Very surprised to see that reproduction cost was <<10% of total budget. I would assume that this would only hold for small scale biology with low variable costs and standard equipment like in vitro cell biology and computational work. A different financial approach is likely necessary for setups with high variable and fixed costs (e.g. animal and human work). Not saying it shouldn't be a required part of the funding but the speed of science changes if you need to spend 50% of your budget on reproduction.

Off-topic, but does anyone know if HN's dupe detector changed recently? I submitted this about an hour ago (mainly to check whether it was already submitted) but it spawned a new thread instead of upvoting this one: https://news.ycombinator.com/item?id=22541423

It's not so hard to click on "past" and then click on a recent thread and upvote it myself, but it was a nice feature. Using the HN bookmarklet to check "was this recently submitted?" was a nice shortcut to get to recent active threads directly from a URL that you're pretty sure will be on HN.

This is great. One side-effect that might be missed is that people who know their work is going to be replicated are not going to fudge their results, so simply by existing this acts as a filter.