MTG Salvation

Of note: Dark Depths/Thespian's Stage isn't a 2-card combo, it's a 1-card combo. The combo is Knight of the Reliquary.

Bonus points if you donate both Platinum Angel and Immortal Coil to your opponent and take out their graveyard.

EDIT: It's not Modern legal, but for people interested in giant mandatory loops, someone has actually managed to construct a universal turing machine out of magic cards (basically, a computer that, using interactions between creature tokens and the stack, can compute basically any necessary problem).
Version 5, which requires some players to always say "yes" to may triggers, can be found here: http://www.toothycat.net/~hologram/Turing/About.html
An explanation of version 6, which removes the may triggers and makes the machine impossible to disrupt.

Because this turing machine built from magic cards can be used to calculate anything, it is actually possible to dedicate the machine (essentially, a giant infinite loop) to an unanswerable problem (say, whether all even numbers can be expressed as the sum of two prime numbers). That would actually make a game state where it's IMPOSSIBLE to tell whether the loop is infinite, and therefore whether the game is a draw.
Unfortunately, the machine needs 5 players, so we can't do this in a tournament. However, I believe all the cards necessary for the machine are legacy-legal, so if they ever make a Legacy-Emperor Grand Prix, we would hypothetically be able to make the head judge hate us forever.

The thing is, while it can get it out fast, it's not trying to. Something like turn 1 thoughtseize, turn 2 thopter, turn 3 transmute muddle for sword, turn four play sword and leave 2 mana up, is quite easy to do, and gives you the ability to take their answer and protect the combo when you play the second part. That's also the reason decks can't recover: Because you generally haven't spent your turns durdling with cantrips and trying to find the combo, you can interact a lot more, which makes it harder to power through thopter/sword. Additionally, since as soon as the combo is out you're under a huge amount of pressure, it's basically impossible to play around countermagic and removal once its out. It's a bit like how you technically have to cast a spell for Pact of Negation once they tap out for Hive Mind, so you're not ALWAYS immediately dead, but realistically you're not winning those games. The other thing is that, because thopter/sword makes the aggro matchup so good, you don't need to play nearly as much spot removal, instead allowing for more countermagic, card draw, and discard, which is exactly the sort of cards other combo-control decks might want to run. Basically, I misspoke: the combo alone isn't impossble to recover from. It's that, because it's not a combo you have to spend a ton of resources finding, you also have to recover from the combo while facing disruption from multiple angles.
And the testing for sword and seething song was this week. Everything else was before the pro tour.

No problem, always glad to make that a bit clearer.
First, we figure out a list to work with. These are usually quite rough. We do these with anything that could possibly be good with the card, we're pretty inclusive. Then, we run each one in 5 matches against a random sample of decks. If a list wins 2 or more, we make revisions and do 3 matches against every deck in our gauntlet. If at any point the deck clearly looks unplayable, we stop testing. If it looks like modifications are necessary, we change the deck and do it again. The most iterations we ever did were 3 runs through the gauntlet for Zoo with Nacatl, as we tried to find the right ones. We don't update our gauntlet to try and simulate what the meta would look like, we just assume that the clunkiness of our lists will offset the lack of hate. This obviously wouldn't work as well with dedicated combo cards like Dread Return, and would tend to overvalue their viability, but since we've never found a combo card to be too dominant, that hasn't been an issue yet. This method is fairly predictive for versatile cards, less so for all-in combo cards. Dread Return is a perfect example of a place where this method would fail. However, for other cards, cards which don't have perfectly obvious checks, it's fairly predictive.

As for the distinction I draw, I believe it's a matter of how resilient it is, and how fast it is. Take Scapeshift, for instance. Scapeshift actually demands quite a few slots in your deck, since you have to get to 7 lands, and play Valakuts, and play additional mountains. While that cost isn't huge, it is present. Thopter/Sword, on the other hand, require far fewer cards, and what cards they do want are things like Muddle the Mixture or Tezzeret, Agent of Bolas, which are already fairly good on their own and perfectly playable without the combo. The other thing is that the thopter/sword combo is fast. We were able to consistently get it out on turn 4 with countermagic protection, and from there, it was essentially impossible for any opposing deck to recover. Scapeshift needs 3 ramp spells and the Scapeshift itself to win on turn 4. It's not a matter of "it's a win condition that's fast, resilient, and reliable", it's a matter of "it's a win condition that's TOO fast, TOO resilient, and TOO reliable".

