I will sometimes start with a zone file, like com.zone. I search for keywords in registered domainnames. Then I filter by nameserver (registrar). Finally, I run a script to fetch the page titles. You would be surprised at how effective this can be in finding websites that you would never be able to find using simple Google searches. Of course, it is cumbersome; search engines can make this process very easy but they deliberately disable this type of exploration. You can query Google's index for a list of all websites with domain names that contain a certain keyword, but you will never be able to retrieve the full list of results, and certainly not in a "neutral" order such as alphabetical.

Arguably a web comprising a large number small, diverse websites, where each user may be visiting a variety of different websites, is less suitable for advertising than one where all web users are funneled through a few large websites that survive by selling online ad services, like Google. It stands to reason that those large, online ad services sites would have little interest in showing users an undiscovered portion of the web. They want users to congregate on "popular" sites. Good for advertising.

OTOH, using zone files instead of a search engine, social media or news aggregator site in the online ads (or VC) business, one can see all websites that have registered an ICANN domain name. No filters. No advertising-related algorithms. Popularity is irrelevant. The user determines relevance, not a third party.