MTG Salvation

Yeah, one of the ways Lithoform Engine goes infinite is by copying an artificial evolution to bounce simian spirit guide without spending red mana. Wouldn't be surprised if there were more ways too.

Edit: How do we recuur Dawnglow Infusion? It needs to not bypass the flashback that triggers the hyperstage.

@Anacalagon - Yes, if you had some way to scale the numbers of everything at the base level, such as Replicate, you would get more layers. You won't though get to any point where untapping Unbender Tine is worth X layers, without a more advanced technicque we call a stage combo; rather some number bigger than 3.

@Deedlit:
Going over every layer in 3090:

Angelheart Vial,Rings of Brighthearth,Howling Gale,Boon Reflection
Probably intended: 1 damage = X vials trigger; 1 vial = X copies of the life gain ability; 1 copy of the ability = X life - 3 layers
Reality: Boon reflection breaks the hyperstage, so can't be used. Vials also get reset in between worldpurges. 1 damage = X vials trigger, which all must be used before the worldpurge, so just gain X life - 1 layer.
If vials can be protected by soul scuplptor (or a pretty convoluted phasing setup) then it's 2 layers.
A concern is that vials might be able given to the opponent via mirrorweave - is that an issue?
Finally, the layer trigger, Howling Gale, doesn't work since we don't have access yet to green mana.
Ashcloud Phoenix could be an option? It would be able to leave behind a morph during a computation, but that should be OK since morphs have no creature types so it would quickly die.

Mournful Zombie
1 white = X copies of ability = X opponent life - 1 layer (without the boon reflection)
This works, although it needs to be given haste so the Marchessa's Smuggler and Eye of Ramos need to be used here.

Plains, Caged Sun, Amulet of Vigor
Intended: 1 plains ETB = X amulet triggers; 1 amulet trigger = X white mana - 2 layers.
Reality: The plains gets bounced by the worldpurge, making the amulet triggers do nothing. 1 plains ETB = X white - 1 layer.

Firebrand ranger
1 green = X plains - 1 layer - this works

Gaea's Cradle: Same issue as the plains. 1 layer rather than 2.

Helldozer - I'm not sure how this actually reuses the cradle

Lotus Petal, Morturary
1 card draw = play lotus petal with March out, remove March before DN triggers; X mana - 1 layer - works

Dragon Appeasement, Pitchstone Wall, Curse of Chaos
Intended: 1 combat step = X discards; 1 discard = X sacrifices; 1 sacrifice = X draws
Reality: Things that trigger on discard don't work for 2 reasons. Firstly, in this deck we're using Gerrard's Verdict. The deck in 3090 wasn't, but it still wouldn't have worked for the second reason: The cleanup step is an infinite discard outlet, since discarding to handsize, if it triggers an ability, gives players priority after which there's another cleanup phase. Only works if somehow the setup were to require casting a spell that gave no max handsize for the rest of the game.

Finest Hour, World at War - each make X combats - works.

You might be misunderstanding what I mean by layers. A layer is a system which takes in one (or a small amount) of a resource (mana, life, triggers that untap a certain thing), and outputs X of another resource, where X increases over time. X of this other resource repeats everything you've done so far X times.

As a simple example, lets say your base combo is that spending a red mana doubles the value of X, spending a green mana generates X red mana, and spending a white mana generates X green mana. As X actually increases as the combo progresses, it's really referring to a different value each time. Say it starts off as 3, and you have 2 white mana. The combo looks like:

Spend a white mana, get 3 green mana
- Spend a green mana, get 3 red mana
- Spend a red mana, set X to 6
- Spend a red mana, set X to 12
- Spend a red mana, set X to 24
- Spend a green mana, get 24 red mana
- Repeat this 24 times: Spend a red mana, set X to 2X. Overall, X becomes about 2^24
- Spend a green mana, get 2^24 red mana
- Repeat: Spend a red mana, set X to 2X - overall, X becomes 2^2^24
[By this point, the action of spending a green mana can be shortcut to setting X to 2^X]
- Spend a white mana, get 2^2^24 green mana
- Repeat that many times: Spend a green mana to set X to 2^X. Overall, X becomes 2^^X

And so on, each layer adding an up arrow.

