I was in a relaxed mood of post-battle euphoria, like a sports spectator who had watched their team win a hard-fought game. Actually, it was more like a gambler, since the stakes were much more real for me, and I also had netted some sweet, sweet gains.

  I had achieved my Essence needs, two skills–Demesne and Sigilmancy–had leveled up, and I had a powerful new alchemical toy to play with.

  And yet, despite all the people who died in my demesne, I had not gained a level up. One step forward; one step standing firmly still.

  My first two acts post-heart Buffett was to absorb the healer’s heartless corpse and to ramp down the Holding’s regeneration to the minimum. The former because I hoped to pick up more useful materials (which, unfortunately, was a bust); the latter because I wanted to give the dwarves as much time as possible to come back and fuck that guys shit up. Not that the dwarves were less dangerous or were well-intentioned, but just–maybe they could get in a big fight again and a giant meteor crashes down and kills them all?

  I had time, quite a bit of time, really. So, I decided to work on my future companion first, dungeon upgrades later (when I had more mana).

  I brought up the previous system notifications when I had first tried making a companion. There had been two parts: the first had been about insufficient essence–a problem now solved–and the second had been “Sigil core not unique.”

  I wondered if the message was in some way related to the “hole” in the dwarves’ sigil arrays. Sigilmancy didn’t give me a definitive answer. Perhaps I wasn’t asking the right question.

  In my first attempt I had put my focus on creating a dragon, because of course I’m going to make a dragon. Dragons were awesome and awesomely powerful. So, now that the other condition was met, I could test again.

  But before I did any actual testing, I wanted–needed–to install a safeguard first. Software programmers often included those annoying “confirmation” dialog boxes for a reason. People were generally rather stupid and a stop-break might prevent some of them from doing something stupid. I had not been immune to this. Also, the system had not been user friendly.

  So, I would use Interface to try to make sure nothing I didn’t want went through.

  Ah, but the only way to do that was to actually do it–what a conundrum.

  I started with an easy foolproof option: something I knew would get denied. I took the Essence and willed it to be spent and demanded a sigil core of “dragon,” just like before. Again, the error message came up.

  I went through the motions again, but this time, I focused on creating a companion that was a “Dark Goblin.” The error message appeared again.

  Dwarf–the same notification came up.

  Those were the three that I knew the answers for already, but confirmation was still useful.

  I chose the Dwarf as a model again. I went through the process as slowly as I could, dragging it so I could feel the mana move, coalesce into a form–a proto sigil--and then collapse. Just as it started to collapse, I used Interface and stuck a notification screen in, trying to catch the last parts and turn it into useful information.

  I picked the last one and felt a slight sting as the mana whipped out and then back into me. This was progress, even if the message was a little weird.

  I iterated. This time, I focused on getting the proper choice dynamics. I iterated a third time and got rid of the useless prompts.

  That was much better.

  Unlawfully taken from Royal Road, this story should be reported if seen on Amazon.

  I thought about what to test next. Using some basic deductive logic, I tried with an untested archetype: goblins. As expected, it failed. That made sense, if “Dark Goblins” were a thing, then they were probably a variant of something else.

  What about the other way?--I tried “Red Dragons,” “Blue Dragons,” “Bronze Dragons,” and a bunch of others. All of them were failures.

  I really wanted to test my Interface design so I thought about what options would probably be unique. I needed to pick something really really specific, something that almost certainly only existed in Earth literature.

  I picked a more ridiculous option and pushed it through. Kender.

  Success.

  But I was not making a Kender; I wasn’t that evil.

  Kender, though, were just a bastardized version of halflings–also available. And halflings themselves were just a bastardized version of hobbits–they too were available.

  I was even more curious about what qualified for the Companion, so I decided on the next test. The system had foisted upon me the monster archetype “undead” even before I had done anything as a demesne core. It was a very common monster trope for dungeons. Yet now, I wasn’t restricted. So, I wanted to know. I started the process again, but this time put my will toward creating an undead companion.

  There were no dialogue options; this was a very different error message. Something was missing; something else was incorrect. I thought about what were the key differences between the first set and this one.

  Having a theory, I tried another option: vampire.

  That was not expected. I tried a different one: zombie. That got the fault error, as did lich, wraith, ghost, and even wendigo.

  I was now confident I had one of the parameters figured out, but there was another test I could to further prove it. One more time I went through the motions, but this time on creating a golem companion. As I expected, the fault message appeared.

  I was convinced my theory was correct: my companion had to be a living, breathing creature. It had to be a living breathing creature of a type that was “unique,” an archetype that did not already exist out there in the world (or maybe in other demesnes). Not that I wanted an undead companion anyways, but it was useful to get this kind of information. I could now survey the system to find what races that I hadn’t seen yet were in this world.

  So that is what I did. I ran the gamut of fantasy races and some of their variants that I could think of.

  After more than an hour of trying different types of creatures, I had a sufficiently large list. A couple of things were apparent to me. First, the system wasn’t necessarily saying that the exact thing I was looking for was or wasn’t available. This was consistent with my previous discoveries about how the system interacts with language. In particular, how could it be that ‘Hobbit,’ a very specific type of fantasy race only in a few books, didn't exist, but ‘Uruk Hai,’ another very specific variant of Orc in the same series of books, did? Another example was angel and demon. The system thought ‘demon’ was a living creature, but that an ‘ angel’ wasn’t. Likewise, undead were excluded because, obviously, they were not living creatures, but vampires existed, which suggested they were living creatures here.

  The system was finding the closest approximate thing to give an answer; the language I used wasn’t determinative. Even my categorizations might be wrong. Fortunately, the accuracy of my categorization or terminology was not important.

  So far, I had already found twenty-five options for my companion. I could probably find even more if I started testing for more esoteric and unorthodox creature types. But there seemed little reason to. Most of the choices I had were merely illusory; I knew I wasn’t going to pick them.

  As for the few viable, interesting options? I stared at the screen.

  This was going to take a while.