Quote from Kirblar

Quote from LegitKarona

I don't quite know how to say this...
There isn't enough graveyard hate. Sure, there's graveyard hate, but if you're relying on interacting with thopter/sword via the graveyard you're going to lose. Decks that play thopter/sword aren't planning on getting it out as soon as possible, they're planning on playing the control game for a bit and then getting it out when there's a window. Graveyard hate doesn't do a whole lot to that strategy, and unless you have Rest in Peace, your dream scenario is to remove the Sword before they can bring it back, in which case they still have the Foundry and can just draw into another one. You're literally just trading 1-for-1. And, as we found out time and time again in testing, trading 1-for-1 to disrupt the combo in a thopter control deck is not a winning strategy. In fact, we couldn't find a winning strategy. Aggro decks fold to it, midrange decks fold to it, combo decks can't play through it (once you get the combo, you can just play draw-go, counter the key spells, and spend all excess mana on thopters), and control would rather just play it. The only deck we found that could break through and win against it was infect, because when you think of a healthy modern format, you think of infect.

I'm sure there are other decks that could beat a thopter/sword control deck. The metagame would adapt, it always does. But unless we missed a HUGE piece of tech, Sword of the Meek needs to stay on the banned list.

I just started ignoring people who thought it was safe. Made my life easier.

NO. PEOPLE ARE WRONG ON THE INTERNET.

EDIT: kidding aside, that's probably a good idea. I do like to point out the theory for why a card should stay banned, rather than just "I played games with it, you didn't." At least it's more productive.

I don't quite know how to say this...
There isn't enough graveyard hate. Sure, there's graveyard hate, but if you're relying on interacting with thopter/sword via the graveyard you're going to lose. Decks that play thopter/sword aren't planning on getting it out as soon as possible, they're planning on playing the control game for a bit and then getting it out when there's a window. Graveyard hate doesn't do a whole lot to that strategy, and unless you have Rest in Peace, your dream scenario is to remove the Sword before they can bring it back, in which case they still have the Foundry and can just draw into another one. You're literally just trading 1-for-1. And, as we found out time and time again in testing, trading 1-for-1 to disrupt the combo in a thopter control deck is not a winning strategy. In fact, we couldn't find a winning strategy. Aggro decks fold to it, midrange decks fold to it, combo decks can't play through it (once you get the combo, you can just play draw-go, counter the key spells, and spend all excess mana on thopters), and control would rather just play it. The only deck we found that could break through and win against it was infect, because when you think of a healthy modern format, you think of infect.

I'm sure there are other decks that could beat a thopter/sword control deck. The metagame would adapt, it always does. But unless we missed a HUGE piece of tech, Sword of the Meek needs to stay on the banned list.

Dredge decks were the first thing we tried. With no way to sacrifice creatures from the graveyard (Dread Return is banned), Bridge from Below lists were horrible. We then tried a couple of dredgevine variants, using first Fatestitcher and Gravecrawlers to bring back Vengevines, then with Skaab Ruinators. Both were bad, with the Fatestitcher version being horrible and the Skaab Ruinator version being playable, but still bad. We tried a Reanimator-style deck that milled Unburial Rites and used Life from the Loam to get the lands to cast it, but that was bad. The last thing we tried was a dredge-based Griselbrand Reanimator Shell, which was actually decent but ultimately not as good as the current version.

@NessOnett: That's true. Maybe I'm underselling GGT. But I will stand by the prediction that unbanning it would have a far smaller impact on the format than the unbanning of Bitterblossom, and that it is the easiest card to unban on the entire list.

While I would disagree with your ordering (GGT is by far the best card to unban, since it's actually just completely unplayable), that's roughly correct. That said, I like having at least a little bit of concrete information to use when talking about unbans, and it gives everyone a good starting point. You're never worse off for knowing with a bit more certainty what's going to happen if X is done.

Also, a couple of people have been arguing for a Sword unban, so I decided to see if there was anything to it.

