Etali is like a box of chocolates, and other bugs [Investigating]

Dodecapod

Etali, Primal Storm

1.  After a handful of games with Etali, I've witnessed at least 4 distinct classes of behavior, which vary based on conditions I haven't isolated yet:

(A)  In some cases, generally with 5 or fewer cards in hand prior to attacking, Etali works as stated; here there are no intuitive or obvious bugs to report.
(B)  In several instances, Etali's ability has resulted in 7 cards in hand when I attacked with a full hand; intuitively, its ability should either fail in this case, or it should destroy/exile the top card of the opponent's library (analogous to attempting to steal a creature using Lay Claim with 3 other creatures on the board already).
(C)  When the AI Laid Claim to my Etali, it seemed to steal the last card in my hand instead of the first card in my library; I couldn't determine whether the AI obtained the stolen card in its hand or its library.
(D)  In one game, when I attacked with a full hand, Etali placed the stolen card on top of my library; in that instance, the hand limit was respected, but intuitively the ability should behave as described in (B), by either failing outright or destroying/exiling the stolen card, not by placing it on top of Etali's controller's library.

In case it's relevant, all of these classes of interactions occurred multiple times, although not all of them occurred in multiple games (e.g. whatever the causes were, behavior within a single game seemed to be generally consistent except for (C), when Etali behaved as in (A) or (B) prior to the AI casting Lay Claim, then behaved differently under the AI's control).

2.  (Edit:  After further testing, this section may be inaccurate, but I'm leaving it visible in case it helps somehow with ruling out possible causes and verifying the underlying behavior)  Additionally, Etali appears to add the stolen cards to the controlling player's permanent deck even beyond (A)-(D); in multiple instances, even when there was no reason to believe a stolen card had erroneously been placed on top of my library (e.g. when (B) occurred on a given turn and presumably excluded the possibility of (D) on the same turn), I later drew a card that was not part of the deck I began the game with, and either could have been or definitively was contained in the AI's deck based on the game state up to that point and/or the enemy PW's colors.  I don't know whether this occurred in all instances (i.e. in games where I didn't draw any cards from the opposing deck outside of (D), I don't know whether that was because those cards were not added to my permanent deck or because the game ended before I would have drawn the added cards) and/or what the causes might have been if the behavior was inconsistent, but in the cases where (presumably opposing) cards were added to my deck (excluding instances of (D)), this precedent seems inconsistent with recent intended interactions involving theft abilities (which now generally seem to steal only a single copy of a card) over the last several months, although it would be consistent with how theft abilities have interacted at some points in this game's history (e.g. during at least parts of the Zendikar and Innistrad blocks to the best of my memory).  Therefore, to speculate a bit, it may involve a broader templating issue with how these abilities are implemented within new sets based on obsolete functionality.

Vraska, Relic Seeker
At level 60, the description of Vraska's ultimate has correctly been changed in accordance with the v2.5 release notes to state "At the start of every turn, ... your opponent loses 5 life", but the effect erroneously deals only 3 damage per turn to the opponent, as it did prior to v2.5.  There may be a similar issue with Vraska at lower levels as well, but I'm unable to test that.

Savage Stomp
The +1/+1 bonus is applied after the creatures "fight", whereas the wording implies it should be applied before "fighting" instead (which is what occurs with the paper MtG counterpart); either the card text should be changed to clarify the functionality, or the functionality should be changed to reflect the current wording.

Eternal Resilient Khenra
When Eternal Resilient Khenra enters the battlefield, it grants +X/+X, where X is its current power, to the first other creature its controller controls, not +4/+4 as the text states; either the text should be changed to reflect the functionality or vice versa.

Walk the Plank
Walk the Plank's text doesn't specify it must target an opposing creature, but upon casting, creatures the caster controls are grayed out as illegal targets and can't be selected.  Either the text should be changed to clarify the functionality, or the functionality should be changed to allow for targeting creatures the caster controls.

//Edited Title -Brigby

Indymon

I’m going to echo the original poster’s experiences. In particular:

ive gotten to a hand of eight cards with this ability. Though it doesn’t break my heart and I guess it could be intended. 

In certain games Etali doesn’t steal a card when it attacks but I continuously end up drawing from my opponent’s deck for the rest of the game. I’m basically blocked from being able to access my own deck. 

DBJones

So, leaving a spot in your hand open should avoid these glitches (except the Last Claim one) until they get fixed right? Annoying, but at least for me I usually do that anyway to avoid skipping my draw (and I'll leave two slots open for that purpose when I'm using Etali).

