| Memory & Logic |
- Based only on the “Context Window” of the LLM.
- Massive memory drift with long texts.
- The AI often forgets items, people or places.
- Frequent hallucinations of things that never happened.
|
- 100% rigid hard logic.
- The system doesn't forget anything as long as the code is error-free.
- No hallucinations possible.
|
- The MySQL database is the irrefutable truth.
- No chat history, instead structured snapshots per round.
- No memory drift and absolute object permanence.
|
| Input & Freedom |
- Completely free text input.
- The AI understands almost everything.
- Often leads to absurd or nonsensical situations.
|
- Strict parser that requires exact commands (e.g. "take key").
- Synonyms or complex sentences result in "Command not recognized".
- Often frustratingly rigid.
|
- Real Natural Language Actions: Free, verbatim writing.
- Gemini interprets the intention, Laravel validates the logic.
- Builds something interesting from any text input.
|
| Rules & God Mode |
- Sycophancy: The AI mostly says "Yes and Amen".
- Players can use “God Mode” and wish for anything.
- Quickly leads to boredom due to lack of challenge.
|
- Strict, unchangeable rules.
- If something has not been programmed, it does not exist.
- Players must find exactly the predicted solution path.
|
- No God Mode! The engine gives objections.
- If you don't have a lightsaber in the database, you don't have one.
- It is precisely these contradictions and the logical failure that keep the story exciting.
|
| Chance & Immersion |
- The LLM “estimates” probabilities.
- Often inconsistent pacing.
- Little influence on the actual mechanics in the background.
|
- Real RNG (Random Number Generator) in the code.
- Mechanics are often presented to the player in a very dry manner.
|
- The LLM is given real random values/entropy from Laravel.
- No numbers or states visible (except in special LitRPG scenarios).
- The mechanics remain invisible, everything is built in narratively and immersively.
|
