The first book/manastone trade-off was utterly anticlimactic. One of the surveyors and one of the other people in that meeting — Goblin Cave had already forgotten everything about how they introduced themselves — returned, along with two others who seemingly only came with them to carry a long cot with two crates of books on it. They dropped it off, and Goblin Cave had created a neatly-stacked pylon of rhombic dodecahedral manastones in preparation.

There was one wrinkle. "Ah, if we may make a request," said the... knight-diplomat? "The... pseudo-dungeon that extends beyond here," — here being the first chamber of the mana bellows, where the exchange was happening — "the initial survey was never completed, but they did report low-level dungeon-style mobs with mana-aligned drops. We would be interested in sending low-level teams to survey and, if possible, farm for resources and experience."

Goblin Cave had not bothered to relocate the flame wisps for this conversation, and at this point the flock had spread out into a uniformly-scattered band of wisps, slowly revolving through the bellows' flow, with one or two passing through the room every several minutes. It was a little concerned that if it said 'YES' they would just start blasting.

It was also concerned about having adventuring teams wandering around through its tunnels. There were no boss mobs, only a scattering of slowly-respawning, low-level, low-tier 'aberration' mobs. It said as much: THERE ARE ONLY A FEW MOBS NEARBY, AND THEY ARE WEAK AND SLOWLY-RESPAWNING. RESULTS OF MY EXPERIMENTATION. But... IF YOU WISH TO ATTEMPT TO FARM THEM, I RUN THE MANA FLOW ON A CYCLE. AT ITS LOW EBB, IT SHOULD NOT BE HAZARDOUS. It would also need to shuffle off the flame wisps to somewhere isolated. As much as it wanted them to level on their own, it was much more likely they'd all die immediately.

This was, perhaps, the duchy's next step towards trying to invade it. Presumably, letting low-level adventurers in was a lie to try to further analyze its capabilities, just like its response was a lie, still pretending it wasn't a dungeon core. But if it knew they were lying, and they knew it was lying, the whole situation existed nothing except as an excuse to confuse and disorient the other.

It did, ultimately, need to know how well the aberrations it had unlocked performed against adventurers, not just goblins.

It was striving with the full force of its power to hold back its mana enough not to instantly obliterate its worms and ants. By contemplating mana control, of the system, it was slowly driving it to unearth insights into the nature of reality. All killing adventurers would teach it was how to to kill adventurers. That was a very specific niche of study. It was unfortunate that it was necessary.

The pile of books was so much new information it had trouble forming any kind of coherent structure of it. Everything new was placed in a context of a dozen other new things, and each of those things had dozens of its own things, outwards and outwards until it was a confused mess. It had usually obtained a few scrolls a year, with books coming as sudden windfalls — sometimes three a season, sometimes none for three years. This delivery consisted of 60 books, ranging from the shortest — a geographical treatise of only a few pages of text interspersed with several maps it was also having difficulty reading — to the longest — an immense tome about the thirteen elemental vortices and their interplay, which seemed incredibly interesting if it could understand anything in it past the first five pages.

It put most its time towards trying to read and comprehend the books, slowly building up a mental map of... everything. Where it was located, who the political actors were, how they conceptualized mana, what they considered unsolved problems in mana dynamics and system studies. Everything.

While it was doing that, it was also preparing. Thinking of the sages who would potentially inhabit it, it tried to imagine what kind of landscape adventurers would want to exist in. Somewhere with light, presumably. And plants? Apparently nearly everything it could glimpse outside of its entrances would fall under the categorization of 'plants'. Budding off a secret door from one of its sixth-floor caves, it shaped a hallway leading to another dirt-filed cave, only this one it lined the ceiling with slabs of lumenrock, and covered the dirt floor with spawns of cavegrass. Some of what it could see outside was a similar kind of plant: green-brown and spiky. It was considering asking the adventurers if they could simply dig up some of that, too. Samples of everything that existed outside that could be easily moved within it.

The primary problem, as always, was mana flow. Cavegrass was a mob spawn, although a particularly pathetic one. (The sole skill Goblin Cave had unlocked for it was [Grasping Roots], which meant that it could theoretically catch and bind adventurers, but in practice they were so weak as to only provide a moment's distraction before being destroyed.) So it would have no trouble bathing in its mana. Outside grass was, presumably, not system-empowered, and would die from exposure to its mana.

Its mana locks were slowly becoming more advanced. It was still struggling to understand how the adventurers' conception of mana harmonics applied to its own sense of mana flow, but it had managed to create locks that had gaps that only permitted a certain... call it a 'tone' of mana through. The result was a room beyond the lock that felt muddy and unclear, but that it could still, slowly, act and command mobs into without them being immediately dissolved by low mana. And it could create a second mana lock past that, that further isolated the mana flow. Four locks deep, and worms no longer dissolved at the ambient mana pressure, and it could still — slowly and haltingly — command an invisible servant within the room. Here, it could plant [Cavegrass] and... grass side-by-side.

