It looks like in order to reuse the Zubera, we'll need to cast Caress one more time without TYS up, then before it resolves, use Worldfire and Riftsweeper to get the Zubera back into our library to draw it. Then from there we'll have to keep using Zubera triggers to redraw the Zubera, using Vedalken Orrery and Twinning Glass to make sure we can take necessary actions at instant speed. (The Caress draw would probably have to be Zubera, Infatuation, and Spite, I guess? Then we could play those from prepared Twinning Glass activations, to rebuild our hand with a Zubera trigger to spare.)
Does that all sound right so far?
Edit: In this scenario, I don't think we can get TYS back until after the first Worldfire. Is that a problem?
After playing around with possible next steps from there, I think I have a better picture of how the deck operates. One thing I'm still not sure of is, what's the second Spellweaver Volute for?
We need two Spellweaver Volutes to enchant both Fated Infatuation and Kaervek's Spite when we do a hyperstage transition; the Fated Infatuation to build the higher stage, the Kaervek's Spite to get our life back while destroying our resources. We need more instants before doing a computation (Artificial Evolution, Arcbond, and {c]Wrong Turn[/c] after a Worldfire), but we can just do two or three hyperstage transitions, casting Karevek's Spite and one other instant each time, I think. But one Volute won't be enough for the normal hyperstage transition.
Sorry, I'm new to the forum but thought I should share that for the past couple days I've been working on a list that uses Devilish Valet, Sedgemoor Witch, Cabaretti Confluence targeting sedgemoor, 25 storm spells (the best of which being Empty the Warrens), plus 9 regrowth effects that exile themselves (like Vengeful Rebirth) copied as many times as I need them to be with Thousand-Year Storm to get all the storm cards back. Not sure if it holds up compared to the multitude of other thoroughly researched strategies, as it's probably been well discussed already, but might as well mention it. Let me know if you want the extremely rough decklist so far.
Once again, this is extremely rough and I do see extreme inefficiencies right now, but this is what I have so far. It's mainly just a proof of concept.
So, it looks like the main advancement in the deck comes from Fungal Sprouting; comparatively, the cards that add tokens on the order of the number of spells so far, will add tokens numbering in the thousands, compared to much more than a googleplex for most of the Fungal Sproutings. Tend the Pests does something similar, but only if you sacrifice the Devilish Valet, which seems to counter our progress.
Each copy of Fungal Sprouting takes Devilish Valet from X power to about 2^X power. So if we resolve X copies of Funal Sprouting in the game, we will achieve roughly 2^^X damage. We can cast 4 Fungal Sproutings, get them back using Dryad's Revival for 8 castings, and retrieve both the Fungal Sproutings and Dryad's Revival 8 times using our exile spells, for 72 castings. Each cast gets a bunch of copies from the Thousand-Year Storms; this increases throughout the combo, but on average it looks like we will get a few hundred per TYS, for a little over a thousand from all 4 on average. So, rough estimate of 100,000 copies of Fungal Sprouting, for about 2^^100,000 damage in the end.
Deedlit, the first draft would be playing lotus into show and tell into omniscience, then casting devilish valet, archamge emerius to draw, thousand year storm for value, and then cabaretti confluence targeting the devilish valet to get 3 more, drawing into any storm spell to get ~11 card cast/copy triggers, draw 11 cards and from there cast the rest of my thousand year storms, sedgemoor witch, copy sedgemoor a bunch, eventually casting platinum angel as to not lose via decking myself from the draw. I'd just cast all my value spells, then cast all my storm spells including Haze of Rage using birgi mana to buyback it quite a few times, then cast my playset of fungal sprouting, after that casting my 3 copies of tend to the pests each sacrificing one of the devilish valet copies from earlier. Then cast one of the self exile regrowths getting all my instants and sorceries + lotus back, and repeat the entire thing until I run out of regrowth effects. Very rough, haven't really gone into specific lines yet.
hyp_cos there isn't really a good reason to run multiple copies of the permanents, that's just out of my eagerness to get a rough functioning list together. Copies would definitely be better. As for the spells, they are harder to copy in meaningful quantities compared to thousand year storm, so I just opted to run multiple copies instead.
