Goblin Cave dug and dug.

Several of its material synthesis skills ranked up, unlocking a handful of further materials — voidglass and lumencrystal, as dark- and light-aligned crystalline materials; orichalcum, which would probably have some useful applications along with gold and mithril for mana machinery; serpent obsidian, which was a mana-sucking variation of voidstone that may or may not be useful; and so on — and it added a few more levels to its primary digging skills — [Material Absorption], [Tunneling], and the like — in the mean time. It leveled up, solely from the pittance of experience it got for digging.

While it was building out a network of tunnels for floor 27 — accessible from an offshoot of its library dump — it reflected on its mana pipes, which over what was now several years of operation had done nothing but be a constant annoyance. It had made a few slight alterations, varying their flow rate and decoherences, but there was nothing. But perhaps, it thought, it was going about things with the wrong idea. It used mana threads to spawn things, and the description for the [Flame Wisp] said they were naturally spawned by disorganized mana. So perhaps there was some structure or formulation of mana that would lead to mana threads naturally occurring. Its mana pipes shattered the mana flow into thousands of short, disconnected threads, so perhaps what it would want to make was something that somehow coaxed out a single unbroken mana loop and twisted it into the correct pattern to spawn something. Or perhaps just even a pattern that moved itself around; given enough time it would end up tangling into something.

Consequently, its expansion of floor 27 ended up being mostly different kinds of mana resonators. It was effortless for it to manipulate mana, but replicating the process with only brute materials— it seemed beyond possibility. This mesh of serpent obsidian and mithril wafted mana, that serrated tube compressed it, and on and on it built a nightmare catacomb of twisting tunnels, all perturbing the mana flow in one way or another. The closest it got to success was a particular kind of cylinder with a notched spiral of mithril and serpent obsidian, which vaguely replicated the pinching, twisting pressures Goblin Cave exerted to control its mana flow. But it was a reedy, unstable thread, and the forces only balanced in perfectly straight lines; if the tube curved the mana thread sheared itself apart. Goblin Cave couldn't imagine the circumstances that would've lead to mana threads naturally forming, if all of this machinery couldn't manage a single thread.

Everything it'd done, and Goblin Cave still couldn't figure out how to spawn a single [Flame Wisp].

Floor 27 ended up being particularly enjoyable upon reflection: as an experimental testbed it was utterly unfit for traversal by adventurers. Inch-thick pipes twisted in helixes through cubes of solid gold. Sawtooth waves of voidglass the size of houses formed interference patterns. Its warrens on its other floors were, nominally, there to be delved at some point, even if they were intended to be disorienting and confusing. Floor 27, by being shaped by the intricate physics of mana dynamics, had absolutely no referent an adventurer could grasp upon.

Consequently, floor 26 ended up being more experimentation. It sculpted out a massive torus, ringing the entire floor, and set up resonating plates in bands. It hooked them up to a simple control node, setting up a timing circuit: at one interval, the mana pulses interfered destructively, canceling themselves out into a still void; at others, they synchronized, building up stronger and stronger until Goblin Cave had to bring it to a halt for fear it would end up exploding the entire floor. Pressure thudded through the rock, rattling the nearby caves — full of vacant idiot hobgoblins, with the closest adventuring party three floors down.

One other aspect of mana it was investigating was keeping something stable outside its dungeon. Inside its dungeon was a poor place to research that, but it would have to make do. It tried constructing various seals and gateways, attempting to slow or stop its expansion of mana into empty space, and when it had managed that — a certain twisted ring shape, an elaboration on its mithril coils around its true entrance — it had a bizarre foreign space within it, mana-dead, that had its dungeon instincts screaming out to fill it. Then... it tried to shape mana to push it out into the space and rebound back, or to create a stable sphere of dungeon mana within it, disconnected from its main bulk. It permuted through various seal structures, filling most of its expanded floor 24 with dead-end mana voids, and failed over and over again to meaningfully expand into them. What it wanted...

Presumably, if it had an intelligent spawn that could exit into the world and return back with information, it would need something that kept the mana around it, or within it, still dungeon-aligned. Goblin Cave didn't know if such a thing was even possible. What it wanted was... a secondary core, in essence? Something that could maintain a spawn's mana structure even completely disconnected from its own mana pathways. It shaped various gems, creating control nodes in manacrystal and trying to impart into them some semblance of homeostasis, and then flung them over and over into various mana voids before breaching the seal and flowing in to check the results. It was difficult to tell, not the least because the act of checking contaminated all its results, but it thought it had gotten a certain pattern that twisted and pulled the mana around it that managed to maintain its coherency for, perhaps, six seconds after exiting its domain.

