I actually think that it's quite fair to criticize a "leading" achievement that falls apart or otherwise fails soon after.

For an achievement to "count", I'd make the bold claim that it has to actually work.

* However, I do agree that often to make progress you absolutely have to push your engineering "comfort zone" and try dangerous things, in order to learn from your mistakes. But this does not change my definition of "success", it just acknowledges realities of R&D.