Apologies for the lack of card tagging or whatever it's called, I just wanted to send this reply ASAP. Let me know if I should add them/it's too annoying to look up the cards.
I try to keep tagging cards in my posts, for other people's convenience, but it's not really a big deal, I don't think.
To expand on what hyp_cos said: So by concentrating on getting a lot of copies of your spells, you've managed to be able to resolve about 100,000 copies of Fungal Sprouting. I didn't include Tend the Pests, but I suppose we would get about 75,000 of those as well, for roughly 2^^175,000 damage at the end of the combo. Perhaps with more optimization, you can pump the latter number to around a million. A million is a large number in most cases, but consider this: What if you found a way to convert a creature's power, or number of creatures, into the number of times you could cast or copy Fungal Sprouting? So if you started with 10 creatures, you would get 10 copies of Fungal Sprouting for around 2^^10; not a big deal, but if you did it again, then you would get 2^^10 copies of fungal sprouting, resulting in Devilish Valet getting a power of 2^^2^^10. This is much bigger than you could ever hope to get by putting in more instants/sorceries, more copies of Thousand-Year Storm, and things like that. And you could perhaps keep going to 2^^2^^2^^10 and beyond. N repetitions of this procdure would get you more than 2^^^N, and you would have achieved 3 Knuth arrows, which will beat any 2 Knuth arrow deck (provided N isn't too small). So this is a better goal than just more copies of your best spell: create more "layers" of recursion, and get more Knuth arrows.
Of course, this is easier said than done. It's not immediately obvious how to turn number of creatures into spell casts, it may require some creative thinking. But that's part of the fun!
You're right! I'll spend a while looking into that. I think I need to get in the right mindset of 'casting a few copies of a spell is nothing compared to actually multiplying', I need to be more focused on making everything exponential. Thanks for the feedback, it's really great to have others helping out!
Also, sort of related but not fully, why do some not open with Black Lotus into Show and Tell into Omniscience? At only 3 cards it seems like the most efficient way to get omniscience out. Is it because other cards have the ability to deal more damage in certain builds?
Cast Alena, Kessig Trapper. Find a way to give her haste, doesn't really matter how. Cast Intruder Alarm. Whenever you get any number of tokens, in between each Intruder Alarm trigger, tap Alena, Kessig Trapper for mana equal to Devilish Valet's power. Generate an absurd amount of mana this way, sink it all into Storm King's Thunder. Cast Fungal Sprouting and get an equally absurd number of copies. In between each Saproling etbing, do the Alena, Kessig TrapperIntruder Alarm thing to get even more mana. Repeat this process as many times as possible. Alternatively, just use Mana Echoes + Maskwood Nexus to get mana, that would be far less efficient compared to the previous combination of cards though, so probably not worth bringing up.
Unfortunately, our decks typically have lots of ways to go infinite, which winds up ruling out a lot of cards we would really like to use. In our latest deck, there's a lot of dividing things things up by various resources, particularly mana color; black mana is the cheapest, and gets the smallest payoff, red mana is harder to get and gets more stuff done, and so on. Right now, we retrieve Mox Ruby form the library to get our red mana; if we had Black Lotus, we could also get the more valuable blue mana and white mana for the same cost, so this goes infinite.
Show and Tell is also a problem, since it allows the opponent to play their Wastes (we assume the opponent has a library of 60 Wastes), and this allows us to donate a Soul Foundry to the opponent, and they can use it in a way that allows us to achieve arbitrarily high numbers. This doesn't allow us to "go infinite" without opponent cooperation, but our rules state that a deck is disqualified if it can deal arbitrarily high damage, opponent cooperation included. So we can't allow the opponent to get mana either.