Your sequence, if it worked as intended, would be somewhere between 8 an 10 layers.
There are ways to get X layers, but they are much more intricate.
However, this issue with your sequence is that parts of it don't scale - which is why I only counted 3 layers. (there might be a few more). Unless I'm missing something, once a creature or enchantment is on the battlefield, you aren't making more copies of it until much later in the sequence. Same with rowan embelems - although you are reusing rowan later in the sequence, you aren't making more embelems every time you run though this part of the combo.

Imagine if in my example layer system the step was instead "spend a green mana to make 1000 red mana". Then, you can see that once X reaches 1000 (which it does quickly), I'm worse off than in my original combo. Now if you imagine the step was "spend a green mana to make an amount of red mana equal to the number of creatures you control", and a later step was "spend a blue mana to make X creatures", where blue mana was only used in a much later sequence, then I'm really in the same sequence - say before I use any blue mana, I can make 1000 creatures, then I'm in the same situation; then later in the sequence I spend a blue mana to make (say) 2^^^^1000 creatures; which appears to make the lower sequence slightly better, but still not as good as a proper layer would have been for the same reason The final count is going to get much bigger than 2^^^1000 so that factor didn't really do much.
This is basically what you're doing with embelems and other stuff.

Edit: Deedlit's post is exploins what exactly is going on here - you're repeating things a number of times described by expressions like 2^D*I*(Y+1)*M*W*(K+1)^2, but none of those numbers are actually proportional to X, which if they wore it would be worth more layers. In the next step, the expression involved the storm count, which IS proportional to X, so it is worth a layer.

Another infinite is that a token copy of Hoard of Notions can repeatedly cast the original from the grave, killing it via legend rule, to generate mana.

You have a lot of cards/steps that basically do the same thing - e.g. casting blue spells and paying energy both individually untap the unbender time, but the impact of this is going to be dominated by whichever of those things you can do more of, making the other not really matter. It's more efficient to chain your resources together in "layers" - for example, using X to represent the amount of each resource accumulated before it, each time you cast your token maker, you get X copies of it (thanks to TYS), each of which gets you X intruder alarm triggers, each of which generates X mana with Selvala - which is 3 layers.

The other thing you want to do is scale your resources (i.e changing the value of X), so that each layer truely makes the next iteration of it better - for example, if you had a way to scale the number of lands you have, then each intruder alarm trigger untapping a Magus of the Candelabra would untap X lands, each of which untaps selvala, which is worth an extra layer. wheras previously untapping a magus didn't actually do much because it would only untap a small number of lands. Similarly, if you had a way to re-use Rowan you could scale the number of embelems which would be worth a few more layers in this section.

Since computations need to be initiated by killing all of our artifacts and creatures with something like Desolation Giant or Death Pit Offering, if the inner stage is creature based then its tokens need to be hidden from this effect. Soul Sculptor can be used for this. Then the hyperstage spell needs to hit all permanents (or all nonland permanents).

I guess if the inner stage were life-based or creature-based and didn't rely on Cowardice, and the hyperstage was mana-based using Worldfire, then there wouldn't be any unwanted bounce effects making a mana-based or creature-based stage easier

Now that standard is done, I've been thinking about this a bit.

Here's the current deck last time we left off:



Things that are off limits include:
- Dealing damage to any creature, other than via Goblin Boom Keg
- Killing any creature or artifact we control without killing them all
- Giving any creature haste, unless it can't attack or can't be blocked
- Controlling a necromancer during a computation
- O-ring effects that can be creatures (else they can allow a necromancer to exist during a computation)
- Doing anything at instant speed that could interfere with a computation
- Returning instants from the graveyard to our hand

Resources used:
- Colourless mana
- Red mana
- Life

We'd like to now build as many layers as possible on top of the life hyperstage.
Doing so will be tricky, because firstly any resources that are permanents will be bounced by the worldpurge, and secondly because the cowardice can bounce any creature we may want to use. #3090 contains some good ideas though not all of them work.

Ah yes, also true.

Ah yes, that's true.

(Crushing Canopy, Embereth Sheildbreaker) -> (Collision // Colossus, Return to nature) looks a little more efficient - same number of names, less mana, and an extra card for cling to dust. But I probably wouldn't bother updating everything just for that now that it's published.

Great!

I'd maybe add a sentence at the end on how to actually direct all that damage at the opponent's face (i.e. sink all excess green mana into a leyline to pump a leafkin). (a few people have asked about that in the reddit thread).

