Thursday, August 9, 2012

Where Did Avacyn Restored Go Wrong?

It's important to recognize when you succeed and determine what you did right so that you can do it again. Fortunately, that's usually pretty easy. People are happy to tell you the good things they've noticed and are attuned to the positive details because they remember thinking "wheee!" when they were experiencing them.

It is at least as important to recognize when you fail and determine what you did wrong, so that you can avoid the same mistakes next time. Unfortunately, that's rarely easy. While the internet is full of people happy to tell you what you did wrong, it's much harder to find useful critical feedback from that same crowd. Part of that is because some get a bigger rush from shouting profanities in a forum than from making coherent arguments, but the real difficulty is that it can often be very hard to pinpoint why precisely you don't like something.

As a designer, it's not acceptable to disregard your critics as specious trolls, or even to accept failure in a general sense. In order to grow—to make your next product better—you must understand what it was about your work that caused people (even if it's only a minority) to be unsatisfied. It doesn't even matter if they're arguments make sense. Something upset them, and you have to figure out what and why. I suspect this is a tough skill for any designer to learn, but I know for a fact that it's one of my big weaknesses.

While I now recognize that I didn't really enjoy AVR, it hadn't occurred to me until I noticed the clamor against it. The same thing happened back in Kamigawa with which I was happy enough at the time, but later realized I'd ignored serious problems in my blissful-ignorance mode. It's a nice mode for a lot of situations (TV, movies, work), but if I'm going to grow as a designer, I can't afford to miss learning opportunities like this.

So let's tackle the question head-on and see if we can't learn something useful.

Why is Avacyn Restored Limited unpopular?

I asked this question on Twitter and was pointed to an Avacyn Restored Limited Overview by Matthew Watkins, an excellent analysis of the Limited environment through (apparently pain-staking, manually-gathered) tournament data from Magic Online. Matthew ends with a positive conclusion: "I am really enjoying Avacyn Restored as a format." While he remains a fan of AVR to this day, the comments at the end of his article start showing some of the crowd's dislike. If you read Part 2, Matthew analyzes how the draft format evolved after its release and calls out specific cards and trends that cause the best color-pair (GU) to be so over-drafted that it actually puts up worse results than the previously-iffy RW deck.

I looked further and was pointed to Why Avacyn Restored Makes You Want To Punch Cute Animals by Jackie Lee which enumerates the set's flaws in an exceptionally coherent and well-reasoned manner. She points to increased variance overall, exacerbated by miracles and the larger quantity of variable-value cards; spells like Thatcher Revolt or Flowering Lumberknot that are nearly unplayable in the abstract but very strong in the right deck (in the ideal circumstances). While every set has some cards (milling being the proto-example) that are bad or useless until you meet some non-trivial synergy threshhold, it would appear that AVR has such a surplus that they get pushed into every maindeck.

Where most players will go for straightforward aggro or control strategies, focusing almost exclusively on the most efficient threats, the most versatile answers and the most profitable utility spells, some will pick up the late Burning Vengeances, Tome Scours and Gnaw to the Bones of the world. Having enough straightforward playable cards to keep the Spikes and Timmies happy, leaves the wacky cards to the Johnnies and everyone gets to build the kind of deck they enjoy.

Avacyn Restored's surplus of weak/weird/conditional cards at common and uncommon means that the strong/safe/obvious picks go fast and players are forced to consider cards outside their comfort zone. If you ignore Constructed formats beyond Block, it's easy to imagine that the overall power level of a set is irrelevant because it's all relative. In a land of Grey Ogres, Shock isn't limited, it's focused. Where Searchlight Geists roam, surely Guise of Fire is solid. While there's some truth to that, there are multiple important kinks in that plan:
• If you power down the commons and
uncommons, you've got to power down the rares and mythics as well. Otherwise, your less capable removal will be an insufficient response to your opponent's rare threats. But you can't nerf the rares because they sell the set.
• Standard and older formats can't be ignored and powering down your set would leave Constructed relatively unchanged, allowing it to stagnate 3 months earlier.
Even if the set would have been fun in a vacuum, players won't forget the power level they're used to and they will feel that set is weaker, reducing their enjoyment of it.

Archetype Diversity
The next biggest complaint I've seen about the set is that it feels like "a bad core set," meaning there aren't enough interesting things to do.