Btw, has anyone seen temporary steal effects trigger the same bug as Lay Claim? I tend to see those more often.

UweTellkampf

Not really, @DBJones I‘ve often encountered a behavior in which Etali won’t put the stolen card in player’s hand even if there was a spot free. Instead, it will put it on top of your library and as such it won’t have the mana gain which is stated in the card’s description.

Dodecapod

Just as an update on Etali, I might have a working hypothesis for what's causing the inconsistent behavior, and it's possible I misinterpreted what was happening with point 2 in my original post, so that section may reflect a PEBCAK error instead of a genuine bug.

It appears as though exiling a stolen card from hand may act as an effective toggle switch between stealing cards into Etali's controller's hand and stealing them to the top of the library instead.  For example (ignoring any other aspects of the board state and turn actions):

Etali's controller (EC) turn 1:  Etali attacks and steals a card to EC's hand
EC turn 2:  EC exiles the stolen card, then Etali attacks; because the first card was exiled, flipping the toggle, the 2nd stolen card will be placed on top of the library (without receiving any mana) instead of into EC's hand.
EC turn 3:  EC casts the 2nd stolen card, then Etali attacks; because no further stolen cards have been exiled since EC's last turn, the 3rd stolen card will again be placed on top of the library without any mana, as on the previous turn.
EC turn 4:  EC exiles the 3rd stolen card, then Etali attacks; because another stolen card has now been exiled since EC's last turn, the toggle flips for the 2nd time and the 4th stolen card will be placed in EC's hand with full mana.

This idea still requires further testing and verification to be certain of what's occurring (my sample size is still fairly small if there are any corner cases or if I misinterpreted what I saw, and I haven't tested detailed variations such as holding 2 stolen cards in hand and exiling one or both on the same turn), but on the surface it could potentially account for all of the behavior aside from possibly case (C) in the original post, and at least the basic test case seems to be reproducible so far based on a few more games since yesterday.

Krishna

shhhhh theres nothing wrong with Resilient Khenra it works just fine *walks away, whistling

Dodecapod

Just to update once again, I'm giving up on trying to understand Etali; now I can't seem to reproduce the behavior I thought I observed yesterday, so it's back to the drawing board in terms of trying to discern an identifiable pattern.

Best of luck to anyone who ends up playing with or against this card; it's a lot of fun to use, but having counterintuitive, inconsistent behavior is also confusing sometimes when trying to plan out upcoming board states and turn sequences.

DBJones

I figured out one of the screwy things going on with Etali. It actually makes your opponent draw a card, then steals it. If your opponent's hand is full, it steals from their deck to yours. If it just did that second part, then drew and filled the card, it should fix several of the issues, including the >6 card hands (the card would then be unfilled on top of your library, but if they changed the wording to "fetch your opponent's next card to your library, then draw and fill it with mana" or similar that would be how it should work).

Was your hand full when your opponent stole from it directly? That could explain it a bit.

Dodecapod

DBJones said:

I figured out one of the screwy things going on with Etali. It actually makes your opponent draw a card, then steals it. If your opponent's hand is full, it steals from their deck to yours. If it just did that second part, then drew and filled the card, it should fix several of the issues, including the >6 card hands (the card would then be unfilled on top of your library, but if they changed the wording to "fetch your opponent's next card to your library, then draw and fill it with mana" or similar that would be how it should work).

Was your hand full when your opponent stole from it directly? That could explain it a bit.

Unfortunately, I don't remember; if it happens again, I'll pay closer attention and see if the results confirm that behavior.  At the least, the explanation certainly seems plausible, and it'll be useful to have some idea of what to look for in upcoming games.

UweTellkampf

I also have no idea really when and why Etali behaves as it should, and when and why it behaves randomly. Even bugged, this is still a strong card. Every second turn or so, it will gift you (maybe) a mana-filled card, but then some of those, the ones that are put on the top of the library, they are really like a box of chocolates, I end up discarding many of them because they dont really fit in my deck. So, it is a big difference, quality wise, in getting the mana and not getting it, obviously.   

@Brigby - is this bug being looked into?

Firinmahlazer

Edit* Pretty sure @DBJones is correct. Watch the opponent hand. You'll see a card flash in quickly before going to yours. Good eye dude.

DBJones

Thanks, I actually noticed it first when my opponent pulled out an Etali in Training Grounds, since your hand is brighter it was a bit easier to see.