For that very reason, Channel is no good either, since we if we had it we could imprint it on a Spellweaver Helix and donate the Helix to the opponent, allowing them to cast Channel for themselves and get colorless mana.
Believe it or not, our start turns out to be: Archaeological Dig / Mox Ruby / Mox Sapphire / Mana Crypt / Shriekhorn / Academy Rector / Caress of Phyrexia. 6 cards to get Omniscience, and the 7th draws 6 cards to barely be able to get going. Obviously, a more powerful start would be preferred, but this is just how it worked out given the cards that we can't use, and our unwillingness to use up card slots on the opening given that this is enough to get started.
Some of the small decks (see the "To Graham's Number and Beyond" thread) purposefully use a different start; for example Black Lotus / Channel / Mycosynth Lattice seems to be inferior, because we still have to pay life for our spells, but it also turns everything into artifacts, and allows us to pay for activated costs with colorless mana/life, which Omniscience doesn't help with. So that could be advantageous. A recently posted deck started with Black Lotus / Channel / Nature's Spiral / Elixir of Immortality, which worked out well to keep recycling Black Lotus, allowing us to pay for activation costs, X spells, and flashback in addition to normal spell casting. Also, it was a good way to maximize the number of instants/sorceries in the deck, which was advantageous in this deck.
So, it's okay to do a little rounding to make things simpler; something you learn when calculating with these fast-growing functions is that, rounding errors don't propagate over time, instead they go away. This is a little hard to explain, but the reason is that as we get to higher and higher number notations, the difference between even close notations get vaster and vaster, so even as rounding errors get bigger and bigger, they don't get bigger as fast as our notations do. This is why we are able to divide things up in terms of layers; there no such thing as half a layer, if an operation fails to achieve another layer, it will be a small improvement that is basically just a rounding error.
So, let's look at the mana produced by Alena. When we resolve a copy of Fungal Sprouting, we'll create X tokens, each of which trigger both Alena and Devilish Valet. Clearly, it is better to resolve the Devilish Valet triggers first. Devilish Valet will have power X, and will double its power X times, to get X * 2^X power. We can tap and untap Alena X times, to get X^2 * 2^X mana. This is where the rounding comes in; X is much smaller than 2^X, so the mana we get from Alena is basically equal to Devilish Valet's final power. Also, the mana from previous Fungal Sproutings are much less; the previous Fungal Sprouting resulted in the Valet having a power of X, so the mana we got was about (log_2 X) * X (or roughly X), and the mana before that was even smaller, and so on. So the mana we get is pretty much the final power of Valet.
So, CaptainMarcia pointed out that Kaervek's Spite won't trigger things that trigger on its casting, like Thousand-Year Storm because Kaervek's Spite will sacrifice our permanents before the time when they will trigger. This seems to cause a problem with Cephalid Shrine. I imagine there's a 1 for 1 replacement that we can make to fix this.
Edit: Oh, I'm being silly - we don't need to counter our instants, since they get copied by Spellweaver Volute.
Its a bit inefficient as we don't cast omniscience in the initial setup, so we need to cowardice bounce and replay it once before truly going off (after copying it with like Fated Infatuation so we don't have to actually spend UUU on it.)
The only other problem with the decklist is that card 45 is intended to be Neurok Transmuter. Which I keep expecting to cause problems by allowing artifacts to be typeless, but its actually just convenient for a donated Coat of Arms to not care about march of the machines anymore.
Edit: Could we get more mileage out of dropping Precursor Golem to run both Zubera and Sphinx? We'd need the Zuberas to die on the opponent's board (with Prodigy tokens on their side as well), but after computations, we can kill damaged typed creatures on the opponent's field with Artificial Evolution.
Edit: Sphinx goes infinite if the opponent cooperates. Psychic Possession doesn't get the benefit of Prodigy, but it should be able to avoid that.
