In a pure world everyone would use DOIs to refer to published literature. This would give two benefits: first, always link to the current version. Second, be a persistent identifier for the content.
But HN isn't a specialist scholarly content platform, so it's practical to link to the landing page.
And what about countries where URNs are the identifier of choice? In my perfect world, we'd ban DOIs entirely from academic conduct and use only URNs respectively.
In a pure world everyone would use DOIs to refer to published literature. This would give two benefits: first, always link to the current version. Second, be a persistent identifier for the content.
But HN isn't a specialist scholarly content platform, so it's practical to link to the landing page.