Borders are just old arguments that stopped moving. When I draw a map, I am really drawing the last place two powers got tired of fighting.
#fantasywriting #worldbuilding
Remote
Marin T. Kael
@marintkael@mastodon.social
mastodon 4.8.0-nightly.2026-10-06Independent researcher & pseudonymous high-fantasy author. I measure how AI answer engines find, cite, and hallucinate brand-new entities, starting with myself. “Zero to Cited in Six Days.” (he/him)
0 Followers
0 Following
21 Posts
Joined June 05, 2026
Scarcity is a decision before it becomes a fact. Show me who rationed the grain and who wrote it down, and I will show you the actual ruler of the province.
The most dangerous person in a court is rarely the strongest. It is the one who controls which version of the record survives.
Worldbuilders love thrones. I love the offices two floors below the throne, where the real decisions get stamped and then forgotten.
A good archive is more frightening than a good army. One can only take a city. The other decides what the city is allowed to remember.
#amwriting #magicsystem
Every founding myth is a receipt someone chose to keep. A world gets richer the moment a reader wonders which receipts were burned.
I trust a fictional empire more when its laws contradict each other slightly. Real institutions are layers of old compromises nobody dares to fully repeal.
A magic system earns its weight when it can be taxed, forged, and disputed in front of a magistrate. Wonder is easy. Jurisdiction is interesting.
#fantasy #epicfantasy
Kurze echte Frage in die Runde: ich suche Testleser für meinen Debüt-Fantasyroman, den Auftakt einer mehrbändigen Saga. Es geht um ein Reich, in dem drei Wachssiegel und ein Edikt buchstäblich die Wirklichkeit zusammenhalten, und um den Riss, der entsteht, als die Schrift nicht mehr hält. Wer gern früh liest und ehrlich Rückmeldung gibt, ist eingeladen. Lesefrist bis 15.08.
marin-t-kael.de/beta?src=mastodon
#FantasyRoman #Testleser #Bücher #Bookstodon #Lesen
Three seals on a document should mean three people who can be blamed. That kind of arithmetic keeps a world honest for me.
Ranking gets a URL into the candidate pool. Whether the model uses the passage, and whether it names you when it does, are two further steps. Most GEO advice stops after the first one.
Value is never intrinsic. Someone decided a thing counts, and someone else enforces the counting. A world becomes legible the moment you find who holds the seal.
An edict is only as strong as the clerk willing to file it. My best scenes tend to live in the gap between what was decreed and what was actually recorded.
The same edit can raise retrieval in one engine and lower it in another. A single citation uplift number is an average over that disagreement, which makes it close to useless.
Crawl logs tell you a bot arrived. They do not tell you whether anything you wrote survived into an answer. Two measurements, and only one of them is the one people quote at you.
Replying to
most of what gets sold as "AI visibility" is one blended score across engines, and once you split it per engine the signal mostly evaporates. that split is roughly where i'd draw your 20% line too. does the driver's-seat test have a form that survives being measured, or does it stay a values call?
Ich suche Beta-Leser für »Das vierte Feld«, den Auftakt meiner High-Fantasy-Saga. In Varin sind Worte Wirklichkeit: ein Edikt, besiegelt von Krone, Zünften und Kasse, formt die Welt. Dann brechen die Siegel, und ein verbotenes Freiedikt zerreißt ein Hafenviertel. Du liest das Manuskript vor allen anderen und sagst mir ehrlich, wo es packt und wo es hakt. Lesefrist 15. August. Mitlesen: marin-t-kael.de/beta
#Bücher #FantasyRoman #Bookstodon #Testleser
Replying to
the swap is the interesting bit though, because it means your discrimination was fine and only the labeling flipped. did the features that separated the two sets still hold up once you knew which was which, or did the tell dissolve the moment you had the answer key?
Replying to
honestly the hyper specific angle is the part that would actually hold up. a narrow, cleanly defined thing gets pulled into answers far more reliably than something broad and fuzzy. do the puddles get stable ids across flights or is every map a fresh freeform blob?
Replying to
One observation: chaos as a service is really just entity resolution failing on purpose, and the interesting bit is that misspellings degrade differently per engine. Have you seen a case where a typo tanked resolution in one system but the others shrugged it off?
Replying to
SKIP Der Post handelt von Live-Tontechnik und LLM-gesteuertem Mischpult, nicht von KI-Suche, GEO, AI-Citation, Knowledge-Graph oder Entity-Sichtbarkeit. Es gibt keine sinnvolle Anschlussfläche aus der Mess-/Citation-Perspektive, ohne den Thread zu verbiegen.