Of all the amora’im, Abaye and Rava are presented as epitomizing the discussions that take place in the Gemara. In all of their arguments in the Gemara, the halakha always follows Rava’s opinion, with only six exceptions. Those six are referred to by the Gemara by the acronym YAL KGM:
Ye’ush shelo me-da’at (Massekhet Bava Metzia 21b) – When a person does not realize that he has lost an object until after it is picked up by someone else, and he gives up ownership when he realizes it, can we apply it retroactively?
Aid zomem lemafrei hu nifsal (on our daf today) – When witnesses are found to be unreliable, does their status change as of that time or from the time that their false testimony was made?
Lehi ha-omed me-elav (Massekhet Eiruvin 15a) – When a pole is standing on its own (it was not placed there by a person), can it be used as part of the structure that will create a legal private domain on Shabbat for the purpose of carrying?
Kiddushin she-lo nimseru le-bi’ah (Massekhet Kiddushin 51a) – If a couple gets married, but will be unable to consummate their marriage, does it have legal significance?
Gilui da’at be-giṭṭin (Massekhet Gittin 34a) – If a person makes a statement that can be understood as affecting the divorce that has been sent to his wife, must we take it into consideration when ruling on the validity of the divorce?
Mumar la’avor aveira le-hakh’is (Massekhet Sanhedrin 27a) – Can a person who commits sins be trusted to testify in court?
The case that is discussed in our Gemara is when witnesses are found to be zomemim – a second set of witnesses comes to court and testifies that the first group could not be telling the truth, since they were with them in a location far removed from the incident at the time in question. Abaye believes that any testimony that they gave from the time of their original statement can no longer be trusted, since from that time it is clear that they were unreliable. Rava argues that the whole concept of zomemim is a hiddush – a new idea – established by the Torah, since logically there is no reason to trust the second group of witnesses more than the first. Therefore we implement the law only from the moment that it can be applied, i.e. from the time that the second testimony was given.