Meanwhile, it had also constructed a vastly simpler mana loop out past its mana bellows, and redirected the flow for a circuit to spin the bobbing flame wisps into that new, isolated loop, before sealing it off behind several secret doors as well. This would keep them safe, hopefully. They didn't ever do much aside from float in the mana flow.

The first squad of adventurers came through and immediately got lost in its mana bellows. Goblin Cave had decided to run it on an automated cycle, mostly just for its own amusement: high flow during the 'day', low flow at 'night', smoothly ebbing and flowing over time. It had posted a modulus diagram showing the current flow rate at the entrances.

The handful of mobs there appeared to be a moderate challenge for the group — certainly vastly more of a challenge than an equivalent amount of same-level goblins — but once they stepped off the primary current of the bellows, they became totally disoriented in the maze of secondary passages. They ended up stumbling back onto the main flow route as it was starting to pick up again near dawn, and became tantalizingly close to actually stumbling up into one of the other two as-of-yet unused secondary entrances, before their... handler, or minder, an adventurer with them who stayed back and didn't engage the mobs with them, informed them they were headed the wrong direction to get out. Someone with some kind of directional or mapping skill, presumably.

It had racked up several tallies in its experience corridor. By number of marks, it was looking like 'elemental construct' was in the lead (4), followed by 'aberration' and 'undead' (3 both), then 'elementals' (2), with nothing else getting any tallies. This seemed to flatly contradict the adventurers' claims that experience was based solely on category, but it also seemed to contradict its understanding that experience was a fixed value per mob, as its spawns' system panes reported. It was also possible an adventurer's 'experience sense' was, in fact, profoundly inaccurate, but they seemed to trust it at least a little more than that. So... more research was required. And it would probably be worthwhile to try to unlock more mobs in all of those categories... as well as more mobs in all the other categories, for comparisons. The process of manually-constructing spawn templates would be pretty time-intensive, and also to be honest it had not been feeling in the mood for that. Too many things had been going on for it to retreat into the calm, measured kind of consideration required to dig through possibility space.

It made several new sub-dungeons on floor 51.

This was, in truth, fairly simple: take the control node it had constructed for the one sub-dungeon, and swap out all the specifics for a different category. The sub-dungeon had only ever placed empty spawn templates, rather than handling the spawns itself, and since Goblin Cave was already doing the actual spawning in of the mobs for the sub-dungeon itself, all it would really need to do would be pick different mobs.

The main issue was that it didn't scale: it required too much attention to simply set up a dozen sub-dungeons and set them going. It picked away at that problem for a day or two, trying to set up some kind of... spawn template template, that the control nodes could copy, but it still couldn't get them to push mana through it with the correct precision to actually establish a spawn. It could get them to randomly pick between a few templates, though, even if it couldn't get them to properly spawn anything. Yet. It seemed possible. There was probably a system skill for it, in fact: somewhere in the thicket of unpurchased skills that it had passed by in favor of spawn rate upgrades. That was slightly annoying, and also moderately humorous. Next time it leveled, whenever that would be, it would try to dig through its skill tree to see if there was anything that would take it a step closer.

In the mean time, it would continue tweaking the control nodes. It wasn't actually certain about it, but it had been hoping that anything qualitatively new a system skill unlocked was theoretically something it could do directly with mana manipulation. Quantitative upgrades — 'allow 5 per depth more mobs per floor' and the like — seemed to directly mediate mana flows to make that possible, in a way that was beyond it, but much like new mob spawns, things like new trap unlocks or new materials all seemed to be things that it could have made on its own before obtaining them via system unlock, if it had known how to do it beforehand. Knowing how to do it beforehand without being told, of course, was the challenge.

All that was to say, it created three new dungeons: one elemental-themed, spawning in [Flame Wisps], [Wind Shades], and [Animated Pebbles]; one undead-themed, spawning in [Biting Skull] and [Grasping Hand]; and one, extremely subdued, plant-themed, spawning in [Cavegrass] and [Chokevine]. Plants, or at least tier 0 plants, were nearly incapable of killing anything; they seemed more like hazards, and at tier 0 the actual hazard they presented was negligible. It added [Bat] to the random selection for that one, so that there would be at least some mobs there capable of movement.

The hope was to unlock new skills for the mobs in question, which would slowly make them more of a threat — [Cavegrass] also had the potential for, say, [Bladed Leaves] — and let some of them level up, eventually. It would also, if it paid a lot of attention, let it record experience gained for its goblins as they killed the new categories of mobs, which would give it another source of raw data.

All of that would take time, however. The spawn rates of the sub-dungeons were still abysmal, and there weren't enough mana goblins around to fully delve the one sub-dungeon it had made, much less the three new ones it had instantiated.