Interesting take, but you're biggest complaint about vertical search engines seems to be "relevance depends on the sociology of the current moment. " Curation

Interesting take, but you're biggest complaint about vertical search engines seems to be "relevance depends on the sociology of the current moment. " Curation is just a another form of this, and a lot more momentary. You also state "With strong opinions on how to organize information, reflected in their choice of filters, vertical search aggregators have distinct advantages that horizontal software can never achieve." I strongly believe google can do this, structured data was a big push to make this easier for search engine, but as machine learning models expand to extract structured data from unstructured text, giants will be better positioned to do this then vertical search engine. "Spotify doesn’t curate what songs make it to their platform. Instead, it takes the entire universe of music and finds endless ways to discover and search across its library, including a mix of manual curation (via playlists curated by its in-house team of curators and its users) and algorithms (like Discover Weekly)." So would the metrics like whether a user follows more links, after the first one, trys another query, hops back to the search results quickly, a form of curation? And I also think most of your boutique search examples, aren't search engines, but in app searches. A search engine needs to search an index external documents to be a search engine. Looking at the query, "What do NFT collectors think about NFTs?", how you describe curated search engine would be terrible at answering that, because it would give very limited scope of the curators opinions. Google should return all the results containing content where people talk about this, and you can form an objective opinion based on what people are talking about. A better model for answer this type of question would probably be a massive language library like GPT3, continually trained on current information.