

Data Analysis

ISD analysed both the type of content (e.g. status updates, photos, videos) and external links, to understand the impact of the news ban on the nature of content being shared on Facebook.

The 50 most shared domains for both Canada and Australia about the Israel-Hamas and Ukraine-Russia conflicts were manually coded to assess the trustworthiness of information, focused on whether a source was high quality, low quality, another platform, or other (namely URL shorteners/ redirectors and petition/fundraising sites).

To determine whether a source was high or low quality, ISD:

Analysed the 10 most engaged URLs from each domain. The presence of any false or misleading content or information presented in a highly-biased manner led to a low quality rating.

An example of misleading content includes reporting that Prime Minister Benjamin Netanyahu acknowledged the 7 October attack was a false flag to justify a genocidal attack against Palestine, while an example of biased content includes a piece labeled “Insight” describing Canada’s Liberal government as “GUTLESS SCUMBAGS” for allegedly “betraying” Israel.

Also, we considered whether the links had clear labels if they were editorial or opinion pieces, as well as if outlets were transparent around funding and leadership.

Scoring was handled by two analysts, who discussed any discrepancies in their ratings until an agreement was reached.

Story selection (i.e. consistently reporting news that favors point of view or ideology) was not considered as a factor. Therefore, outlets like Al Jazeera or The Times of Israel, which disproportionately reported headlines favorable to Palestinians and Israel, respectively, were still categorised as high quality given their factual reporting with neutral language.

Changes in News Consumption

Decline in Credible Coverage

Meta’s news ban contributed to a lower volume of content from high-quality news outlets (e.g. Reuters) and created space for a higher volume of content from other social media platforms (e.g. YouTube), in contrast to parallel datasets from Australia during the same period and compared to news about the Ukraine-Russia War during its early phases. While high-quality news comprised just 6% of the Israel-Hamas dataset for Canada, it comprised 34%-41% of the other datasets (Israel-Hamas data set for AUS, and Russia-Ukraine for both CA and AUS; see Figure 1 below).

Shares of content from other platforms, meanwhile, reached 73% in the Canada Israel-Hamas dataset, compared to ~40% for the other datasets. Notably, some of this content included news coverage, such as a post linking to Toronto Star reporting via a redirect link generated through X (Twitter; a workaround of the news ban). However, of that 73% linking to other platforms, in a randomised sample of 100 posts from the same Canada Israel-Hamas dataset, only 21% linked to high-quality news providers. Therefore, in terms of numbers, a decrease in high-quality sources has been registered since Meta’s news ban, even when considering this workaround being used.

Figure 1: Percentage of shares from various news events in Canada and Australia. Data collected from Facebook groups 7 October 2023 – 7 March 2024, on the Israel-Hamas conflict and from 24 Feb 2022 – 24 July 2022, on the Ukraine-Russia conflict.

Shares of Substack and Rumble content about the Israel-Hamas war grew significantly in Canada in proportion to the Russia-Ukraine dataset (+467% and +46%, respectively). These platforms have relatively lax content moderation and have become known as hotbeds for extremist discourse and misinformation in recent years. For example, the most shared Substack post about the Israel-Hamas conflict in the Canadian dataset included unsubstantiated claims that the 7 October attack was a “premeditated psyop” and a quote from conspiracy theorist David Icke who claimed it was a “false flag.”

With content from other platforms– namely YouTube, Substack and X (Twitter)– heavily consisting of analysis of the Israel-Hamas conflict, it is possible that Meta users in Canada were left less informed and more susceptible to influence from highly partisan sources and propaganda about the war. For example, one of the most shared videos came from the Israeli Prime Minister’s YouTube account, framing the Palestinian Authority as violent and alleging that children were being brainwashed to hate Jews.

Shift in Content Type: Less Links, More Photos

Compared to the datasets from Australia and surrounding the Ukraine-Russia War, the Canadian Israel-Gaza sample contained notably less sharing of links, such as news articles, and a corresponding increase in the sharing of photos, typically bite-sized news blurbs and partisan talking points. Consequently, Canadian Facebook users are seemingly at risk of being less informed about the conflict than users in other countries, on account of much of the news they share and consume consisting of brief snippets of information lacking context.

