Building Hacker Datasets

I have been working on Hacker Datasets, a directory of data resources found through Hacker News discussions. The idea grew out of seeing projects such as HackerTube and HackerBooks. Old community conversations contain a lot of useful material that is difficult to discover through the usual feed.

My first instinct was to make the directory much larger. Thirty entries looked too small, so we expanded it, then expanded it again. That brought up a harder question: what should count as an entry?

A dataset can appear under several URLs. A link might lead to a platform, a tool, or an article about the data. Someone mentioning a resource also does not necessarily recommend it. Even counting people becomes less obvious when submissions and comments are mixed together.

I spent more time discussing these questions with AI than I expected. They affect what the site is actually promising its readers.

I also decided that every resource does not need a long introduction before the site can be useful. Basic entries can provide a source and relevant discussions; a smaller selection can get more editorial attention. That makes the project feel manageable again.