@jfb1337: So the idea behind Smoke Spirits' Aid and Precursor Golem, is that we we would have a bunch of Precursor Golem triggers for SSA on the stack. We resolve one of them, putting a bunch of enchantments on all the Golem creatures. Then we kill the creatures (say with Kaervek's Spite or the computation), triggering the creating of a bunch of Treasure tokens, which we can sacrifice for a bunch of white mana.
@CaptainMarcia: Floating-Dream Zubera gets us more than Consecrated Sphinx (if it worked) or Psychic Possession off of Caress of Phyrexia. Each time we kill a FDZ, we can draw many cards thanks to all the copies of Harmonic Prodigy, and we can do so thousands of times. The opponent draws three times off of Caress of Phyrexia, and we get a bunch of draws from Consecrated Sphinx or Psychic Possession, similar to FDZ off of Harmonic Prodigy. So it's 3 versus thousands.
FDZ going into Psychic Possession is better than using Caress of Phyrexia. Having Psychic Possession adds a layer, same as Precursor Golem, so it's the same number of layers for both, and the difference is how many initial draws we can accomplish. Looking at it again, it seems like, for both us and the opponent, the library can be refilled back up to 60 cards. (I guess that's something that should be checked carefully.) So both ways we can kill FDZ 246*60 times once we get the full combo up and running. But, I guess the difference is that with the Psychic Possession version, we have the 3 draws from Caress of Phyrexia as well. So it looks like BB_{w^3 + w2 + 3}(14,760) with Precursor Golem, and BB_{w^3 + w2 + 3}(14,763) with Psychic Possession, assuming we can get started with just the 3 CoPs on our side. So yeah, good call.
I think the reason we've been prioritizing other layers over Psychic Possession, is that, when we had Worldpurge, the opponent could keep 7 cards in their hand, so we could only refill the library to 53, and the input number was therefore 53*number of creature types rather than 60*number of creature types. But with Worldfire it looks like we can get 60.
Opening hand: Mox Ruby, Mox Sapphire, Mana Crypt, Archaeological Dig, Shriekhorn, Replenish, and Caress of Phyrexia
Play the mana sources, mill Omniscience and Thousand-Year Storm, reanimate them.
Caress draws Floating-Dream Zubera, Archon of Falling Stars, Comeuppance, Divine Congregation, Fated Infatuation, and Kaervek's Spite. Play the first four. With four Congregations on the stack, play Infatuation to make five more Zuberas, then gain 56 life. Play Spite x6, killing our creatures, putting our opponent at 5 and ourself at 55, and letting us reanimate Omniscience and draw 36 cards.
It looks like in order to reuse the Zubera, we'll need to cast Caress one more time without TYS up, then before it resolves, use Worldfire and Riftsweeper to get the Zubera back into our library to draw it. Then from there we'll have to keep using Zubera triggers to redraw the Zubera, using Vedalken Orrery and Twinning Glass to make sure we can take necessary actions at instant speed. (The Caress draw would probably have to be Zubera, Infatuation, and Spite, I guess? Then we could play those from prepared Twinning Glass activations, to rebuild our hand with a Zubera trigger to spare.)
Does that all sound right so far?
Edit: In this scenario, I don't think we can get TYS back until after the first Worldfire. Is that a problem?
Edit: Wait, no, Spellweaver Helix can let us reuse Replenish. That helps.
https://www.moxfield.com/decks/eC-89_RT1kGoay0419KNIA
Common quantitative reasons may read like "we add more CARDNAME so that we can benefit more from it".
Qualitative reasons are various, including:
Quantitative reasons are bad, because you may replace the extra cards with some cards that makes a higher layer.
