The two-party vote: Reporting election misinformation on social media platforms

Stijn Peeters
Abdoul Yerbanga, Rong Sun

Introduction

In the lead-up to the Dutch parliamentary elections on 28 October 2025, a meme circulated regarding how one was to cast a vote for GroenLinks-PvdA – a new party emerging from a merger of GroenLinks and PvdA, two established left-wing parties. The meme claimed that to cast a vote for the party, one would need to fill two boxes on the ballot paper – one for both parties. In fact, this would make the vote invalid, as it is only counted if exactly one box is filled.

The meme is a version on an older ‘meta-joke’ regarding the voting process; another popular variation alleges that the elections have been spread over two days; one day for the side of the political spectrum the teller of the joke supports, and another one for the other side. The latter would then, of course, be at a later, non-election day. In many cases these jokes will be understood as such or as satire, and those who spread it may often do so with this expectation, rather than in an attempt to mislead voters.

However, in this case the joke had a kernel of plausibility, given the fact that GroenLinks-PvdA truly is a merger of two parties that could previously be voted for separately. Combined with a broader concern about election-related misinformation, this prompted a number of ‘fact checks’ debunking the meme. According to reporting by AFP, an international news agency, the Dutch Ministry of Internal Affairs also contacted social media platforms to ask them to flag such content due to its potentially misleading nature.

This followed earlier, similar requests from the Ministry during the previous election cycle in 2023, when a version of the same meme was circulating. In both cases, the Ministry used its status as a ‘trusted flagger’ – a designation introduced by the EU’s Digital Services Act that requires platforms to treat reports with higher priority. Under the DSA, a trusted flagger can flag content but does not have the power to make moderation decisions; the decision of whether something is to be moderated always remains with the platform.

The AFP report mentions two specific posts that were flagged this way; one on Facebook, one on X. The Facebook post is no longer available (visiting it shows a generic "This content isn't available right now" note). The X post is available, but has been flagged by the platform with a notice saying “Visibility limited: this post may violate X’s rules against Civic Integrity” (see Figure 1 for a similar post with the same label).

Figure 1. A post on X containing the ‘double vote’ meme, labeled as having its visibility limited and potentially violating X’s policies. Post text has been paraphrased.

As such, the meme can be considered ‘reportable’; it is an example of content that violates the platforms’ policy, which has been reported as such, and indeed flagged or moderated by the platform. Nevertheless, various other versions of the same meme – sometimes simply copy-and-pasted from the original – were also circulating, and had not been reported to the platform or in any case not been flagged. As these would be in violation of the same policies, the goal in this case study was to confirm whether reporting these posts via another reporting mechanism – the one available to ‘ordinary’ users – would likewise result in moderation.

Platform policies

To do so, we looked at three platforms; X, Instagram, and Facebook. X and Facebook have been shown to consider the meme in question worthy of moderation when flagged. Instagram, as a platform with the same owner as Facebook (i.e. Meta) may be expected to have a similar content moderation policy.

For all three platforms, we collected 10 posts published in October 2025 containing (a variation of) the ‘vote twice’ meme. For X, we collected a second set of posts from 2023, as previous reporting indicates that the meme was considered reportable back then as well, and we were interested in whether X would moderate the posts ‘after the fact’.

All three platforms explicitly disallow election-related misinformation:

• Facebook claims on its ‘Our approach to elections’ page that “[w]e remove content that attempts to interfere with voting, such as incorrect voting information”. Additionally, on its page describing ‘community standards’ concerning Misinformation, it notes that “we remove misinformation that is likely to directly contribute to a risk of interference with people’s ability to participate in those processes” including “misinformation about the dates, locations, times, and methods for voting, voter registration, or census participation”.
• Instagram has mostly adopted the same policies as Facebook; links to its ‘Community guidelines’ point towards the same documents as those that apply to Facebook, which are collected in the ‘Meta Transparency Center’. As such, Instagram is considered to follow the same policies as Facebook.
• X, in a section in its policy documents titled ‘Civic integrity’, notes that it does not allow “advancing verifiably false or misleading information about how to participate in an election or other civic process” including “misleading claims that cause confusion about the established laws, regulations, procedures, and methods of a civic process”, where political elections are explicitly noted as an example of such a civic process. This policy document is also linked to in the warning label added to the post that was mentioned by the AFP as having been reported by the Ministry, suggesting that it is indeed the relevant policy here.

Method

The goal was to report posts that re-used the meme that had been flagged according to the AFP report verbatim, or used a close variation of it. This included the use of a photo of a ballot paper (see Figure 1). To find such posts, a mix of query strategies was used. The platform’s own search feature could be used to find posts containing relevant phrases such as “twee kandidaten” or “twee bolletjes”. On Instagram, which has no platform-wide text search feature, Google was used with a site search query, e.g. [twee bolletjes site:Instagram.com].

While election-related misinformation is defined quite specifically as not allowed on all three platforms, they do not actually offer a way to report content as violating this specific policy. All three platforms’ reporting mechanisms – in any case those available to an ordinary users – only allow reporting within broader categories such as ‘Privacy’ or ‘Child Safety’. The most relevant reporting pathways were as follows:

• Facebook: Report Post → Scam, fraud, or false information → Shares false information
• Instagram: Report → False information
• X: Report Post → Spam

When used in the EU, X offers an additional reporting link, titled ‘Report EU illegal content’. In contrast with the ‘ordinary’ reporting option, following this link takes the reporter to a form in which more specific information can be provided to motivate the report. Here we chose to report the posts for ‘Negative effects on civic discourse or elections’, after which a more specific subcategory can be chose. Two of these were equally relevant, and so five posts were reported for ‘Violation of EU law relevant to civic discourse or elections’ and five for ‘Misinformation, disinformation, foreign information manipulation and interference’.

The reporting pathways used in the end are visualized in Figure 2. The goal was to test each of these with a set of 10 posts containing the ‘double vote’ meme, and then observe 1) whether any response to the report would be sent by the platform and 2) whether the reported content would be removed, labeled, or otherwise moderated.

All posts were reported on 23 October 2025, one day before the elections. We then monitored the reported posts as well as any responses to our reports from the platforms for a period of four weeks, until 25 November 2025.

Figure 2. Reporting mechanisms used.

Results

In summary, the reporting had no clear effects. Posts were checked one week after reporting and then again four weeks after reporting.

• On Instagram, all posts are still online, with no warning labels or other indications that action was taken. While Instagram’s help pages claim that the status of reported posts can be followed in the website’s Settings page, no records of the reports could be found here.
• On Facebook, some posts had been labeled or disappeared one week after reporting (Table 1). However, those that had been labeled noted that they had been flagged by dpa-Faktencheck, a fact-checking outfit part of the Deutsche Presse-Agentur that works with Meta directly; it is thus unlikely that the posts had been labeled following our reporting. We received no responses to or notices regarding our reports.

Status as of 25 Nov 2025	n	Notes
Available	5
Available, but hidden with a flagging notice	2	Flagged by ‘dpa-Faktencheck’, a trusted third-party flagger
Unavailable	3	Displays ‘This content isn't available right now’. No indication whether it was removed, hidden by the owner, or moderated.

Table 1. Status of reported items on Facebook as of 25 November 2025 (four weeks after reporting).

 

• On X, a single post from October 2025 had been labeled as potentially violating X’s rules (see Figure 1). Two other posts were no longer available as its author’s account had been suspended from the platform (i.e. all their posts had disappeared, not just the one reported). It is unclear if the labeling of posts or account suspensions (see Table 2) were a result of our reporting. That many identical or near-identical posts that we also reported in the same way are still available suggests that this is not the case. For 11 out of 20 post reported with the ‘EU illegal content’ feature, we received a rejection of the report within an hour after reporting. For the other reports, we received no update or response.

Status as of 25 Nov 2025	n	Notes
Posted in 2025, reported via standard reporting features
Available	10
Posted in 2025, reported via the ‘Report EU illegal content’ feature as misinformation
Available	4
Labeled as violating rules	1	Post had 22 views and 1 like.
Posted in 2025, reported via the ‘Report EU illegal content’ feature as breaking EU laws
Available	4
Author account suspended	1
Posted in 2023, reported via standard reporting features
Available	9
Author account suspended	1
Posted in 2023, reported via the ‘Report EU illegal content’ feature as misinformation
Available	5
Posted in 2023, reported via the ‘Report EU illegal content’ feature as breaking EU laws
Available	4
Author account suspended	1

Table 2. Status of reported items on X as of 25 November 2025 (four weeks after reporting).

