Since changing its algorithm in January 2018 to boost the content of family and friends over other content (including news), Facebook has signaled that it is less interested in news. However, the field is still trying to understand the long-term impacts of this change for news publishers. This is a problem because policymakers and legislators across the world are becoming concerned about the relationship between platforms and publishers. In particular, there are worries that platforms’ ability to make unilateral decisions about how their algorithms operate may harm the economic sustainability of journalism. This article provides some clarity around the relationship between these two parties through a longitudinal study of the Australian news media sector’s relationship with Facebook from 2014 to 2020, with a particular focus on the January 2018 algorithm change. We do this by analyzing Facebook data (2,082,804 posts from CrowdTangle) and external traffic data from 32 major Australian news outlets. These data are contextualized by additional desk research. We identify a range of trends including the decline of news sharing, the collapse in the performance of “social news,” the variable position of social media as a source of referral traffic, and, most critically, the diffused nature of the 2018 algorithm change. Our approach cannot make direct causal inferences. We can only identify trends in on-platform performance and referral traffic, which we then contextualize with industry reportage. However, the data provide vital longitudinal insights into the performance and responses of individual media outlets, news categories, and the Australian media sector as a whole during a critical moment of algorithmic change.

自 2018 年 1 月更改其算法以将家人和朋友的内容提升到其他内容（包括新闻）后，Facebook 已表示对新闻不太感兴趣。然而，该领域仍在努力了解这一变化对新闻出版商的长期影响。这是一个问题，因为世界各地的政策制定者和立法者都开始关注平台和出版商之间的关系。特别是，有人担心平台对其算法如何运作做出单方面决定的能力可能会损害新闻业的经济可持续性。本文通过对 2014 年至 2020 年澳大利亚新闻媒体部门与 Facebook 关系的纵向研究，阐明了这两个政党之间的关系，特别关注 2018 年 1 月的算法更改。我们通过分析 Facebook 数据（来自 CrowdTangle 的 2,082,804 个帖子）和来自澳大利亚 32 家主要新闻媒体的外部流量数据来做到这一点。这些数据与额外的案头研究相关。我们确定了一系列趋势，包括新闻共享的下降、“社交新闻”表现的崩溃、社交媒体作为推荐流量来源的可变地位，以及最重要的是，2018 年算法变化的扩散性质. 我们的方法不能做出直接的因果推断。我们只能确定平台性能和推荐流量的趋势，然后我们将其与行业报道相关联。然而，这些数据提供了对个别媒体、新闻类别、