Each copy of Fungal Sprouting takes Devilish Valet from X power to about 2^X power. So if we resolve X copies of Funal Sprouting in the game, we will achieve roughly 2^^X damage. We can cast 4 Fungal Sproutings, get them back using Dryad's Revival for 8 castings, and retrieve both the Fungal Sproutings and Dryad's Revival 8 times using our exile spells, for 72 castings. Each cast gets a bunch of copies from the Thousand-Year Storms; this increases throughout the combo, but on average it looks like we will get a few hundred per TYS, for a little over a thousand from all 4 on average. So, rough estimate of 100,000 copies of Fungal Sprouting, for about 2^^100,000 damage in the end.
Deedlit, the first draft would be playing lotus into show and tell into omniscience, then casting devilish valet, archamge emerius to draw, thousand year storm for value, and then cabaretti confluence targeting the devilish valet to get 3 more, drawing into any storm spell to get ~11 card cast/copy triggers, draw 11 cards and from there cast the rest of my thousand year storms, sedgemoor witch, copy sedgemoor a bunch, eventually casting platinum angel as to not lose via decking myself from the draw. I'd just cast all my value spells, then cast all my storm spells including Haze of Rage using birgi mana to buyback it quite a few times, then cast my playset of fungal sprouting, after that casting my 3 copies of tend to the pests each sacrificing one of the devilish valet copies from earlier. Then cast one of the self exile regrowths getting all my instants and sorceries + lotus back, and repeat the entire thing until I run out of regrowth effects. Very rough, haven't really gone into specific lines yet.
hyp_cos there isn't really a good reason to run multiple copies of the permanents, that's just out of my eagerness to get a rough functioning list together. Copies would definitely be better. As for the spells, they are harder to copy in meaningful quantities compared to thousand year storm, so I just opted to run multiple copies instead.
Apologies for the lack of card tagging or whatever it's called, I just wanted to send this reply ASAP. Let me know if I should add them/it's too annoying to look up the cards.
To expand on what hyp_cos said: So by concentrating on getting a lot of copies of your spells, you've managed to be able to resolve about 100,000 copies of Fungal Sprouting. I didn't include Tend the Pests, but I suppose we would get about 75,000 of those as well, for roughly 2^^175,000 damage at the end of the combo. Perhaps with more optimization, you can pump the latter number to around a million. A million is a large number in most cases, but consider this: What if you found a way to convert a creature's power, or number of creatures, into the number of times you could cast or copy Fungal Sprouting? So if you started with 10 creatures, you would get 10 copies of Fungal Sprouting for around 2^^10; not a big deal, but if you did it again, then you would get 2^^10 copies of fungal sprouting, resulting in Devilish Valet getting a power of 2^^2^^10. This is much bigger than you could ever hope to get by putting in more instants/sorceries, more copies of Thousand-Year Storm, and things like that. And you could perhaps keep going to 2^^2^^2^^10 and beyond. N repetitions of this procdure would get you more than 2^^^N, and you would have achieved 3 Knuth arrows, which will beat any 2 Knuth arrow deck (provided N isn't too small). So this is a better goal than just more copies of your best spell: create more "layers" of recursion, and get more Knuth arrows.
Of course, this is easier said than done. It's not immediately obvious how to turn number of creatures into spell casts, it may require some creative thinking. But that's part of the fun!
Also, sort of related but not fully, why do some not open with Black Lotus into Show and Tell into Omniscience? At only 3 cards it seems like the most efficient way to get omniscience out. Is it because other cards have the ability to deal more damage in certain builds?
Cast Alena, Kessig Trapper. Find a way to give her haste, doesn't really matter how. Cast Intruder Alarm. Whenever you get any number of tokens, in between each Intruder Alarm trigger, tap Alena, Kessig Trapper for mana equal to Devilish Valet's power. Generate an absurd amount of mana this way, sink it all into Storm King's Thunder. Cast Fungal Sprouting and get an equally absurd number of copies. In between each Saproling etbing, do the Alena, Kessig Trapper Intruder Alarm thing to get even more mana. Repeat this process as many times as possible. Alternatively, just use Mana Echoes + Maskwood Nexus to get mana, that would be far less efficient compared to the previous combination of cards though, so probably not worth bringing up.
