To start: I fully support Kagi’s initiative to clean up the web. The influx of "deceptive or low-value AI-generated content" is a problem, and having a default filter that downranks this "slop" is a net positive for the majority of search queries. However, the inability to disable this filtering is concerning for a tool aimed at power users. The announcement blog states that "Our ethos at Kagi is to put humans in control." Yet, by making this filter mandatory, Kagi is implicitly stating that users cannot be trusted to discern quality for themselves. Even with manual human review, "better results" remains a subjective metric that users should be able to audit.
I pay for Kagi to have granular control over my search experience. I should be able to see the search index in its raw state simply because I want to. There is a distinct difference between a "search engine" (which indexes the web) and a "curated feed" (which selects what is good for you). Without an opt-out, Kagi behaves like the latter. Kagi allows users to customize almost every other aspect of the ranking and display (personalized results, lenses, and custom CSS). "Slop ranking" should be treated no differently: a helpful default setting that the user ultimately has the authority to override. Furthermore, the SlopStop documentation notes that users can filter AI media (images/video); this same agency should apply to text/web results.
