Given the enormous amount of papers that come out, I personally tend to read papers that come with code (and data) first.

For me, it shows the authors are confident yet also open to critique. Which is a wonderful thing.

Secondly, I usually need the code to really understand the paper.