While it's unfair to compare any set to Innistrad, the pinnacle of all Magic set design, it's very fair to say Avacyn Restored offers significantly fewer viable deck-building options than most sets—including Magic 201x (which I love, for the record). The primary cause is the imbalance in archetypes. As Watkins' analysis shows, blue and green are dominant (because blue's bounce and green's soulbond tend to trump most other strategies) and black is The Worst Color (loner is trumped by blue's bounce, too slow against fast red-white decks and doesn't play well with souldbond). There are fewer legitimate strategies and so there's less to explore, causing players to tire of the format sooner.

Removal and Bombs

Olivia Voldaren and Drana, Kalastira Bloodchief were brutal to face in their time, but you had Tragic Slip and Flame Slash to deal with them. How do you deal with Sigarda, Host of Herons? Deadeye Navigator? There aren't really answers in the set. Not below rare. While it's not secret that I've been curious about making a set or block with weaker removal than we're used to, I also propose that the top-end of bombs have been too strong and would simultaneously need to be cut. The goal being that more interaction will happen during combat, where the game is most fun.

Avacyn Restored does feature weaker removal, but the bombs remain just as strong as ever. The problem is that once the bombs hit the table, the game tends to be more one-sided than normal because you have very few outs to undo your opponent's advantage. That makes the games more swingy, which is arguably good for new players but clearly bad for established players.

Before I move on, I'll point out that Ian may also have been referring to the games in which you get bowled over by the weight of the soulbond deck or the speed of the humans deck. While that's independent of the rares in the set, it's another real problem that's enabled by reduced removal.

Hidden Value
Zac Hill wrote Bringing Flashback Back in December to explain the trap that Odyssey fell into but they were careful to avoid in Innistrad. Specifically, that hidden value increases the gap between strong and still-learning players. The good players recognize the incremental value in cards like Morbid Hunger that look bad to new players, allowing them to get more of them and then defeat the new players in a way that isn't even readily apparent.

Abundant Growth, Alchemist's Apprentice, Borderland Ranger, Butcher Ghoul, Crippling Chill, Driver of the Dead, Fleeting Distraction, Geist Snatch, Joint Assault, Peel from Reality, Scroll of Avacyn and Scroll of Griselbrand are all spells that cantrip or net you an extra card or virtual card that a new player might undervalue. That's not counting the
obvious two-for-ones like Amass the Components, Grave Exchange, Gryff Vanguard, Mental Agony, Undead Executioner and Voice of the Provinces, and that's just covering common. Soulbond tends to lead toward greater-than-apparent results; The myriad blink effects that seem to do nothing actually multiply these incremental effects; We've got strong evidence that AVR is not an easily grokkable set and many players will lose a lot of games without ever figuring out why. A recipe for discontent.

The solution Zac offers and that we all know worked in Innistrad is to make cards generally do what they say they do and look as good as they are. Sometimes, that means simplifying a card (Naturalize over Natural End), sometimes avoiding incremental advantage (Avacyn's Pilgrim over Borderland Ranger) and sometimes making a great card obviously great (Travel Preparation over Trusted Forcemage). There are always exceptions (how long did it take to figure out how to use Gnaw to the Bone?), but you can't let the exception become the rule.

There's something visceral about how bad AVR is for draft. I don't blame anyone who hates it and can't explain why. I hated it and didn't even realize it. Given the opportunity to draft it online, I would opt for Innistrad or M12 instead. Yet I never said "man, I sure do hate AVR." That's a big part of why I did the research and wrote this article. I need to get better at realizing when something's wrong, rather than just automatically adapting.

The eloquent Mr. Becker was kind enough to spare some time to share his insights on AVR Draft with me and he had a lot of good points. He covered most every point I've mentioned so far, but had a few more nuggets. Not unlike Rise of the Eldrazi, a lot of the best cards in the set are quite expensive (Avacyn, Angel of Hope, Griselbrand, Craterhoof Behemoth). Entirely unlike Rise, there is only one card in the set to ramp your mana to cast these big spells, many fewer mana sinks to make use of all that mana once you've got it and—once again—not enough solid removal to answer the threats that do hit the board. We also discussed soulbond and miracles, and the result was enlightening…

With just 19 soulbond cards in the set (11 at common), there wasn't a lot of room for a range of quality. They had to push the mechanic so that it would be exciting and draw players to the set. Diregraf Escort is the only one of the bunch that you would actively avoid playing, and the inclusion of Flowering Lumberknot even validates that piece of cardboard.

