Another useless Telegram Bot.
This project is an implementation of Conway's Game of Life using a Telegram bot. The Game of Life is a cellular automaton devised by the British mathematician John Horton Conway in 1970. It is a zero-player game, meaning that its evolution is determined by its initial state, requiring no further input.
TL;DR;
Every message sent to the group can start the process to create a new ecosystem or evolve the current ecosystem. The ecosystem can die because of the normal algorithm or custom variations. An organism can die because of the normal algorithm or custom variations. Every new ecosystem, evolution or die (of the entire ecosystem, not organism) will be notified to the group. End.
In this implementation, every message sent to the group where the bot is added can start the evolution process based on some probabilities.
When a message is sent to the group, different actions come into play:
NOTE: all the probabilities can be set on .env file.
If there are no ecosystem alive, the probability of a new ecosystem being born is 20% by default. This offers some uncertainty in not knowing when a new ecosystem will be created.
By default, the size is 5x5, and up to 3 organisms of every group (Flora and Fauna) can be born.
When there is a living ecosystem, the probability of new evolution depends on the number of the messages received after new ecosystem born. By default, every message increase 1% this probability, so, at least every 100 messages the ecosystem will evolve.
When a new evolution occurs, there are again some actions:
- By default, there is a 2% that the entire ecosystem die (epidemic, meteorite, natural disaster...)
- Every evolution can increase this probability, in order to add some "oldness" or "degradation". By default, every evolution increases 0,1%.
When an ecosystem evolve, the basic algorithm for "Game of Life" starts, but with some additions.
In this implementation, there are not only "organisms" in general; there are 2 main groups: Flora and Fauna. And there are a "free" interpretations about the "Trophic levels". In addition, every level has a "survival" value (1 to 4), that represents the probability to survive in the entire ecosystem. For example, a flower is more fragile than a wolf.
Flora
The flora organisms are all the organisms related, well, with flora: plants, flowers, trees, mushrooms... This organism is categorized as "Trophic Level 0", and the survival level is 1.
Fauna
In this group there are basically animals, and has different subgroups (Trophic Levels):
Level1: basically insects like bees, flies, ants, bugs... The survival level is 1.
Level2: small animals like rabbits, squirrels, small birds, goats... These animals are generally herbivorous. The survival level is 2.
Level3: bigger animals, like pigs, cows, monkeys, giraffes, elephants... These animals are generally herbivorous or omnivores, but not as dangerous as predators. The survival level is 3.
Level4: basically, big predators and mostly carnivores, like lions, tigers, sneaks, wolfs... The survival level is 4.
With this, when the basic algorithm determine that an organism keep alive, the following actions starts:
-
The neighbors are relevant. In resume, if neighbors are more dangerous than the organism, the organism can be devoured, and the probabilities to survive are as follow (home made calculations):
- Calculate the average "survival_value" for all the neighbors, but take in account only those values equal or higher than the organism. For example, if there are 3 neighbors with values 1, 3 and 4, and the "survival_value" for the organism is 2, only 3 and 4 will be used, but this value will be divided by 3 (the other organisms can be devoured instead of you).
- Then, calculate the percentage between the organism and this average. In this example, 2 vs 2,3 ~ 0,86. So, the organisms has about 86% to survive (or about 14% to be devoured).
-
If the organism is not devoured, there is a default probability of 5% that the organism die suddenly (ill, accident...). This value can be modified with every evolution (by default, increases 0.1%).
-
If the organism is not devoured or die suddenly, stays alive.
The data for all the ecosystem management is stored in SQLite. There are, for now, 2 tables: users and ecosystems:
-
Users:
- ID: Telegram user ID
- Ecosystems created: number of ecosystems born thanks to a message from him/her
- Ecosystems evolved: number of ecosystems evolved thanks to a message from him/her
- Ecosystems killed: number of ecosystems died thanks to a message from him/her
-
Ecosystems:
- ID: autogenerated ID
- Ecosystem: the current ecosystem, to be able to recover it in case the bot stops. It also serves to know the state in which the ecosystem was when it died.
- Evolutions: records the total number of evolutions (used in some probabilities)
- Messages: records the number of messages received since the last evolution of the ecosystem (used in some probabilities)
- Total messages: records the total messages that have been received since the creation of the ecosystem
- Born date: the date on which the ecosystem was born
- Extinction date: date on which the ecosystem became extinct
- Creator: User whose message gave life to the ecosystem
- Killer: User whose message extinguished the ecosystem
The idea is that all this data can be exported, and extract some statistics (for fun).
GenAI? What can the GenAI do in an simple algorithm to create end destroy ecosystems?
By default, when some events occurs, a message is sent to the group. For example, "New ecosystem is being born...", "The ecosystem has died" or "Ecosystem evolving...".
Well, the idea here is to generate this messages using GenAI. For example, instead of "New ecosystem is being born...", is more interesting "In a newly planted forest, animals and plants begin to adapt and relate, forming a new ecosystem."
This feature is in beta, so, you know. And for now, there are options to user OpenAI or AzureOpenAI services.
The prompts and pre-prompts are defined on utils/prompts
The recommended use is using Docker:
- Clone the project
- Rename .env.template file to .env
- Fill the required parameters on .env, and change the probabilities if you want.
- Run docker compose with
docker compose up -d - Wait for your ecosystem to be born, and enjoy.
If you want to run directly with Python:
- Clone the project
- Rename .env.template file to .env
- Fill the required parameters on .env, and change the probabilities if you want.
- Run
pip install -r requirements.txt - Run
python app.py - Wait for your ecosystem to be born, and enjoy.
The probabilities defined may seem chosen randomly, and in some ways they are, but not entirely. Here are some links: