Algorithmic Misjudgement in Google Search Results: Obscuring Locality-Specific Electoral Information

Core Concepts

The search algorithm frequently omits and mistargets non-federal government websites, especially at the local level, obscuring the important role these authoritative sources play in providing voters with relevant electoral information.

Abstract

The study audited 3.45 million Google Search results across 786 locations in the US to examine the representation of government-maintained websites in the online environment of electoral information during the 2022 midterm elections.
Key highlights:

40.6% of the 4,556 unique domains that appeared in the organic search results were government websites, contributing 39.97% of all organic results.
However, a small subset of 10 non-federal government domains accounted for 35.96% of all non-federal government domain appearances, with high rates of mistargeting.
Overall, 71.18% of organic results containing non-federal government websites were mistargeted, with county and local government sources suffering the highest mistargeting rates at 91.94% and 76.08% respectively.
The search algorithm showed inconsistency in its treatment of locality-specific government sources, leaving most government domains, especially at the local level, appearing minimally, inconsistently, or missing entirely from election-related search results.
The authors frame this omission and frequent mistargeting of non-federal government domains as a form of algorithmic misjudgement that obscures the important role these authoritative sources play in providing voters with relevant electoral information.

Stats

"40.6% of the 4,556 unique domains that appeared in the organic search results were government websites, contributing 39.97% of all organic results."
"71.18% of organic results containing non-federal government websites were mistargeted."
"County and local government sources suffered the highest mistargeting rates at 91.94% and 76.08% respectively."

Quotes

"We frame the omission and frequent mistargeting of non-federal government domains as a form of algorithmic misjudgement that obscures the important role that these domains play in providing official, locally-relevant information to voters."
"While correctly-targeted government sources are usually ranked higher than those that are mistargeted, 71.19% of government website occurrences were mistargeted."

Key Insights Distilled From

Algorithmic Misjudgement in Google Search Results

by Brooke Perre... at arxiv.org 04-09-2024

https://arxiv.org/pdf/2404.04684.pdf

Deeper Inquiries

How can the search algorithm be improved to better surface relevant, locality-specific electoral information from authoritative government sources?

To enhance the search algorithm's ability to surface relevant, locality-specific electoral information from authoritative government sources, several improvements can be implemented:

Geotargeting Refinement: The algorithm can be refined to improve geotargeting accuracy by incorporating additional signals such as user location data, search history, and context. By leveraging more precise geolocation data, the algorithm can better match users with relevant local government websites.

Contextual Understanding: Enhancing the algorithm's understanding of context can help prioritize locality-specific information. By analyzing the search query, user intent, and the nature of the information sought, the algorithm can better identify when locality-specific government sources are most relevant.

Authority Recognition: Implementing mechanisms to recognize and prioritize authoritative government sources can ensure that users are presented with trustworthy and reliable information. By assigning higher weights to government domains in the ranking algorithm, the visibility of these sources can be increased.

User Feedback Integration: Incorporating user feedback mechanisms can help the algorithm learn and adapt over time. By allowing users to provide feedback on the relevance and accuracy of search results, the algorithm can continuously improve its ability to surface locality-specific electoral information.

Transparency and Accountability: Providing transparency into the algorithm's decision-making process and ensuring accountability for the surfacing of authoritative government sources can build trust with users. By clearly communicating how search results are generated and highlighting the importance of government sources, users can have confidence in the information presented.

By implementing these improvements, the search algorithm can better serve users seeking locality-specific electoral information from authoritative government sources, ultimately enhancing the civic information environment.

What are the potential civic harms caused by the algorithmic misjudgement identified in this study, and how can they be mitigated?

The algorithmic misjudgement identified in the study can lead to several potential civic harms, including:

Misinformation and Disinformation: By mistargeting locality-specific government sources, users may be exposed to inaccurate or misleading information, leading to misinformation and disinformation about elections. This can undermine trust in the electoral process and democratic institutions.

Voter Suppression: If users are unable to access accurate and relevant electoral information from their local government sources, it may result in voter suppression. Lack of access to essential voting logistics and candidate information can deter individuals from participating in the electoral process.

Erosion of Civic Engagement: When users are not presented with locality-specific government sources in search results, it diminishes their engagement with the civic process. This can lead to apathy, disengagement, and reduced participation in elections.

To mitigate these potential civic harms, several strategies can be employed:

Algorithmic Accountability: Implementing mechanisms for auditing and monitoring the algorithm's performance in surfacing government sources can help identify and rectify misjudgements. Regular audits can ensure that the algorithm is functioning as intended and surfacing relevant information.

User Education: Educating users about the importance of seeking information from authoritative government sources can help mitigate the impact of misjudgement. Promoting awareness about reliable sources and encouraging users to verify information can empower them to make informed decisions.

Collaboration with Government Agencies: Collaborating with government agencies to optimize the visibility of their websites in search results can enhance the accessibility of accurate electoral information. By working closely with election authorities, search engines can ensure that users have easy access to reliable sources.

By taking proactive measures to address algorithmic misjudgement and its potential civic harms, search engines can contribute to a more informed and engaged electorate.

In what ways might the findings of this audit apply to the online information environment around elections in other countries, where the governance structures differ from the US?

The findings of this audit can have implications for the online information environment around elections in other countries with different governance structures. Here are some ways in which the findings may apply:

Localization Challenges: Similar challenges with geotargeting and surfacing locality-specific information may exist in other countries where governance structures are decentralized. Countries with federal or regional election administration bodies may face similar issues in ensuring the visibility of relevant government sources.

Trust in Government Sources: The importance of surfacing authoritative government sources in search results is universal, regardless of governance structures. Ensuring that users have access to reliable electoral information from official sources is crucial for fostering trust in the electoral process.

Algorithmic Transparency: The need for algorithmic transparency and accountability in surfacing electoral information is a global concern. Countries worldwide can benefit from implementing measures to audit and monitor search algorithms to ensure the accuracy and relevance of search results.

User Engagement: Promoting user engagement with government sources and encouraging the use of reliable information can be a common goal across different countries. Strategies to enhance user awareness of trustworthy electoral information sources can be applicable in diverse contexts.

By recognizing the broader implications of the findings of this audit, countries with varying governance structures can learn from the challenges and opportunities identified in the study to improve the online information environment around elections.