Hi all! I wanted to ask a question to the policy people regarding backing up through drawn cards. In general, we try not to do this because drawing cards presents decision points, but there are cases in which we can do so easily and safely.

My question is: Assume the player is drawing from a randomized library, and we need to back up through their card draw. According to policy, we take a random card from their hand and put it on top of their deck, but don't shuffle. If the library was random before the draw, why do we not re-randomize it to minimize the information gained by the player by the rewind? This seems like something we should be doing; am I missing something?

Edited Lyle Waldman (June 28, 2017 11:13:24 PM)

If we put a card that was already in the hand back, we don't want it to be shuffled away.

Information gained can be disruptive.

But changing the actual state of the game is generally way more disruptive.

Imagine a Modern game between AP on Counters Company and NAP on Eldrazi Tron. AP is holding Vizier of Remedies and Devoted Druid, but NAP has a Walking Ballista with a few counters on it, so AP is waiting for a Stony Silence or a Collected Company or some other answer to come up so she can combo off and win. As AP is considering attacks, she notices that NAP paid insufficient mana for his Reality Smasher - he confused Power Plants for Towers, and cast it for 4. This is a Game Rules Violation, and we consider a rewind through a draw step.

If we take a random card from AP's hand and put it back on top, we'll probably end up at about the same place we should be. AP might make some different choices knowing what she drew, or she might end up with an instant where she didn't have one before and have some new options during NAP's end step, but it'll likely turn out okay.

If we take a random card from AP's hand and shuffle, there's a good chance that we take either the Druid or the Vizier. Now we've changed AP's entire game plan - she's been playing the last few turns with the assumption that she'll be able to combo off shortly, but now she's lost that option. NAP has been observing how she's been playing, and assumed that she had the combo, but wasn't sure, so NAP has made plays around then, and might notice the sudden change in AP's tactics.

The latter scenario is a lot more disruptive - by removing a random card and shuffling it away, you potentially change a player's gameplan. Early on, it can turn a hand you kept into a hand you'd mulligan, and in the late game, it can potentially remove a card you'd been holding for 4 turns, and create an incentive to hold onto a land or two just to maybe protect your card from rewinds.

Information is problematic, but not as problematic as the game state itself.

I've been ruminating on this lately, mostly considering the differences between GRV and HCE:

GRV rewind – random card put on top of library

HCE fix – opponent chooses card shuffled into library

Why are these different? Applying John's logic, HCE disrupts the game state by potentially shuffling an important card that player had since the beginning of the game—in fact, it seems more likely.

I think it must be due to the severity of the infraction and potential for cheating, but it still feels weird that two infractions that are sometimes nigh indistinguishable from each other should have such similar-yet-different fixes.

Looking forward to the thoughts on this!

Originally posted by David Poon:

I've been ruminating on this lately, mostly considering the differences between GRV and HCE:

GRV rewind – random card put on top of library

HCE fix – opponent chooses card shuffled into library

Why are these different? Applying John's logic, HCE disrupts the game state by potentially shuffling an important card that player had since the beginning of the game—in fact, it seems more likely.

I think it must be due to the severity of the infraction and potential for cheating, but it still feels weird that two infractions that are sometimes nigh indistinguishable from each other should have such similar-yet-different fixes.

Looking forward to the thoughts on this!

I think this is also where my source of confusion comes from as well. It seems like the philosophy guiding the HCE fix and the philosophy guiding this GRV backup philosophy are contradictory to me as well.

The main difference is that in HCE, the violation *always* has a card in an
unsearchable place, whereas the GRV only has an extra card incidentally as
a result of a card draw being backed up. Since the card draw isn't the
error, we don't penalize GRV the same way as HCE.


*David Poon*

Let's start with why we don't shuffle away random cards for GRV.
We don't do that because the error wasn't the drawn card. It was some earlier error that had a subsequent legal card draw (once both players accepted the error as correct). It is considered too dispruptive to take away a random card from their hand and shuffle it away, as the players entire game plan for a few turns could revolve around having that card. We *want* to be here over shuffling a random card away. That's too punitive,


Now HCE is deliberately more punitive. The error *is* the card going into the hand (or set). There is no opportunity for the opponent to catch the error before the data is lost. That's the key here. There is no way for the opponent to know something is wrong until the information is lost. It's often much harder to see an extra card drawn than a spell cast for the wrong mana or with illegal targets.
That additional angle of potential abuse is what leads to the harsher penalty.

There isn't an inconsistency here.
We are taking into account the opponents ability to monitor what is going on and correct things.

First, it's always good to refer back to “The Source” (I wonder if Toby's OK with that nickname? heh):
https://blogs.magicjudges.org/telliott/2016/04/04/revisiting-the-hidden-corners-of-hce/

You can find other blog posts from Toby & the Policy team, linked from that blog post; keep it handy, read it thoroughly, read it often.

Now, for an over-simplification of GRV vs. HCE:
If an error occurred before a card was put in hand, it's GRV; if the error is a card or cards being put in hand, it's HCE.

That extends beyond just drawing an extra card or three, to all the permutations of HCE - but the basic principle serves well as a guiding principle. If the opponent had a chance to notice and stop an error before a “non-public card” became part of the problem, then it's the GRV that we focus on. With HCE, the opponent has little or not chance to detect an error until the card becomes the problem.

d:^D

Oh, one more thing…
Penalties do take into account effectiveness in deterring behavior.
A cheater who wants to draw an extra card has to weigh several factors..
1) can he avoid detection
2) if detect can he lie his way out of a DQ?

If your hand is crap, it might be worth the risk to draw and extra card. Best case is you aren't caugh. Worst case is you are DQ'ed. Average case is you get HCE, lose a random card. If I'm inclined to cheat, those might seem good odds.

But if I change to losing my best card…suddenly the advantage goes way down. I have to rely completely on my opponent not noticing to gain an advantage …even if I avoid a DQ and just get an HCE I will most likely end up worse off.

Sorry Scott, but I don't believe anyone was asking for the defining differences between GRV and HCE—my apologies if I implied that—but rather the differences in carrying out the fixes.

Bryan, would it be appropriate to summarise the philosophy as 1) trying to preserve the game state for GRVs, and 2) making the HCE fix harsher than that for GRVs to curtail cheating in those scenarios?

I had been seeing the HCE fix as a “revision” of GRV where a shuffle is involved for previously unknown cards, but it looks like it would be better to view it as a deliberately more punitive fix to account for the ease of cheating with this infraction?

Actually, we've made the HCE fix *less* harsh than its predecessor, Drawing Extra Cards (which was Game Loss).

Sorry to have wasted your time supplying foundational philosophy, David. :/

d:^D