To determine that, it made linked corridors of mana locks, with a void in between. Then it shot its construction from one to the other: varying speeds across varying distances. If it managed to regain contact with the mana infused into the gem on the other side, then it had survived the time between. If it was just inert rock, or shattered fragments, it had not.

It would need better tools, and further experimentation, and presumably an intelligent creature capable of understanding commands, to practice further. It would also be profoundly dangerous. It would be one thing to simply have a spawn die in its task as an assistant. But experiments with its mana boundary would inevitably involve failure, and that failure would mean its spawn would die outside of its zone of control. Whatever happened to its soul then, Goblin Cave doubted it could recover it.

Goblin Cave hadn't grown particularly attached to its goblins, as such, but it had grown attached to their souls. Not simply in a mechanical sense, although it was certainly that as well; they were scarce and valuable. But insofar as any of its spawns were... beings, with desires and drives and wants; beings with ambition, it was not the animating shell that had those ambitions, precisely, but their mass of soul stuff. Dying was simple; it could capture the soul and reincarnate it. Allowing one to vanish into nothingness would be a profound neglect on its part, and Goblin Cave wasn't in any rush to hastily experiment in ways that could lead to that. Even the barest sliver of often-forked soul, the single unit that was required to spawn a tier-0 creature, had within it decades and decades of lifetimes; the loss would be immense.

That being said, the very nature of pushing outside its boundaries, and the total lack of knowledge it had of the outside world, meant that it was likely inevitable. But that didn't mean Goblin Cave had to rush into these risks unthinking.

It was strange to take in the full scope of its dungeon: the simple simulacra cave of its original goblin warrens, digging down and down to floor 50, but now it was increasingly surrounded by a bizarre lacework of esoteric materials, laid out in intricate furls. Its digging had only gotten more elaborate as it had dug back up, and by the time it would close in on its initial floors — well. It would have dug out one hundred and one floors, by its reckoning if not by the system's, and it was certain the contrast between the two styles could not be more stark.

Its nerves began to act up as it finished up the mana lacework around its 14th floor (a mess of interconnected caves that had seemed expansive when it had first dug them out, but now seemed remarkably quaint — a mess of dead-end chambers, one of which was home to its first [Goblin Warlord] as the empowered floor boss). It had noticed the shape of the floor nearly tiled itself, and so it had carved out fifteen duplicate floors in various rotations, connected through hazy corridors, each with slight variations in the layout. Which was an amusing joke, but... 14 was close to the surface. Already it was digging through a vastly different strata of rock than before: thinner, older, richer in silica than the fresher granite from the depths. It had been reigning in its expansions, keeping them shrinking in size at roughly the same rate its original floorplans had shrunk, and for the most part — floor 14 notwithstanding — it had settled on a roughly eightfold expansion in size for each floor.

Soon, though... well, soon it would finish, and then it would have to consider what to do next. Its original thought had been thoroughly revise its upper floors, melding its new functional, mechanical style with its old cave style. For its lower levels, it had thought to seal all its old floor connections and reroute: force the pathway down to its core to meander not just through its original floors but through the mazy, nightmarish pathways of its expansions. The problem, of course, was that its new creatures were utterly incapable of defending its lower depths.

There was a system submenu that appeared to suggest that some dungeons were trap-focused, with few or no spawns at all. Goblin Cave wasn't sure what to think of that. Deepmine Delve, not too far up the mountain chain, was ostensibly trap-focused, but Goblin Cave wasn't sure of the mechanics of that: what was the trap-to-monster ratio? What kind of traps? Did it have a specialized skillset that made traps more practical? But it was certainly true that in some respect it was hoping it would be able to protect its core from whatever consequences would come by dint of sheer size, which... may or may not have been true. Certainly it was harder for adventurers to tunnel through dungeon-reinforced stone, but it was by no means impossible, and for all it had dug its core chamber was still there within the entrance to floor 50, not fundamentally any more protected than it had been at the beginning of all this.

Goblin Cave enjoyed control. It liked being in control of its surroundings. It liked knowing what it was doing and how it was doing it. But in a world of other beings, there was always so much up to chance. In five minutes some level 5000 titan could stroll through its entrance and shatter apart the mountain that housed it, cutting straight to its core in a single blow; there was no evidence that something like that couldn't happen. The fear had always been that by visibly showing its change, it would attract attention it couldn't handle. But that possibility was always there; it was only a matter of how it chose to prepare for it. And looking at its complex manifolds of space, woven together — well, it had certainly prepared for something.