This page provides project updates for the Australian Search Experience project.
To join the project visit the main To find out more about The Australian Search Experience and to join the project visit the main webpage here
3 May 2022
Published on the Analysis & Policy Observatory (APO) webpage, this background paper outlines the motivations for this study, presents the broader context of the research, introduces prior work by our research partner AlgorithmWatch, and describes the practical implementation of the Australian Search Experience project. Outcomes from this research will be presented in scholarly publications and public updates throughout the lifetime of the project.
30 September 2021
Initial Observations presented at the ADM+S News and Media Symposium
Our analysis of the hundreds of millions of search results donated by the more than 1,000 citizen scientists who have installed the browser plugin so far is still ongoing, and we will release detailed results from this research at a later stage. However, at the ADM+S News & Media Symposium on 30 September 2021 we presented the following very preliminary observations about key patterns in the dataset. You can also watch the full presentation below.
Very different patterns across platforms Google Search, Google News, Google Video and Youtube
- Google search largely stable
- Google news very fast-moving
- Google Video quite static
- YouTube often stable in top ~5 results, then highly changeable.
- Detailed analyses per platform, and cross-platform comparisons.
- Further breakdown by demographic attributes and browser features
- Extension to non-organic search results
- Evaluation of search result quality
Limited evidence of search personalisation so far
- Personalisation in Google Search largely driven by user location
- Critical search topics appear manually curated, at least in part
- Some results differences based on browser type (desktop vs mobile)
- Other platforms (especially YouTube) need further detailed analysis
- Demographic profile of citizen scientist cohort unrepresentative
- Need to compensate for participant attrition over time.
- Gradual variation of search terms to address emerging topics
- Focus especially on major upcoming events – e.g. federal election
This project is being conducted in collaboration with partner AlgorithmWatch.
AlgorithmWatch is a non-profit research and advocacy organization that is committed to watch, unpack and analyze automated decision-making (ADM) systems and their impact on society.