Some of you may remember my team did some (sadly, limited) testing of possible unbans, testing which managed to predict with some accuracy what would happen given their January unbans (Bitterblossom would be too weak to perform in modern, and Nacatl would be strong enough, but still a safe unban). That testing, which we've done a bit more of, says the following about possible unbans:

Golgari Grave-Troll is the safest unban you can get. It's absolutely unplayable in its historic decks, and no current deck wants it. We got some decent results with a Goryo's Vengeance reanimator deck using a dredge engine to draw, but it was still unimpressive, and if it were modern legal the results we got were not good enough for me to pursue the deck further.

Ancestral Vision is a safe unban. It's not unplayable like Grave-Troll, and indeed I found it to be quite strong, but it's not banworthy. We got solid results with it in UW/X control (esper and UWR), and it would certainly be worth playing there, but the results still weren't amazing enough to make the deck too dominant. We also did some limited U/B faeries testing with it, but still didn't get strong enough results for us to worry about it. I'm going to say this explicitly: I don't believer faeries is a good deck in modern. It's playable now, and might be decent with Ancestral Vision, but unless we missed a major card choice that significantly ups its power level, Faeries will not be tier 1.

Bloodbraid Elf is probably a safe unban. We got some scary results with Jund (a little bit better than our testing with Nacatl), but it didn't feel particularly broken. The most notable thing was that it had a quite poor matchup against UWR and especially UWR twin, which suggests that there's a fundamental check on the power of Jund in the format with Deathrite Shaman gone. I'd definitely want to see more testing of this card before I felt comfortable unbanning it, were I the DCI, but it's certainly not inherently dangerous.

Seething Song is debatable. We tried storm with seething song, and found it to be quite powerful, probably powerful enough to make it a recurring contender in the format. However, we found a 15% turn-3 goldfish rate so far, and this was done against our standard modern gauntlet with no sideboard changes. Given that it performed well but not incredibly well (worse, for instance, than zoo with Nacatl did during our testing), it would probably only mean people would have to devote more sideboard hate to storm. From a format dominance and enjoyability perspective, Seething Song is not a threat. From a turn-4 rule perspective, it could be in violation, but I don't think it has a high enough percentage. It's also worth noting that I am a longtime storm combo player, as are two other members of my team, which in my opinion gives this testing more credence (we're less likely to misplay a storm deck).

Sword of the Meek is NOT a safe unban. Dear god is this not a safe unban. The combo is so good, a control shell that can consistently get it out by turn 4 (not that hard if you play any Muddle the Mixtures) doesn't really have to worry about aggro matchups, so they can just play a ton of counters and discard with Snapcasters and Thirst. We tried it in UWR, Esper, and U/B tezzeret. Every single one was extremely strong, with Tezzeret testing roughly equally to zoo with Nacatl and the other two testing vastly better. Note: this does not mean that Sword is absolutely not unbannable. It's entirely possible we missed a couple of powerful hate cards or interactions that make Thopter/Sword weaker, or that new printings over the next couple months will make it weaker. What it does mean is that it's not a safe unban, meaning that, based on our testing, there are real and credible risks that sword will be dominant.

There are other cards we've done minimal testing for (Mental Misstep, Chrome Mox, and Blazing Shoal), but none of those are actually for potential unbans, but rather for reasons of figuring out more general things about approaching banned lists. We haven't done enough testing for those to be relevant, so I can't make any predictions about them.