ILikePancakes

Had 4 cards in my hand after casting Etali, it attacked, then no card stolen. Upon my next turn, I drew the stolen card from my library (of course empty of mana). Second turn, same thing. Got killed, was able to cast a new one. Worked properly that turn, then the ai obliterated me. I hate training grounds.

Enygma6

The Etali bug looks like the same root cause as the Hostage Taker one to me.  

The card (Etali) or creature (HT) tried to enter your opponent's hand, then is move to yours.  Where with Etali it is filled with mana.  
In the case of Etali, if the opponent hand is full, the stolen card is not changed from Library to Hand before being switched to you.  So since it does not transfer into Hand, it does not gain the mana.  You just draw it next turn.  Really annoying when you end up playing half of someone else's deck without being able to draw any of your own.  
For Hostage Taker, if the opponent hand is full and the creature is forcibly bounced, it is flagged to be destroyed.  But before destruction happens it is transferred to your hand.  When you cast it, the "destroy this card" action is triggered, thus it dies when cast.  
  
In both cases, it never looks at the size of your hand, just appending the extra card to what you already have.  Which can result in hands of 7 or 8 cards.  

AmoXJS

It is really sad...the bug(s) really make this card, which would be otherwise toptier, unusable

Szamsziel

Unusable??? Even bugged it is a staple in all my red decks. If you are worried about not being able to draw your cards - there are simple workarounds. Example: put discarding creature before etali - naheb, galeon, dreamstealer. Use discard abilities (angrath). Use archive to draw more.

Dodecapod

After having had more time to experiment with Etali, the explanations @DBJones and @Enygma6 posted seem to coincide with my experience too; apparently the sequence works like this when Etali attacks, as far as I can tell:

1.  The opponent draws a card (which fails if the opponent has 6 or more cards in hand)
2.  The drawn card is filled with mana (which fails if step 1 didn't occur)
3.  The card is transferred from the opponent's hand to Etali's controller's hand (if step 1 didn't occur, the top card of the opponent's library is transferred without mana to the top of Etali's controller's library instead).

Discussing this on Discord and Slack, I've seen one or two people mention corner cases that didn't seem to conform to this behavior, but haven't been able to replicate any deviation from those 3 steps personally; if the interpretation proves to be correct (at least in the vast majority of realistic scenarios), then Etali's ability has the interesting side effect of depending only on the opponent's hand size in order to work, and would allow Etali's controller to have an arbitrarily large hand in principle.

By taking advantage of that quirk, I was able to set up a story mode game where I eventually acquired 261 cards in hand before arbitrarily deciding to quit (in principle, it seems likely that accumulating at least 95 more cards would have been possible, assuming the app didn't crash before that); I have several more screenshots and a couple of brief videos if it would help for investigating further (and would also be able to reproduce a similar game state at will in order to record additional information), but this was the deck, essential game plan, and a few of the highlights illustrating the results, in case it helps with identifying the current and intended functionality:

Nicol Bolas



Deck strategy

1.  Dowsing Dagger, Turn to Frog, and Imprisoned in the Moon are used to fill the AI's board with 3 creatures that can't deal damage (Tightening Coils reduces the Frog token's power to 0) and lock the AI's hand.

2.  The combination of Dreamstealer in the first slot and Etali in the second slot then ensures Etali's ability will always add a card to the hand instead of the deck because Dreamstealer ensures the AI always has 5 or fewer cards in hand when Etali attacks

3.  The Plant token ensures only Dreamstealer deals damage, so ostensibly it should be possible to steal a number of cards roughly equal to the AI's life total minus 1 (pending any other damage sources and life gain that affect the AI).

4.  I used this approach against Brood Monitor (BFZ Encounters Chapter 3, level 8), which lacks any way to disrupt the lock or remove creatures as far as I saw, but there are likely other opponents which would be susceptible to the same technique as well.

5.  The 4 supports other than Dowsing Dagger are arbitrary (Prism Array is even counterproductive for this purpose), and are simply included in order to increase the deck size to 10 cards; I chose options that I thought would help establish the combo more quickly, but they have no material relevance to exploring Etali's behavior, and could effectively be blanks instead.

Sample results

DBJones

Wow, that's a crazy number of cards in hand! Nicely done.

Thésée

@Dodecapod you are my hero of the day :-). Thanks for sharing this 

Tilwin90

So we can agree that Etali's text might as well write:
"Whenever Etali attacks, if your opponent has five or less cards in hand, put the top card of your opponent's library into your hand and it gains full mana. Otherwise, put that card on top of your library instead."

Tilwin90

Oh god... so it behaves differently when the player or the opponent controls it?
Brilliant coding skills, I'll say