By now, you should know the mantra.

Clear and obvious.

For the NFL to overturn a call, there must be indisputable visual evidence that confirms a mistake with 100 percent certainty. If there is a shred of ambiguity, the original call stands.

In a nutshell, that's why referee Bill Vinovich upheld the Denver Broncos’ game-winning two-point conversion Sunday at the Superdome. The play lifted the Broncos to a 25-23 victory over the New Orleans Saints, even though it appeared that defensive back Will Parks might have stepped on the sideline and, thus, run out of bounds during the return.

When you watch the available replays, there does not appear to be any green turf between Parks' white shoes and the white sideline. But the angles were not perfect. It's possible that there was an unseen gap that would have been visible from above or perhaps from a direct sideline angle.

The NFL's replay system relies on television broadcasters to determine where cameras are placed and how many are installed for a given game. In this case, CBS didn't provide an angle that could have definitively answered the question.

It's fair to wonder why a $13 billion industry can't install its own cameras in stadiums for just that purpose, something New England Patriots coach Bill Belichick has championed for years. I looked into the issue in 2015 and found that it would be more expensive than you think and more complicated than it sounds.

Among other issues, the various stadium configurations around the NFL mean that it would be almost impossible to put cameras in the same place in each building. That would create the kind of imbalance that the league's competition committee tries to avoid.

It's also reasonable to ask why the league has established such a high standard for overturning calls on replay. After all, are referees on the field 100 percent certain about every call they make? Of course they aren't. They make the best call they can based on whatever they saw.

But the NFL long ago decided it did not want to use replay as a tool for reofficiating games from the booth. There is a difference between scanning replays for obvious mistakes and analyzing them for probability shifts. As a result, the NFL typically only overturns about a third of all plays that originate in the booth, as Parks' return did.

Entering Sunday's games, 33 of 96 booth reviews had been overturned (34.4 percent). That's down from 39.3 percent at the end of last season, but relatively close to previous years.

There's nothing more maddening that knowing that a game-winning play might have been called incorrectly. But like it or not, the NFL's replay system isn't set up to address issues of "might." It is intended only to fix cases that were clearly and obviously wrong, and Sunday's play in New Orleans simply didn't rise to that standard.