Oof, this is tricky…sometimes, I think both categories are great for different use-cases. Here’s an example:
I have some familiarity with the WebSub protocol and plan on implementing it. I’m interested in comparing WebSub Hub implementations to use as a starting point.
I searched for “WebSub Hub”. Normally, I’d expect results to be too broad, tailored to someone who doesn’t know what WebSub is. That was column A and what I’d expect from a search engine; it seemed to interpret the query as “What is a WebSub Hub?”
Column B was more useful to me specifically: it felt tailored towards people already familiar with WebSub and displayed more specific information (blog posts about specific experiences with WebSub, certain WebSub implementations, etc.), eschewing the more generic results. It seemed to interpret the query as “I typed ‘Hub’ for a reason; skip results about WebSub in general, and show me results about specific WebSub Hubs”.
I feel like Column A is what users would expect and probably makes more business sense for an engine trying to compete with Google/Bing. But I also don’t think we need yet another engine tailored to broad, simple questions. There’s an untapped space for engines focused on long-tail queries, with less SEO garbage, that may be much less helpful for introductory questions. An engine for “I already know what this is; I’ve Googled it. Now tell me more about it.” Alternative engines shouldn’t replace existing ones.
(I’m actually entertaining the idea of a sibling engine to Mojeek that also queries Google, and deliberately shows results that Google wouldn’t.)
It’d be cool if Column A was the default, and there was a “long-tail” switch we could flip if we’re already familiar with a topic and would like to see more information about it; specifically, more recent information that combines the topic with other related topics.