It's very hard to make a card attractive that boosts two cards without doubling the result. Even the common Trusted Forcemage is a serious threat as an efficient Centaur Courser who also boosts your Wandering Wolf. With few good removal spells, you're depending heavily on blue's bounce package to counter soulbond, since the flicker effects in the set have a range of 0. And right there, you can see why blue and green are dominant.

The response to miracle is interesting. What's the difference between winning a game by drawing a Wrath of God and tapping four mana versus drawing Terminus and tapping one? Granted, that three-mana difference will often let you follow up the sweep with a better threat but the basic effect on the game is identical. So what is it that makes the fun of playing Wrath exceed the unfun of losing to it, where the fun of playing Terminus doesn't exceed the unfun of losing to that? The leading theory is one of luck vs skill, or at least the appearance of it.

When you Wrath my board—unless you windmill slammed it from the top of your deck—I don't know how long it's been in your hand; Maybe you've been slow-rolling me, Wrath in hand, forcing me to build up my team with the explicit intention of wrecking me. When you miracle Terminus out, we both know it only happened because you managed to mise it that very turn. Whether you've been playing toward that event or not, it was pure luck that it actually happened.

That's fascinating to me as a designer, because that draw isn't any more or less random than every other draw in the game. Magic is random draw after random draw
every turn of every game. The best players "make their own luck" by playing the odds, crafting game states that will end well for them if they draw one of several cards they know they've got. I think the fundamental problem is that adding cards to your hand and planning around them feels great for the winner and feels fine for the loser (it's the definition of "card game"), but dramatic top-decks feel terrible for the loser—miracle or not. "You top-decked the winning Lightning Helix? Really? Fuuuhhh."

A hard-cast Terminus isn't really any worse than a Wrath of God from your hand, but top-decking either at the critical moment is always going to leave a mark. What makes miracles frustrating for players is that the mechanic exacerbates the frustration of losing to a lucky top-deck by highlighting that moment. It's not just "phew, that was just the card I needed to pull that out. GG. Let's go to game three," it becomes "It's a miracle! Har har! Hey everyone, check out this amazing comeback!" Miracle puts the focus on the part of the card that is least fun to lose against, making that loss much more memorable, and leaving a bad taste in your mouth that gets associated with the mechanic and with the set.

Miracle was a mistake for the same reason you'd never print a card that says "If you win this game, win the match instead." Your opponent smashing you by top-decking Planar Cleansing on the last possible turn isn't fun. Why would you add, "dance in your chair and point out how hard you just stomped your opponent" to that? The fact that Bonfire of the Damned and Entreat the Angels are designed not just to be swingy but complete blowouts is a big poster for hating miracle. Hell, I remember feeling bad as the one playing the miracle, because even I felt like there was no skill on my part: "Hey, I opened this and managed not to put in my sideboard. How clever am I?"

It seems like there was a confluence of unfortunate choices and—I'm forced to surmise—insufficient development that reduced the set's overall fun and depth. Miracle exacerbates one of Magic's little flaws, soulbond is neat but too strong, the bombs outclass the removal, incremental advantage fights with greater variance, and the format is lopsided enough to cut off archetypes.

If I were going to try to improve this set, I would reconsider miracles entirely; either keep soulbond and the bombs but temper them with better removal, or keep removal weak but soften soulbond and the bombs; I would cut Mist Raven, make it uncommon or increase its cost; I would either ditch blink or swap it with a flicker than can target opposing creatures (as a trick against soulbond); I would save loner for a set without soulbond (or vice-versa) (or nix it permanently—it doesn't even combo with black exalted!); and I would replace some of the hidden value cards with straightforward threats and answers.

I'm really glad I looked into this question because I really hadn't put my finger on any of these problems in a concrete, actionable way. Thanks so much to everyone who shared their thoughts on Twitter, on the phone or via articles. I definitely learned a lot and hopefully didn't waste too much of your time telling you something you already knew.

If you have anything to add to the conversation, please do. What did I get wrong? What did I miss?


  1. Thank you so much for writing this! I've heard people bashing AVR for so long now and every time I ask them 'why is it a bad format?', being the troll/devil's advocate that I am, and nobody seems to give me a logical answer. It's always stuff like 'it's not fun' or 'it takes no skill'.

