The repeatability of results is considered to be a test for the veracity of any formula or theory. The Gettier problem can be resloved by following this empirical method of repeatbility of results. However before doing this, an explanation should be in order as to what it means to believe in something with a justification. A hasty generalization should not constitute a belief. Coincidences do not constitute knowledege and it was just a matter of chance that the time shown by the watch was actually true. These coincidental truths give a semblance of knowledge and not true knowledge. A counterexample in this case would be the repeatability test.
If I were to look at the watch some time later again, then I would have known my earlier knowledge to be just a mere coincidental truth. There is always an element of probability involved whether the clock is showing current time or not. One can not consider any coincidental truth to be knoldge without first eliminating the element of chance that may be present in the case. These counterexamples are thus valid in all cases where there are elements of probability involved. Performing a simple repeatability of result would in most cases reveal whether one is jutified in having a coincidental truth to be true knowledge or not. To classify anything as “knowldege”, one must therefore consider all the elemnts of probability at play besides considering the three cirteria of justified true belief. Any framework that takes into acount the element of chance would reveal whether any observation is just a coincidental truth or knowledge. The definition of knowledge as a “justified true belief” is otherwise fine for many cases but to know something for sure in all cases, one must first eliminate the elements of chance through repeatablity of results.
Lemos, N.M. (2007). The traditional analysis and gettier problem. An introduction to the theory of knowledge. Cambridge University Press, 2007. 22-43.