Figure 2: Percentage of posts about varying news events, broken down by type (categorisation provided by CrowdTangle). Data collected from Facebook groups ranging from 7 October 2023 – 7 March 2024, on the Israel-Hamas conflict and from 24 February 2022 – 24 July 2022 on the Ukraine-Russia conflict.

Increased Traction for Low-Quality News

While high-quality news sites like The Associated Press and Canadian Broadcaster CBC are blocked by Meta, inconsistent enforcement of the ban means that low-quality sites were disproportionately left untouched. Facebook either failed or chose not to identify them as news sites. This includes highly partisan news outlets like spencerferraro[.]com or counterpunch[.]org, as well as sites known for driving misinformation and conspiracy theories, such as globalresearch[.]ca.

ISD’s investigation, which involved using a VPN and browsing through posts where various news sites were shared onto Facebook, revealed that 80% of sources rated as “high quality” were blocked, compared to only 36% of those rated as “low quality.”

With Canadians unable to share information directly from news sources, ISD found that various redirection tools were being used. One such domain, newsproxy[.]site (now defunct), gained significant traction by using this strategy to circumvent the ban. However, as with news articles, Facebook ended up restricting shares of the site.

Conclusion

Research has indicated that echo chambers are prevalent on Facebook, but they may not necessarily be the driving force of polarisation. However, by disincentivising users from sharing credible news and driving shares of analysis, opinion and content from low-quality sources containing misinformation, the Facebook news ban has reduced the quality of information shared and risks exacerbating polarisation surrounding the Israel-Hamas conflict already prevalent on Meta platforms.

Figure 3: Shares of news from The Guardian and newsproxy[.]site, both blocked under the ban.

Figure 4: Videos from Rumble shared in Canadian Facebook groups regarding the Israel-Hamas conflict.

Appendix

English Query – Israel-Hamas

(israel OR israels OR israeli OR Israelis OR idf OR hamas OR palestine OR palestinian OR palestinians OR palestines OR gaza) AND (conflict OR war OR terrorism OR terrorist OR terrorists OR hostage OR hostages OR killed OR kill OR dead OR injured OR injuries OR fighting OR ceasefire OR genocide OR violence OR violent OR operation OR operations OR exodus OR refugee OR refugees OR “united nations” OR UN OR peace OR crisis OR humanitarian OR airstrike OR airstrikes OR bomb OR bombs OR bombing)

French Query – Israel-Hamas

(israël OR israélien OR Israéliens OR idf OR hamas OR palestine OR palestinien OR palestiniens OR gaza) AND (conflit OR guerre OR terrorisme OR terroriste OR terroristes OR otage OR otages OR tué OR tuer OR mort OR blessé OR blessures OR combat OR “cessez-le-feu” OR génocide OR violence OR violent OR opération OR opérations OR exode OR réfugié OR réfugiés OR “nations unies” OR ONU OR paix OR crise OR humanitaire OR “frappe aérienne” OR “frappes aériennes” OR bombe OR bombes OR bombardement)

English Query – Ukraine

(ukraine OR ukrainian OR ukrainians OR kiev OR Russian OR Russia OR Russians OR Moscow) AND (conflict OR war OR killed OR kill OR dead OR injured OR injuries OR fighting OR ceasefire OR genocide OR violence OR violent OR operation OR operations OR refugee OR refugees OR “united nations” OR UN OR peace OR crisis OR humanitarian OR airstrike OR airstrikes OR bomb OR bombs OR bombing OR invasion OR invade OR invading OR invaded OR attack OR attacked OR attacking)

French Query – Ukraine

(ukraine OR ukrainien OR ukrainiens OR kiev OR russe OR russie OR russes OR moscou) AND (conflit OR guerre OR tué OR tuer OR mort OR blessé OR blessures OR combat OR “cessez-le-feu” OR génocide OR violence OR violent OR opération OR opérations OR réfugié OR réfugiés OR “nations unies” OR ONU OR paix OR crise OR humanitaire OR “frappe aérienne” OR “frappes aériennes” OR bombe OR bombes OR bombardement OR invasion OR envahir OR envahissant OR envahi OR attaque OR attaqué OR attaquant)