Here's the thing about using replay in sports - the results are only as good as the system you design for the reviews.
You have to give the NHL credit: They are trying to address a small but noticeable problem where 10-15 games a season were being decided on an iffy goaltender interference call or non-call (witness the Dallas Stars loss to the Detroit Red Wings just last season).
But beyond the inherent problems of reviewing a fundamentally subjective situation, the rule the league has in place about reviewing goals that involved potential goalie interference has two extremely evident flaws, one of which was on display in the Stars 4-2 win over the Florida Panthers.
Here is the review rule as it pertains to goalie interference:
(b) Scoring Plays Involving Potential "Interference on the Goalkeeper"
(i) A play that results in a "GOAL" call on the ice where the defending team asserts that the goal should have been disallowed due to "Interference on the Goalkeeper," as described in Rules 69.1, 69.3 and 69.4; or
(ii) A play that results in a "NO GOAL" call on the ice despite the puck having entered the net, where the on-ice Officials have determined that the attacking team was guilty of "Interference on the Goalkeeper" but where the attacking team asserts: (i) there was no actual contact of any kind initiated by an attacking Player with the goalkeeper; or (ii) the attacking Player was pushed, shoved, or fouled by a defending Player causing the attacking Player to come into contact with the goalkeeper; or (iii) the attacking Player's positioning within the goal crease did not impair the goalkeeper's ability to defend his goal and, in fact, had no discernible impact on the play.
Assuming this is the entirety of the rule, and there's no reason to believe otherwise, there are two big problems with how this is written that should have been immediately apparent to the NHL.
The first is that the goal and no goal overturn criteria operate on different standards. Something could be classified as goalie interference under rule 69.1, 69.3 or 69.4 and still meet the criteria needed to overturn a no-goal call.
The most obvious example here is crowding, as defined in in rule 69.3 and elaborated on in Table 16, Section 6 of the 2014-15 NHL rulebook (this season's version doesn't appear to be available online). In seven of the eight situations in that table (three of which result in a disallowed goal), it is not necessary that the offensive player be in the crease - just that the goalie is in it. But by the overturn of a no-goal criteria, an offensive player must have been in the crease or else they are cleared via scenario (iii).
Put another way, a player could be guilty of crowding sufficient to meet the goalie interference criteria but also meet the criteria to have the no-goal overturned because he does not initiate contact and he was never in the crease to begin with.
This dual standard is a huge problem with how this rule is written.
Here's the circular logic that could come into play eventually: A goal is allowed in the face of crowding, as described in Table 16, Scenario 6.D (which involves the goalie in his crease initiating contact with an attacker, who does not need to be in the crease). Under criteria (i), the review of an allowed goal, this should be changed to a no-goal as it violates rule 69.3. However, if the crowding player was outside the crease at the time, this could be re-challenged and changed back to a goal because it fits both (i) and (iii) of the no-goal challenge criteria.
The other problem is with how the rule is defined with regard to the length of the play being examined, which is what was illustrated in the Stars-Panthers game.
The Aleksander Barkov goal arguably met the standards of penalty goalie interference at 19:07 remaining in the first, when Jonathan Huberdeau backed into Kari Lehtonen with both players clearly in the goal crease (see the picture with this article). The contact knocked Lehtonen on his rear end. This is covered in rule 69.2, 69.3 and Table 16, Scenario 1.D (other than incidental contact in the goal crease), and is a minor penalty. It got missed. it happens.
Because he was knocked backwards, Lehtonen wasn't able to get back to his skates easily and indeed doesn't get even settled back on his knees until 19:03, right as the puck starts to get passed across the crease to the wide open Barkov. Lehtonen is still unbalanced, not set on his skates, and is unable to make a good move across to make his best effort on the save.
Here's the NHL's explanation for the review and overturn:
After reviewing all available replays and consulting with NHL Hockey Operations staff, the Referee determined that Florida’s Jaromir Jagr did not interfere with Dallas goaltender Kari Lehtonen before the puck crossed the goal line. The decision was made in accordance with Note 3 of Rule 78.7 (iii) which states, in part, "the attacking Player’s positioning within the crease did not impair the goalkeeper’s ability to defend his goal and, in fact, had no discernible impact on the play."
That misses the point so badly that it's almost comical. The attacker in question was not Jagr but instead Huberdeau, who did initiate contact with Lehtonen earlier in the sequence. The question under review should never have been about Jagr but instead about whether the Huberdeau contact and missed penalty was long enough before the eventual goal that it was irrelevant from an overturn point of view.
Given the five-second time frame between the missed penalty and the goal itself, you can make a decent argument that Lehtonen should have been able to meaningfully recover or that it's not enough time given the lateral nature of the pass leading to the goal. That's the gray area here. How Jagr got into the equation at all is extremely odd.
Given the word "immediately," overturning the no-goal call based on time passed is defensible (though by the opposite of the criteria laid out in the overturned goal section as opposed to the criteria laid out in the overturned no-goal section). It does bring into question, though, how long is too long, and why is this not clearly defined?
Rule 69.3 does not put a time limit on the contact being evaluated, simply stating: "If an attacking player initiates contact with a goalkeeper, incidental or otherwise, while the goalkeeper is in his goal crease, and a goal is scored, the goal will be disallowed." Table 16 is the place that uses the word "immediately" to qualify the necessary timing.
The NHL has made it clear they're using review to look for non-penalty goalie interference, not missed minors. There is no provision for a minor to be retroactively assigned on review. But as illustrated Saturday night, missed minors in the moments leading to a goal can arguably have the same impact on a goalie's ability to make a save as contact as the puck goes into the net.
Here's another example where an earlier missed penalty call could impact a scored goal: Say a forward knocks the stick out of the goalie's hand in other-than-incidental contact, and the stick gets pushed away. This should be a penalty, either slashing or goalie interference, but those get missed in crease battles. If the goalie cannot get his stick back before a goal is scored, even 10-15 seconds later, not having the stick could absolutely impede his ability to defend his goal. But at what point does the initial missed call "expire" and no longer affect the play in a case like that? How would an official be instructed to use video review in that case? It's very unclear.
Taken together, the criteria for video review of goalie interference as it is written right now are extremely problematic. They should be simplified to rely only on the existing goalie interference rule rather than trying to re-write the rule on the fly, and there should be a time frame added to give guidance for plays similar to the one in the Stars-Panthers game.
The NHL has put themselves in a bit of a silly position where they are allowing the overturn of 50:50 plays where there cannot be, by definition, objective evidence as justification. If that's the hill they're going to die on, they should at least make the ride down as smooth as possible.