    It's really refreshing to see some actual analysis put into it!

    (sidenote: flickering your opponents creatures (at least with effects like ghostly flicker and cloudshift) doesn't mess up their soulbonds because the creatures just return and they can repair. you'd need to exile until end of turn or something. if that's what you meant. sorry!)

    1. I might have gotten my 'blink' / 'flicker' terminology mixed up. Yes, it would need to exile until EOT.

  2. Thank you, Jay! I struggled with much the same thing (not realizing I didn't enjoy the format until I heard it bashed) and you've done an excellent job of pinning down the reasons behind its lack of appeal.

    I also like most of your ideas for improving the set, but I think WotC got themselves stuck in a pretty tough spot with Soulbond as a mechanic and would need to go even farther.

    First off, the mechanic is a lot of fun to play with and fits the setting well, so you don't want to ditch it. But you also can't make the cards it appears on much weaker because the mechanic is stronger than it looks and would then be unappealing. I also explicitly wouldn't put more Doom Blades or ways to flicker opponents' creatures in because then the feel-bad moment of getting blown out when attacking with a bonded creature might outweigh the fun inherent in the mechanic.

    The only real solution I see is to add a lot of good sorcery speed removal so that a pair of creatures won't take over the game, but still will have an impact, and theoretically that would reduce the bomb problem as well.

    As for the marginal value, I think that's pretty much unavoidable if you're using flickering as a mechanic, which you want to because of Soulbond. At the moment I don't really see a solution beyond cutting Miracle, Flickering, and the Loner mechanic and rebuilding from just the Soulbond + Sorcery speed removal plan. If we're leaving the baddies with undying maybe they could have a bunch of sacrifice outlets so that you could unpair Soulbond creatures at will and the number of flash creatures could be cranked up to add some more bonding tricks.

    Anyhow, there're my two cents on the subject.

  3. Your comments on Miracle (which I think are spot-on) brought some questions to my mind:

    1. Would it help if the incremental advantages were moved to Miracle? Would that put enough of a spotlight on them to help new players understand what's going on?

    2. Would it help if Miracles had to be set up? For example, what if you needed a cleric in play to Entreat the Angels?

    3. In general, is the problem with Miracles that they involve too much luck, or is it that they draw attention to the variance that's healthy for the game at the macro level but sometimes frustrating in individual matches? If it's the former, then weaker Miracles might be okay; if the latter, it implies that this is a design avenue that we've learned should be closed off.

    4. The biggest question of all: how should incremental advantage be introduced to new players?

    1. It occurred to me that by posting this just as questions it might be seen as a kind of challenge, which was not my intent. I'll put forward my own thoughts, if only to fairly reveal my own ignorance. :-)

      1. I think this would be fine, but the set would need another pass on its theme. Instead of "Avacyn beats down," it would have to be "Avacyn helps the humans beat down."

      2. I like this as a way to capture the idea of faith rewarded: you the player make sacrifices in the belief that they will pay off in the long term, and then they do. I would love to see this backed up with "minor Miracles," to enable a thematic "going to church" deck.

      3. I haven't played enough of AVR to really know, but I think it's the former--a problem of implementation rather than inherent design. We all know and accept that sometimes our opponenets will topdeck the $16,000 Lightning Helix. The problem, I suspect, is that the opponent has to set up the $16,000 Lightning Helix, whereas the Miracles don't seem to have a strong connection to good play.

      4. This one stumps me. I don't see a good answer beyond having the "aha" moment, either through experience or explanation.

    2. The idea of Miracles requiring the Worship condition (you control at least one creature) seems flavorful and an interesting avenue to pursue.

      #4 is definitely a question to keep in mind for the future.

    3. Thomas, these are all excellent questions. Thought-provoking and well said.

      1. As I writing this, I considered offering the solution of implementing Miracles more like the first time we saw them: the horribly named Approaching mechanic from Muraganda. It's basically exactly what you suggest in putting the incremental effects on the draw trigger. The problem with that is that Avacyn Restored didn't need a draw trigger, it needed "miracles." And miracles need to be big. "Gain 4 life" "Draw a card" "Deal 1 damage" just don't feel like miracles.

