Remix.run Logo
▲ teraflop 2 hours ago

I don't think anyone's disputing that Dewey Decimal can place "related" books in different places. The question is whether this new proprietary system improves on it.

Like, consider the parent post's example of "a book about the history of gardening in England". That covers at least two topics, gardening and English history. If you choose to group all the gardening books together in one place, then you force the English history books to be scattered instead. This is just an inherent limitation of reducing a multi-dimensional "topic" space to a linear ordering.

The meaningful question is whether on average, this system is a more useful grouping than alternatives, based on how well it captures the "related-ness" that people actually care about, and how difficult it is for them to find what they're looking for. You can't answer that question by looking at the classification hierarchy alone. It's an empirical question about how users interact with the system.