Sinking Feeling is way too good in this build. Instead of putting it on Master Biomancer, it can go on any lower creature with at least 2 toughness (hint: this will be all of them), which also holds the Splinter Twin. Make a copy of itself, untap, make another copy of itself. Now one copy turns into two, so one of them is free to play whatever its role in the sequence normally is, and this will easily produce enough colorless mana to move both Auras onto the second token, which has the same initial conditions as the starting point of the loop. Repeat forever.
Hm, right. Exiting into Red Paragon from there doesn't work, because the Paragon will be incapable of cleaning up all of its ranks and that will make the Tunneler incapable of doing its thing as we tunnel back down through the steps (after one iteration of a higher step, there will be at least one lingering permanent that forces the Tunneler up to 3 power, and no way of fixing that except with help from something even higher than that). But Dralnu's Crusade, even with Black Paragon following it up, does work because Dralnu's Crusade is an enchantment and can be dispatched rather cheaply when we need to go across that transition.
Hair-Strung Koto, of course, means that the nontoken creatures (the ones without haste) can tap freely at any time. This invalidates what Royal Assassin is doing; since the Mad Auntie/Goblin Tunneler section just beyond it already has a problem, it may end up being paved over from a lower point anyway.
Even if it did work that way, how does the land get back in hand? Monastery Loremaster can only return noncreature, nonland cards, and adding a slot for Crucible of Worlds is pretty wasteful. If life payment is the goal, Spatial Binding achieves that in a more compact way.
Elspeth Tirel can also serve as a one-shot source of life gain after 19 uses, with no way of animating or destroying it.
The differences between instant and interrupt weren't particularly relevant. The main difference in play was that if you wanted to hold up a counterspell in the interrupt days, it had to be a counterspell: holding up an instant card draw spell with the aim of drawing into a counter wasn't good enough.
Then there are certain variations of counters, like overloaded Counterflux, that couldn't be done sensibly in the interrupt days at all.
Mirrorworks seems disappointingly useless here. I'd like to just cut the card entirely, but there are still two things it's necessary for, which can't be replicated by simply moving around Splinter Twin: getting Mimic Vat imprinted such that it can make copies of itself, and being able to have more than one untapped copy of Crown of the Ages at any given time. A card like Mark of Mutiny or Metallic Mastery could solve the first problem while adding an extra targeted spell, but not the second.
Let's actually run through the possible start sequence to make sure it's enough:
Opening hand: Chrome Mox, Mana Crypt, red card (let's say Garza Zol), Chaos Warp, Shriekhorn, Consecrated Sphinx, Words of Wisdom
Get 2R, Chaos Warp the Mana Crypt into Omniscience
Play Sphinx, then play Words of Wisdom to draw 2: Karn and Mana Crypt (52 cards for opponent)
In response to Sphinx trigger, tap Shriekhorn, mill Emrakul, and shuffle Chaos Warp and Words of Wisdom back in
Redraw Chaos Warp and Words of Wisdom
Chaos Warp the Shriekhorn into Leyline of Anticipation
Replay Mana Crypt, tap for 2 mana
Play Karn, and pay 1 mana to animate Mana Crypt into a 0/0 creature
Words of Wisdom, draw 2: Shriekhorn and Mimic Vat (51 cards for opponent)
In response to Sphinx trigger, play Shriekhorn, tap it, mill Emrakul, and shuffle Chaos Warp, Words of Wisdom, and Mana Crypt all back in
Redraw Chaos Warp and Words of Wisdom
Chaos Warp the Shriekhorn into Mana Crypt, tap for 2 mana (3 total)
Pay 1 to animate Mana Crypt, killing it
Play Mimic Vat
If Words of Wisdom always draws Shriekhorn and another card, and the Sphinx trigger always draws Chaos Warp and Words of Wisdom, this is a loop that gains access to one new card and one more colorless mana each time through.
When opponent goes to 50 cards, draw Psychic Battle, and go to 4 mana (back down to 3 to animate)
When opponent goes to 49 cards, draw Horobi
Play Horobi and Psychic Battle
Play Chaos Warp targeting Consecrated Sphinx, which triggers Psychic Battle and Horobi
Destroy Sphinx, and imprint it on Mimic Vat
Spend all 3 mana and make a hasted copy of Sphinx
Psychic Battle changes the target to Shriekhorn, where it resolves and brings back Mana Crypt again
Tap Crypt for 2 mana, then back to 1 to animate
When opponent goes to 48 cards, draw Splinter Twin
Play Splinter Twin targeting Horobi
Horobi kills itself
Psychic Battle changes target to the Consecrated Sphinx token
Tap the token to make another token
At this point, each loop of Words of Wisdom gets to draw 3 new cards instead of 1. But the mana generation is still slow as ever, as this deck seems to have no actual way of speeding it up with only a colorless 2-mana producer that costs 1 to recycle. Land/Mana Vault starts might be able to do something to take advantage of Mirrorworks, but Mana Vault cannot be recycled so easily with Karn, so that route may actually fall flat even sooner.
