Deciding Whether to Adjust the Home Feed After Complaints About Climate Summit Posts

A climate summit is taking place this month, and members of a professional social network are seeing many climate-related posts in their home feed. Some members have reported that they do not want to see these posts. How would you determine whether the platform should take action and adjust the home feed? Assume the feed is ranked and mixes posts from accounts the member is connected to or follows with posts the ranking system surfaces, such as posts the member's network engaged with and recommended posts from outside the network. Assume the feed's negative-feedback actions (hiding a post, "not interested", unfollowing or muting an author, reporting) are logged per impression. Explain what data you would examine, how you would tell a real problem from a vocal minority, and what evidence would lead you to recommend acting or not acting. ```hint Size it before judging it A handful of complaints is a prompt to measure, not a verdict. Ask how many members are affected, how concentrated their exposure is, and how this month compares with an ordinary one. ``` ```hint Where the posts come from Ask whether the climate posts reach members through their own network or through the ranking system, and what each answer implies about what the platform controls. ``` ```hint Stated versus revealed Compare what complaining members say with what members actually do when a climate post appears in their feed. ``` ### Clarifying Questions - How many complaints came in, through which channel, and how does that compare with the usual complaint volume? - Is the objection to the topic itself, to the number of climate posts, or to their quality, for example repetitive or misleading posts? - Are the climate posts mostly from members' own connections and follows, or mostly surfaced from outside their network? - Which actions are on the table: a ranking change, a member-facing control, or a content-policy response? ### What a Strong Answer Covers - Framing the decision: separating a volume or relevance problem from a content-quality or policy problem, and naming the possible actions, including doing nothing - Sizing the complaints and the climate exposure against a baseline, with attention to how concentrated the exposure is rather than only its average - Behavioral signals of dissatisfaction beyond complaints, measured per impression and per member, segmented by members with and without interest in the topic - A diagnosis of whether ranking amplification or members' own networks drive the exposure, and the causal pitfalls in comparing heavily and lightly exposed members - A decision rule agreed in advance, and an experiment with guardrails for members who value the content and for the creators posting it - Recognition that the surge is tied to a time-limited event, and what that means for the urgency and the shape of any fix ### Follow-up Questions - If you test a limit on how much of one topic a member's feed can show, what are the primary metric and guardrails, and how do you get a useful readout before the summit ends? - How would you identify climate-related posts at scale, and how would classification errors bias your analysis? - Many members who complain still react to and comment on climate posts. How do you weigh their stated preference against their behavior? - How would you turn this case into a general way to detect and handle event-driven topic surges?