Hello everyone!

AP controls a Twilight Prophet, but he hasn't the City's Blessing.
During his upkeep, he reveals the first card of his library (a Swamp) and adds it to his hand. Then he draws a card.
During his first main phase, both players realize that AP resolved a “ghost” trigger and they call for a judge.

What's the infraction here and how you'd fix it?
Thanks in advance!

I think the illegal play here was drawing the second card from top, since AP revealing cards from his hand (he was drawing the swamp for his turn) isn't really an infraction.

I would rule this as an HCE, with a warning for AP. The fix is NAP choosing one card from APs hand and shuffling it back in the deck, unless AP knew it would be there (by an earlier scry, for example), in which case, the card returns to the top of the deck.
Notice AP can't choose from either the swamp nor cards he revealed prior that game.

One could rule differently based on HCE's philosophy:

Be careful not to apply this infraction in situations where a publicly-correctable error subsequently leads to an uncorrectable situation such as a Brainstorm cast using green mana. In these situations, the infraction is based on that root cause.

But I don't see any error on revealing the swamp, so there's no publicly-correctable error to be based on, there.

TL,DR: I believe this should be GRV/FtMGS with a backup to the beginning of the upkeep (attempting to put the trigger on the stack).

I agree with parts of what Yurick has explained above. You can, if you choose, reveal every card you draw to your opponent, provided you are drawing the card legally. The action was performed on a card that AP would have drawn legally otherwise. The heart of the matter is, what is the root cause of what happened, and how do we fix.

While the root cause is a GRV (AP did a thing they were not supposed to do, and NAP allowed them to), the result of the error ultimately was that an additional unknown card was drawn. Unless NAP (I think this might have been a typo in Yurick's response as it indicated AP) knew the second card of their library as well, this could be handled as HCE, since an unknown card moved from one hidden zone to another, and the error is not fixable with publicly known information.

Fortunately, in this scenario, the card revealed was a land; if it had been a card with a casting cost, HCE would not be enough to fix the game state, since it only resolves the additionally drawn card - but we technically aren't supposed to take the game state into consideration when evaluating a fix for a penalty (though this is done in cases where backups are considered, which this would be). As such, I don't think we should be handling this case differently simply because the other part of the trigger did not have an effect on the board state - the fact that it could have had an effect on the board state makes me believe we should be backing this up for consistency's sake.

I would rule this GRV/FtMGS with a backup to putting the trigger on the stack. Per the rules for GRVs and backing up, this does not fit into the partial backups prescribed by the former, so to back up, we would select a random card from AP's hand (with the Swamp set aside) to put back on top of the player's library, and then place the Swamp on top of that, then have the players proceed from where the trigger would have been (it wouldn't be put on the stack if AP doesn't have Ascend).

I believe this solution seems the most realistic. Both players are responsible for a legal game state, and NAP failed to maintain the game state allowing AP to resolve a trigger. Giving them a chance to see the content of APs hand to ‘fix’ the situation feels like it favors NAP.

I would have an internal debate over the HCE infraction here, especially because of this line "A player commits an error in the game that cannot be corrected by only publicly available information and does so without his or her opponent’s permission." (Emphasis Mine).

One could relatively argue that both players had a chance to notice the error and did not. So the opponent gave his permission to the (ghost) trigger of the Prophet.

So I agree with Bernie's infraction and Proposed Fix.

The description of the scenario is not very detailed on the actual communication between the players. If I understand it correctly, all actions that were performed by AP were observed by NAP and confirmed by him. In that case, I cannot treat it as HCE. It would be GRV with a possible backup to the point of error (which was putting the trigger on the stack). In the backup, I would put the Swamp aside, return a random card on top and then place the Swamp on top of it.

In case I learn that AP rushed through the actions without giving NAP reasonable time to correct the error, I would treat it as HCE, where the Swamp was the correct card to be drawn (and was just revealed by AP) and the other card was an extra draw. The remedy would be to put the Swamp aside, reveal the rest of the hand to NAP and let them choose one card to be shuffled into the random portion of the library.

“Information about cards previously known by the opponent, such as cards previously revealed while on the top of the library or by a previous look at the hand, may be taken into account while determining the set of cards to which the remedy applies.”

I agree with Milan, here.

If AP reveals the Swamp and the players both acknowledge that NAP loses zero life, then they're allowing a GRV. Jacopo says that “both players realize”, during Main Phase, and that suggests they both agreed on the reveal & draw. Otherwise, as Milan says, AP might have hurried into an HCE.

d:^D

If we consider this a GRV, how are we going to perform the backup?
Are we going to shuffle away the land that was revealed (and letting AP draw the second card he saw) or AP will draw that land for that turn (leaving the second card on the top of his library)?

Originally posted by Jacopo Strati:

If we consider this a GRV, how are we going to…

Originally posted by Milan Majerčík:

In the backup, I would put the Swamp aside, return a random card on top and then place the Swamp on top of it.

I also agree with Milan's proposed application of policy:

Originally posted by MTR 1.4:

To perform a backup, … cards being returned to the library as part of the backup should not be shuffled at that stage if their identity was known to only one player.

d:^D

Thanks to everyone for the answers! :D

Originally posted by Jacopo Strati:

If we consider this a GRV, how are we going to perform the backup?

Originally posted by Scott Marshall:

I also agree with Milan's proposed application of policy:

Scott, you've explained how we perform a backup, but can you clarify why we should back up here?

Originally posted by IPG 1.4:

A good backup will result in a situation where the gained information makes no difference and the line of play remains the same (excepting the error, which has been fixed). This means limiting backups to situations with minimal decision trees.

My read on the situation is that putting a card from hand second from the top–a card that might have been in his hand already, and now won't be drawn till next turn–has a lot of potential to completely disrupt decision trees in the game. Assuming that both players acknowledged the trigger, my instinct would be to leave the game in its current state.

Originally posted by Eli Meyer:

My read on the situation is that putting a card from hand second from the top–a card that might have been in his hand already, and now won't be drawn till next turn–has a lot of potential to completely disrupt decision trees in the game.

You don't think leaving the player with an extra card in hand is worse?