… is practice.
We have a rule in the lab that any demonstration must first be trialed at least 5 times without error before being made public.
Why 5? 5 is a common threshold to begin using statistical methods to establish confidence (a.k.a. coverage factors and uncertainties).
In science, a single test can show a phenomena, verify a theory, and increase understanding of the Universe.
Engineering though, doesn’t care so much about understanding, as much as reproducibility. That that something, based on science, can be done over and over again, including by the public, without fail.
A key difference between science and engineering is practice. And that practice leads to confidence. If it didn’t take practice, it’s not engineering.