I disagree somewhat with your last point. In clinical trials, negative as well as positive results must be reported. When a drug or other therapy doesn't work in a subject, that data cannot be excluded per clinical trial protocol.

In the basic and applied sciences, one problem is that failed methodologies are rarely published. It's a shame, too, as this information is useful to others investigating a specific question and could save much time. Unless one serendipituously comes upon another successful methodology while investigating the failed one, any manuscript they submit will be rejected by reviewers. It is the current culture in the basic and applied sciences.