Try Things is a mantra that reminds you to gather data about what works by experimentation. If a proposed intervention works and you learn that it works, you've just gained something that you'll be able to use over and over again. If it doesn't work, you've only had it not work once.
To the knowledge of one editor of this article, the phrase was introduced to the LessWrong community by CFAR, at least as early as 2016.
See also: Value of Information