Still, even without reusing Chrome Mox, Goblin Tunneler could target the Mad Auntie directly, but only if we can avoid making a Gideon emblem, and have just the one Mad Auntie with no tokens of it, no Dralnu's Crusade, and no black Paragon. Just add two Psychic Battles, then both of them can keep targeting each other without end.
The 24-stager in post 335 doesn't quite work because after bouncing Cowardice and Horobi, Krosan Constrictor's ability can resolve to bring the higher black creatures down into Goblin Tunneler range.
It obviously doesn't make a big difference, but as far as the "just needs to match creature types and colors for imprint" slots, Lumberknot looks better in green Treefolk, and Garza Zol, Plague Queen in blue Vampire (particularly if there ends up being a way to selectively give it first strike).
What does the Archetype/Mine Excavation/Izzet Guidmage/Avarax construction accomplish? Both Archetype and Avarax are red creatures, so there should never be a need to stick Avarax on the Chrome Mox.
Also, we seem to have lost the only land, so there's no way to play Mana Vault at all, other than using the Chrome Mox mana with is needed for color.
For the same reason we can't use Mana Echoes (it allows a token to usefully hold Splinter Twin to perpetuate mana and still do whatever it was planning on doing), I think we have to cut all the main-sequence effects that don't require a tap. In the previous full deck, this is Slayer of the Wicked and Enslaved Dwarf. The later 23-stack looks better even if coincidentally, but what is Ranger of Eos doing here?
If we aren't going to have anything that cares about toughness (other than it being greater than 0), Oathsworn Giant is workable as an anthem effect. It doesn't interfere with power at all, and multiples wouldn't even matter.
Random thought--could we use Karn, Silver Golem in place of March of the Machines? Instead of having to keep bouncing March back and forth to use abilities of artifacts in the face of summoning sickness, Karn means we can just let them be noncreatures by default (Mirrorworks will still trigger off them just fine in that state), and only animate the things we need to for the purpose of bringing them back with Cowardice or having them hold the Splinter Twin (note that Karn's ability specifically targets noncreature artifacts, so it can never cause a Cowardice or Horobi trigger by itself). The 1 mana per individual artifact animation sounds steep, but workable.
If that works, then since there's nothing like Beloved Chaplain or a second Starfield in the deck any more to necessitate protection tricks, and no abilities where we need to bounce Cowardice so they actually resolve, cutting Allay should be possible too, and then we have a free slot. Worst case, the startup demands too much to get to critical mass, and the extra card ends up bridging the gap to that.
Clones actually replace the process of entering the battlefield, which means they're limited to choosing something in the "before" state in order to determine what they look like in the "after". When there's simultaneous entry, there are no Goliaths in the "before", so Clone has nothing to choose from.
Other than that, the difference amounts to a right triangle of triggers, instead of a square missing one diagonal, which isn't actually a very noticeable difference in the grand scheme of things. Even if it's the difference between 17 digits and...let's just say 1000, there are still far more powerful mechanisms to be discovered that can render any such difference moot.
First off, Primal Surge is not simultaneous (that's the entire reason it's possible to make 53 happen with clones at all). The first one will trigger 52 times, then the second one triggers 51 times, the third triggers 50 times, and so on down to the last one, which doesn't trigger at all. You do get to put all 1378 triggers on the stack in the order of your choosing.
For best results (at least within this fairly rigid framework), the surge should play out like so:
1 Hamletback Goliath
2 Dack's Duplicate
3-6 Quicksilver Gargantuan (copying the Duplicate so they're 7/7 haste, as everything after this point will be)
7-9 Dack's Duplicate (copying the Gargantuan)
10-13 Phyrexian Metamorph (so that Sculpting Steel and Copy Artifact can be brought on board)
...and so on.
Each trigger has a "source" who will get the counters, and a "cause" which is the creature whose power is being looked at. It turns out the optimal stack ordering on the triggers is simply to put the source-52 trigger on top, then the two source-51 triggers underneath it, and so on--within each source, it doesn't matter what order the causes go in. This produces a total power of 63,050,394,783,187,009 (63 quadrillion), but the largest creature which accounts for almost half of that power must necessarily be the original, which has summoning sickness. When you factor in the dethrone triggers, the attack will be good for at most 31,525,197,391,593,591.
Is there anything that can be gained by swapping out Gaea's Anthem for Mirari's Wake? Maybe we cut Meloku and use Candelabra of Tawnos to untap Goblin Burrows, assuming it can start making enough mana to be worthwhile?
Elspeth Tirel can also serve as a one-shot source of life gain after 19 uses, with no way of animating or destroying it.
Then there are certain variations of counters, like overloaded Counterflux, that couldn't be done sensibly in the interrupt days at all.