Also it looks like the comments aren't visible in the public version.

Edit: lovestruck beast is slightly more efficient than embreth sheildbreaker

We'd like to re-draw the gemrazer between every few untap triggers, rather than only on draw triggers, for the extra layer.
I don't think the extra mutator is required for pinging nikya, as the porcuparrot can be mutated onto something else with red mana.

Do we want to use all the draw triggers of the spells cast before all the stages are set up? We have to in order to get to the dances to get the prophecies back, unless we cast Assure somewhere first.

If so the sequence will be something like

Use black stage for f_{w}(X) prophecies where X is small (about 5-10)
DCU; Use krakens to untap skull prophets for X -> f_{w2+1}(X) [1 layer: 1 draw trigger = X kraken triggers; worth 2 black mana each]
Dance prophecy, copy it a bunch
Gemrazer - X draw triggers - X -> f_{w2+2}(X) [2 layers: 1 draw = X krakens; worth 1 life each [untapping parrot]; 2 life = X black [flicker]]
Jaya - X -> f_{w2+2}(X)
Awaken - X -> f_{w2+3}(X) [3 layers: 1 draw = X krakens; 1 kraken = 2 red [can't yet use Nikya without being able to recur gemrazer after getting rid of nikya]; 1 red = X life [leafkin]; 2 life = X black]
Light up the stage - X -> f_{w2+4}(X) [4 layers: Nikya works now - flicker kills nikya tokens, porcuparrot kills nontoken]
Greathorn - X -> f_{w2+4}(X)
Chop Down - X -> f_{w3+2}(X) [full life stage]

And the rest of the 25 names have a full set of stages and layers under the draws for f_{w3+2} each time.

(Assure could be used next so that the rest of the draws can be saved for whenever we want)

Hmm; so the last state is 9G1U10B1W with 8 revivals on the stack and some soots under it (including the real cards) and pixie + crushing canopy in the yard.

Resolve revival: 12G3U10B1W
Kinnan Shipwreck: 6G2U10B1W
Quasiduplicate: 5G10B1W x9 - say something like 4 TYS 3 leyline 2 dowser (quaduplicate, crushing canopy)
Crushing Canopy: 2G10B1W
Resolve Revival: 8G2U10B1W
Repeat until out of revivals: Quasiduplicate, Crushing Canopy, Resolve Revival: lots of G, 2U10B1W; real revival is in yard now
Quasiduplicate to return revival: 10B1W
Resolve Soot
Revival: 2U10B1W
Canopy, resolve revival: 4U10B1W
Kinnan Mystic: 3U10B1W; soot and revival in yard

Now we can do enough of the full black stage to profit enough blue for the next part; DCU has not been needed.
Dance Prophecy
Copy it a bunch
Kinnan proliferate creature
Name DCU on all prophecies, then proliferate and cast DCU.
Don't resolve any triggers yet

Now we can do the full blue and black stages, then when it finally comes to resolve a draw trigger we can get a ton of tolarian krakens first, and set up the life stage from there.
The prophecy gets destroyed but we can then eventually dance it back off of TYS triggers, and use Assure next if we want.


So; we can use DCU as a prophecy name. I don't think it's worth as much as the other names just since the number of prophecies is limited by not having the full blue stage available, but it is worth something.

As for the discard, I'm not sure how to meaningfully compare the impact between using bad deal (thus only discarding very late in the sequence) vs using cavern whisperer (drawing the same number of cards from cling but much earlier in the sequence; though slightly lessening the efficiency of the life stage) vs using harvest fear (which loses a card for cling to exile making it probably the worst option)

We don't need cavern whisperer since their hand is going to be discarded to the Bad Deal in the end step.

We could replace Courage in Crisis for a creature that proliferates such as Pollenbright Druid to get an extra name for prophecy.

Invert goes infinite by allowing a porcuparrot to kill itself without a chop down, but can easily be replaced by another split card. Flower // Flourish looks like the most efficient way to burn excess green mana into storm count.

Ah, good point.

We could get the last few draws from Cling if we sac the excess enchantments to the citadel in the end step after having resolved all of the gadrack triggers, where we won't have to ever resolve the citadel ability and cost the opponent life.

Edit: Cutting the extra enchantments (and using crushing canopy) still seems worth it as it trades one name (Colossus) for 4.