A quick check of the stories on the front page suggests this is the case; editorial content and news covering yesterday's events is all /19 (today) whereas today's news and online content (for tomorrow's paper) is all /20 (tomorrow).
Over the years it seems there have been a fair few 'Hacker News for X' (see https://hn.algolia.com/#!/story/forever/prefix/0/hacker%20ne...). Presumably each time people implement the whole site from scratch (users, commenting, ranking, etc). Have there ever been any open-source attempts at creating a generic Hacker News style site that people could self host for specific topics? The only thing that I can think of is self-hosting your own copy of reddit.
From the discussion yesterday on "HN for data scientists"[0], there was mention of a service which enables one to deploy "HN for X"[1], as well as some discussion that (some version of) the code for HN is available[2].
Stamplay is a BaaS similar to Firebase (think of it like IFTTT for backend dev). They open-sourced a HN-clone with their service that gets you up and running super quickly: https://github.com/Stamplay/stamplay-hackernews
Well, reference counting is a form of garbage collection, just not a very good one.
It is incomplete in the sense cycles are not collected, one would need extra machinery for those. It is also not great about throughput, a part of the problem is that it dirties the cache too much. Locking the count (or avoiding it somehow) is another issue. On the other hand for many use cases better latencies are achieved with reference counting than with alternative forms of garbage collection, (but this depends on use cases).
If runtime data cannot have have cycles, garbage is not created at a phenomenal rate and you are quite worried about GC pauses, then reference counting is often a cheap and effective solution.