Amusing as that may be today, that was a very different format. I won't stand by that now, but there was absolutely a window, between the printing of Avacyn Restored and Return to Ravnica, where this deck was just unstoppable. My team stayed on it the entirety of that time, and over the course of several tournaments with it the four copies of the deck we scraped up paid for themselves several times over. You had enough of a control suite to play poor man's UWR if you had to, control the game with the best of them, while also throwing out a Griselbrand and winning the game even if they dealt with it immediately. Or just go off with a huge Past in Flames. Deathrite Shaman, followed by the banning of Seething Song and the printing of Scavenging Ooze, changed all of that. But for those few months, Ritual gifts was absolutely insane. Your opponents could never tap out for fear of Griselbrand or just going off. You could easily play through massive amounts of storm hate. You could present up to four different angles of attack (Gifts Ungiven+Past in Flames, Gifts Ungiven out a Griselbrand, Gifts Ungiven to set up a twin kill, Mystical Teachings for Teferi, Mage of Zhalfir and either Meloku the Clouded Mirror or Inferno Titan), making sideboarding basically impossible even if your opponent knew all the cards in your list. You had basically no bad match ups, since fast aggro at the time largely couldn't deal with a Griselbrand and control matchups couldn't afford to play draw-go or you would win, which meant they had to tap out, so you would win. Fast combo was hosed by countermagic and wasn't particularly good at the time either, and those that were had to deal with very specific, powerful, targeted hate that other decks couldn't afford to play, but Ritual gifts could just slip a singleton into the sideboard and consistently have it by turn 3 the majority of games in the match.

Setting aside the past for a moment, does anyone have some matchup information on the BUG Volute lists people are talking about? I haven't tried these out, but I'll add them to the primer if someone can give me a good list and demonstrate some results.

Also, in other news, URw is proving surprisingly good. I've wound up cutting down on rituals, leaving it mostly up to Goblin Electromancers, Manamorphoses, and a couple dedicated rituals to go off, with more control elements. The plan is basically: play a poor man's UWR control for a bit, while using free cycling on Manamorphose, eventually Gifts up a big Past in Flames or Griselbrand, which leads into a huge Past in Flames. I'm also really liking Aetherize, and it's helping a lot to keep the aggro decks in check.

copied Lightning Bolts, chaining Manamorphoses and cantrips, and finishing with a grapeshot in the 12-13ish range. Combined with the bolts, that's usually enough. It was basically a much more controlling version of storm.

There are a couple of other concerns about Stoneforge: the first is that even without batterskull it would be trivially easy to slot it into a UWR package with a Sword of Fire and Ice, Sword of Feast and Famine, and maybe a Loxodon Warhammer or something to bring the lifegain, and then be able to get multiple activations with Restoration Angel. Also, it DRAMATICALLY restricts the kind of equipment R&D can print forever, in a way that a lot of other banned or bannable cards just don't for other types of spell, and equipment is something R&D rather likes being able to play with. Finally, the ability to just search out any equipment and have a body to put it on is a lot more powerful than people think, and would give a ton of decks a whole new angle of attack, mainly RUW twin and midrange/control. It's not like Stoneforge/Batterskull swap would be the dumbest idea ever, but it still scares me a fair amount.

Knowing what pros you test with is completely relevant. If they were biased towards combo-control and favored decks like Scapeshift in the past, it would completely explain the disparity between your testing and the real world, and put to rest this whole matter. If you named your test partners, it would become a lot less likely that you're just a bad player who makes bad predictions and more likely that you're a good player whose testing was inaccurate for some quantifiable reason.

And all he said was that it uses the same mechanism of tutoring/mana cheating. That also describes Natural Order, which is strong but hardly format-ending, Defense of the Heart and Pattern of Rebirth, which were strong but are unplayable today, Bribery, which is terrible, and the Moggcatcher cycle, which are unplayable. Notice a trend? It turns out that just because a mechanic is broken doesn't mean all the cards that have it are good, and even the cards that are good have relative power levels. Storm is broken. Mind's Desire is broken. Astral Steel is not. Tutoring and cheating into play is broken. Bribery, however, is not broken. Stoneforge Mystic is broken. Tinker, though, Tinker is one of the best cards of all time. To say Stoneforge Mystic is on par with Tinker is as ridiculous as saying that's what Aaron Forsythe said in that video.

It's not, that would be arguing from authority, which is either the fourth or sixth formal fallacy, depending on what authority you're taking the list from. But, no one's really doing that here. When people quote pros, they aren't saying "this is true because the pro says so", they're saying "this pro said it better than I could, read his article". It's possible I'm wrong about that, but if I am, people need to stop quoting pros.

Who did you test with after valakut, specifically, by the way? If they were dedicated combo players, that would easily explain that discrepancy and we could all put the whole "Bocephus was wrong about Valakut" thing aside.

Can I have a list for that? Also, Leveler?