Opening hand: Chrome Mox, Mana Crypt, red card (let's say Garza Zol), Chaos Warp, Shriekhorn, Consecrated Sphinx, Words of Wisdom
Get 2R, Chaos Warp the Mana Crypt into Omniscience
Play Sphinx, then play Words of Wisdom to draw 2: Karn and Mana Crypt (52 cards for opponent)
In response to Sphinx trigger, tap Shriekhorn, mill Emrakul, and shuffle Chaos Warp and Words of Wisdom back in
Redraw Chaos Warp and Words of Wisdom
Chaos Warp the Shriekhorn into Leyline of Anticipation
Replay Mana Crypt, tap for 2 mana
Play Karn, and pay 1 mana to animate Mana Crypt into a 0/0 creature
Words of Wisdom, draw 2: Shriekhorn and Mimic Vat (51 cards for opponent)
In response to Sphinx trigger, play Shriekhorn, tap it, mill Emrakul, and shuffle Chaos Warp, Words of Wisdom, and Mana Crypt all back in
Redraw Chaos Warp and Words of Wisdom
Chaos Warp the Shriekhorn into Mana Crypt, tap for 2 mana (3 total)
Pay 1 to animate Mana Crypt, killing it
Play Mimic Vat
If Words of Wisdom always draws Shriekhorn and another card, and the Sphinx trigger always draws Chaos Warp and Words of Wisdom, this is a loop that gains access to one new card and one more colorless mana each time through.
When opponent goes to 50 cards, draw Psychic Battle, and go to 4 mana (back down to 3 to animate)
When opponent goes to 49 cards, draw Horobi
Play Horobi and Psychic Battle
Play Chaos Warp targeting Consecrated Sphinx, which triggers Psychic Battle and Horobi
Destroy Sphinx, and imprint it on Mimic Vat
Spend all 3 mana and make a hasted copy of Sphinx
Psychic Battle changes the target to Shriekhorn, where it resolves and brings back Mana Crypt again
Tap Crypt for 2 mana, then back to 1 to animate
When opponent goes to 48 cards, draw Splinter Twin
Play Splinter Twin targeting Horobi
Horobi kills itself
Psychic Battle changes target to the Consecrated Sphinx token
Tap the token to make another token
At this point, each loop of Words of Wisdom gets to draw 3 new cards instead of 1. But the mana generation is still slow as ever, as this deck seems to have no actual way of speeding it up with only a colorless 2-mana producer that costs 1 to recycle. Land/Mana Vault starts might be able to do something to take advantage of Mirrorworks, but Mana Vault cannot be recycled so easily with Karn, so that route may actually fall flat even sooner.
It obviously doesn't make a big difference, but as far as the "just needs to match creature types and colors for imprint" slots, Lumberknot looks better in green Treefolk, and Garza Zol, Plague Queen in blue Vampire (particularly if there ends up being a way to selectively give it first strike).
Also, we seem to have lost the only land, so there's no way to play Mana Vault at all, other than using the Chrome Mox mana with is needed for color.
Random thought--could we use Karn, Silver Golem in place of March of the Machines? Instead of having to keep bouncing March back and forth to use abilities of artifacts in the face of summoning sickness, Karn means we can just let them be noncreatures by default (Mirrorworks will still trigger off them just fine in that state), and only animate the things we need to for the purpose of bringing them back with Cowardice or having them hold the Splinter Twin (note that Karn's ability specifically targets noncreature artifacts, so it can never cause a Cowardice or Horobi trigger by itself). The 1 mana per individual artifact animation sounds steep, but workable.
If that works, then since there's nothing like Beloved Chaplain or a second Starfield in the deck any more to necessitate protection tricks, and no abilities where we need to bounce Cowardice so they actually resolve, cutting Allay should be possible too, and then we have a free slot. Worst case, the startup demands too much to get to critical mass, and the extra card ends up bridging the gap to that.
Other than that, the difference amounts to a right triangle of triggers, instead of a square missing one diagonal, which isn't actually a very noticeable difference in the grand scheme of things. Even if it's the difference between 17 digits and...let's just say 1000, there are still far more powerful mechanisms to be discovered that can render any such difference moot.
For best results (at least within this fairly rigid framework), the surge should play out like so:
1 Hamletback Goliath
2 Dack's Duplicate
3-6 Quicksilver Gargantuan (copying the Duplicate so they're 7/7 haste, as everything after this point will be)
7-9 Dack's Duplicate (copying the Gargantuan)
10-13 Phyrexian Metamorph (so that Sculpting Steel and Copy Artifact can be brought on board)
...and so on.
Each trigger has a "source" who will get the counters, and a "cause" which is the creature whose power is being looked at. It turns out the optimal stack ordering on the triggers is simply to put the source-52 trigger on top, then the two source-51 triggers underneath it, and so on--within each source, it doesn't matter what order the causes go in. This produces a total power of 63,050,394,783,187,009 (63 quadrillion), but the largest creature which accounts for almost half of that power must necessarily be the original, which has summoning sickness. When you factor in the dethrone triggers, the attack will be good for at most 31,525,197,391,593,591.