Discussion

The results suggest that reporting content, even if it has been demonstrated to violate platform rules, has no clear effect if the reporting is done via the user reporting mechanism. There are obvious limitations to this analysis’ approach that require some reservations regarding this claim. The reports concerned variations on a single post; platforms may more readily act on reports on other types of content, such as CSAM or hate speech.

Nevertheless, all platforms included have clearly banned election misinformation and express a concern for election integrity in their policy documents. That content which violates this policy – judging by the platforms’ own actions – is then not moderated after reporting, the day before the election this misinformation concerns, is cause for concern.

The sample size of this analysis – 10 posts per reporting method – was relatively low, but since all 10 posts are clear and close variations on a demonstrably ‘reportable’ post, it is unlikely that a higher sample size would have led to different results. If the platforms would consistently follow their own policies regardless of reporting method, one would expect these posts to have been moderated, regardless of how many were reported.

That this did not happen indicated that either these reports are reviewed but not considered actionable; or that the reports were automatically rejected or ignored. That similar posts were labeled because of flagging from other sources (the Dutch Ministry of Internal Affairs), and dpa-Faktencheck) indicates that this content is indeed reportable – but clearly, it matters who is doing the reporting.

Conclusion

We find that on various social media, reporting content that has been proven to be ‘reportable’ - that is, other instances of the same content have been flagged or removed by the platform - are not flagged or removed after reporting. This suggests that while such content can be and is occasionally moderated by the platform, such moderation does not happen (primarily) because of ‘crowd-sourced’ user reports, but rather via other mechanisms such as third-party fact checkers, ‘trusted flaggers’, or algorithmic moderation.

Having reported proven reportable content through a variety of reporting mechanisms on three different social media platforms, it seems likely that such user reporting is largely ineffective or in any case not a viable mechanism for flagging such content in a timely manner. The content that was reported here – misinformation concerning voting methods for the 2025 Dutch parliamentary elections – violated all these platform’s policies, and was reported the day before these elections. It is nevertheless largely still online four weeks later, and insofar as posts were labeled or taken offline, this seems to have been due to other factors such as reports from third-party fact checkers or the suspension of the post author’s account for other reasons.

While the content discussed here – a humorous meme – is perhaps not a significant danger to the electoral process, this nevertheless gives cause for worry regarding the platforms’ stated concern about safeguarding civic processes and limiting the spread of misinformation, election-related or otherwise. (Systemic) risks concerning the spread and weaponization of such misinformation have been demonstrated time and time again; that platforms seem reluctant to proactively address this through providing effective reporting mechanisms is thus problematic.

Post scriptum

As of November 2025, X includes a new category for which content can be reported: ‘Civic Integrity’, summarized as ‘misleading content related to voter participation, suppression, or intimidation in elections and other civic processes’. This category, which would have perfectly fit this case study, was not available yet at the time.

Sources

AFP Factcheck Nederland. “Kiezers van GroenLinks-PvdA moeten géén twee vakjes kleuren, dat maakt de stem ongeldig.” AFP Factcheck Nederland. Accessed November 26, 2025. https://factchecknederland.afp.com/doc.afp.com.79MY6NQ.

Facebook. “Can I Check the Status of Something I’ve Reported to Facebook or Cancel a Report?” Facebook Help Center. Accessed November 25, 2025. https://www.facebook.com/help/338745752851127/.

Instagram. “Check the Status of Something You’ve Reported to Instagram.” Instagram Help Center. Accessed November 25, 2025. https://www.facebook.com/help/instagram/484700649522086/.

Meta. “Community Standards.” Meta Transparency Center. Accessed November 25, 2025. https://transparency.meta.com/policies/community-standards/.

Meta. “How Fact-Checking Works.” Meta Transparency Center. Accessed November 26, 2025. https://transparency.meta.com/features/how-fact-checking-works/.

Meta. “Our Approach to Elections.” Meta Transparency Center. Accessed November 25, 2025. https://transparency.meta.com/features/approach-to-elections/.

Ministerie van Binnenlandse Zaken en Koninkrijksrelaties. “Inzet trusted flagger - Rapport.” Rapport. Rijksoverheid.nl, Ministerie van Algemene Zaken, April 8, 2024. https://www.rijksoverheid.nl/documenten/rapporten/2024/03/22/bijlage-10-rapport-inzet-trusted-flagger-status-bzk-tk23.

X. “X’s Civic Integrity Policy.” X Rules & Policies. Accessed November 25, 2025. https://help.x.com/en/rules-and-policies/election-integrity-policy.