Show and Tell is also a problem, since it allows the opponent to play their Wastes (we assume the opponent has a library of 60 Wastes), and this allows us to donate a Soul Foundry to the opponent, and they can use it in a way that allows us to achieve arbitrarily high numbers. This doesn't allow us to "go infinite" without opponent cooperation, but our rules state that a deck is disqualified if it can deal arbitrarily high damage, opponent cooperation included. So we can't allow the opponent to get mana either.
For that very reason, Channel is no good either, since we if we had it we could imprint it on a Spellweaver Helix and donate the Helix to the opponent, allowing them to cast Channel for themselves and get colorless mana.
Believe it or not, our start turns out to be: Archaeological Dig / Mox Ruby / Mox Sapphire / Mana Crypt / Shriekhorn / Academy Rector / Caress of Phyrexia. 6 cards to get Omniscience, and the 7th draws 6 cards to barely be able to get going. Obviously, a more powerful start would be preferred, but this is just how it worked out given the cards that we can't use, and our unwillingness to use up card slots on the opening given that this is enough to get started.
Some of the small decks (see the "To Graham's Number and Beyond" thread) purposefully use a different start; for example Black Lotus / Channel / Mycosynth Lattice seems to be inferior, because we still have to pay life for our spells, but it also turns everything into artifacts, and allows us to pay for activated costs with colorless mana/life, which Omniscience doesn't help with. So that could be advantageous. A recently posted deck started with Black Lotus / Channel / Nature's Spiral / Elixir of Immortality, which worked out well to keep recycling Black Lotus, allowing us to pay for activation costs, X spells, and flashback in addition to normal spell casting. Also, it was a good way to maximize the number of instants/sorceries in the deck, which was advantageous in this deck.
Edit: Nice combo!
One other thing, how would you go about calculating the mana produced by Alena? The numbers seem too big to reasonably work with.
So, let's look at the mana produced by Alena. When we resolve a copy of Fungal Sprouting, we'll create X tokens, each of which trigger both Alena and Devilish Valet. Clearly, it is better to resolve the Devilish Valet triggers first. Devilish Valet will have power X, and will double its power X times, to get X * 2^X power. We can tap and untap Alena X times, to get X^2 * 2^X mana. This is where the rounding comes in; X is much smaller than 2^X, so the mana we get from Alena is basically equal to Devilish Valet's final power. Also, the mana from previous Fungal Sproutings are much less; the previous Fungal Sprouting resulted in the Valet having a power of X, so the mana we got was about (log_2 X) * X (or roughly X), and the mana before that was even smaller, and so on. So the mana we get is pretty much the final power of Valet.
Edit: Oh, I'm being silly - we don't need to counter our instants, since they get copied by Spellweaver Volute.