      2. That's a very interesting idea. I agree with Pasteur that it's flavorful. My concern is that it's when you have no creatures that you're most in need of a miracle to get you back in the game. Maybe they could have tied it into Fateful Hour? "When you draw ~, if you have 5- life, you may cast it w/o paying its mana cost." Except we're just tweaking the play and the flavor, not addressing the primary issue with the mechanic.

      3. While I don't disagree that variance is healthy for the game, I'm thinking that reducing the power of Miracle (with the exception of the horribly busted Entreat and Bonfire) isn't a satisfactory solution since that would make the mechanic less epic and less worthy of the name "miracle."

      I could be totally wrong, of course. Maybe minor miracles (with a few big ones thrown in there) would be awesome. I'd certainly test it.

      4. This is a hard question, and an important one to think about. Marginal value is a major attraction point for Spike and removing it from the game would reduce its depth, turning off some of the most invested players, hurting tournaments, and ultimately hurting the game. That said, part of New World Order is hemming in the prevalence of invisible advantages to help get new players comfortable with the game. Mainly that means keeping the hard-to-evaluate instances to expert sets and out of core sets.

      Note that obviously good two-for-ones don't push the skill gap and are actually a pretty good introduction to the concept and benefits of card advantage. I guess my answer to this question would be, make fewer but more obviously good card advantage effects in the core set and fall sets (where new players are most likely to try the game), helping to train them to look for those kinds of advantages in the winter and spring sets.

      Thanks for the challenging questions. I've got at least two more perspectives on the subject now. Designer gold.

  4. Great article, Jay. Like you, I am pretty adaptive and didn't see the problems with AVR, even after many people attempted to point them out.

    My main defences of AVR either sounded like

    (a) "it's removal light because of soulbond and soulbond is awesome design space and the perfect setting to try some variety in amount of available removal (on the low side), and I'm all for that", or

    (b) "sure miracles are high variance but variance is a huge part of the game and people who want to pretend that it isn't should be playing chess instead - I think it's healthy to highlight the variance"

    Despite these two responses, and despite enjoying AVR a lot, I have to admit that it's probably comparable to drafting a core set in fun level (to be fair, I like drafting core sets but that is at least one notch below where an expansion should be).

    What your article made me realize is that while I still stand behind both of my defences above, these two things should NOT have been in the same set. Tinkering with removal on the low end OR upping variance would have been fine, but both at once was way too much for people who have come to take certain things for granted from Magic sets.

    Perhaps what this highlights is not so much a failure to develop the set enough but a bias inside Wizards and the FFL - maybe there is a preponderance of people like ourselves who are easily adaptable to such variation, because after all these are just tweaking sliders on things that are moved around in every set (amount of removal / amount of variance).

    Perhaps the lesson learned is that designers should be aware of what players take for granted and be careful not to twist too many of these knobs too far at once.

    If I'm right, then I think one solution (other than dropping miracle entirely) would have actually been to actually connect it to soulbond... miracles cost (1) less for each pair of souldbonded creatures you control. Soulbond would need to be evenly distributed among the colours, including black. Just turf the loner mechanics and also turf black being the "evil" colour - it made sense in Innistrad, but not in Avacyn RESTORED to harmony. This would result in a challenge to give colours an identity but I think that could still be done with careful use of additional soulbond creatures and associated miracles...

    1. Well said, Trevor.

      We know for a fact that R&D can fall into the trap of catering to the most established/advanced players because of Time Spiral, though they've been much much better at avoiding that since.

      The trouble with making Miracle dependant upon Soulbond, apart from the general fiddlyness, is that it turns a modular mechanic that could be used in any deck into a parasitic one that only works in soulbond decks.

      I do think that pushing the variance up OR pushing invisible advantage might have been fine deviations on their own, but mixing them just caused a mess.

    2. New World Order has helped avoid R&D catering too much to advanced players because they put conscious attention to it.

      What we might have identified here is a different bias in R&D (being more comfortable with variation in core game elements than most players) and in this case, being unaware of it actually causes more problems for the advanced players than new/casual players.

      Being parasitic in this case might have been warranted. That's another knob that there are good reasons to keep down in general but occasionally crank up. Maybe they could have been linked a little less explicitly (like how infect and proliferate work very well together but aren't explicitly connected)? I'm curious what the other mechanic was they tested and dropped before miracle...

  5. Good Read Jay, thanks for the thoughts

    - Becker