1 Coat of Arms
2 Xathrid Necromancer
3 Artificial Evolution
4 Arcbond
5 Comeuppance
6 Dralnu's Crusade
Starting a Computation
7 Opalescence
8 Wrong Turn
9 Kaervek's Spite
10 Goblin Boom Keg
11 March of the Machines
Utility: Enchantments/Artifacts
12 Vedalken Orrery
13 Palladium Myr
14 Twinning Glass
Utility: Spells
15 Thousand-Year Storm
Utility: Creatures
16 Bloodbond March
17 Cephalid Shrine
Foundry Stage
18 Armor Thrull
19 Soul Foundry
20 Emblem of the Warmind
21 Cowardice
22 Psychic Battle
23 Riftsweeper
24 Shriekhorn
25 Spellweaver Volute
26 Spellweaver Volute
27 Fated Infatuation
28 Lingering Souls
29 K'rrik, Son of Yawgmoth
30 Archon of Falling Stars
31 Pious Interdiction
Helix Megastage
32 Spellweaver Helix
33 Worldfire
34 Divine Congregation
35 Hungry for More
36 Hungry for More
37 Mox Ruby
38 Panharmonicon
39 Chromescale Drake
Green Mana Stage
40 Groundskeeper
41 Firebrand Ranger
42 Forest
43 Mountain
44 Gauntlet of Might
45 Neurok Transmitter
46 Kurkesh, Onakke Ancient
47 Harmonic Prodigy
48 Ruin Ghost
49 Weathered Wayfarer
50 Archaeological Dig
51 Patron of the Moon
52 Dryad Arbor
Start
53 Mana Crypt
54 Mox Sapphire
55 Replenish
56 Omniscience
57 Caress of Phyrexia
Layers
58 Smoke Spirits' Aid
59 Precursor Golem
60 Floating-Dream Zubera
What is Twinning Glass for, and how do the Smoke Spirit's Aid / Precursor Golem layers work at the end?
The only other problem with the decklist is that card 45 is intended to be Neurok Transmuter. Which I keep expecting to cause problems by allowing artifacts to be typeless, but its actually just convenient for a donated Coat of Arms to not care about march of the machines anymore.
If the concern is getting started, I think we could do an opening hand of Archaeological Dig, Mana Crypt, Mox Sapphire, Mox Ruby Shriekhorn, Replenish, and Caress of Phyrexia, milling Omniscience and Thousand-Year Storm to let Caress draw Consecrated Sphinx, Vedalken Orrery, Spellweaver Helix, Hungry for More x2, and Cephalid Shrine. Then we can replay Caress, get four copies, and counter the original so both players would end up at 9 poison. (The last two would wait until we could get large numbers of Consecrated Sphinx tokens.)
Edit: Chromescale Drake is better than drawing artifacts directly.
Edit: Oh, Harmonic Prodigy can copy the Zubera triggers.
Edit: Could we get more mileage out of dropping Precursor Golem to run both Zubera and Sphinx? We'd need the Zuberas to die on the opponent's board (with Prodigy tokens on their side as well), but after computations, we can kill damaged typed creatures on the opponent's field with Artificial Evolution.
Edit: Sphinx goes infinite if the opponent cooperates. Psychic Possession doesn't get the benefit of Prodigy, but it should be able to avoid that.
@CaptainMarcia: Floating-Dream Zubera gets us more than Consecrated Sphinx (if it worked) or Psychic Possession off of Caress of Phyrexia. Each time we kill a FDZ, we can draw many cards thanks to all the copies of Harmonic Prodigy, and we can do so thousands of times. The opponent draws three times off of Caress of Phyrexia, and we get a bunch of draws from Consecrated Sphinx or Psychic Possession, similar to FDZ off of Harmonic Prodigy. So it's 3 versus thousands.
FDZ going into Psychic Possession is better than using Caress of Phyrexia. Having Psychic Possession adds a layer, same as Precursor Golem, so it's the same number of layers for both, and the difference is how many initial draws we can accomplish. Looking at it again, it seems like, for both us and the opponent, the library can be refilled back up to 60 cards. (I guess that's something that should be checked carefully.) So both ways we can kill FDZ 246*60 times once we get the full combo up and running. But, I guess the difference is that with the Psychic Possession version, we have the 3 draws from Caress of Phyrexia as well. So it looks like BB_{w^3 + w2 + 3}(14,760) with Precursor Golem, and BB_{w^3 + w2 + 3}(14,763) with Psychic Possession, assuming we can get started with just the 3 CoPs on our side. So yeah, good call.
I think the reason we've been prioritizing other layers over Psychic Possession, is that, when we had Worldpurge, the opponent could keep 7 cards in their hand, so we could only refill the library to 53, and the input number was therefore 53*number of creature types rather than 60*number of creature types. But with Worldfire it looks like we can get 60.