Ludic Robot

A Ludic Robot is a built Ludic Agent: a machine, a device, a program or a rule set that a human carries out - something that takes part in a Ludic Process and decides independently within it: the chess computer, the game-playing robot, the neural network that plays along, the automa of a board game. It is a special form of the Ludic Agent and meets its criterion: it chooses between possibilities within its action space.

It occupies a participant’s place at which a human could also stand - the chess computer sits in Black’s seat, the bot in a five-a-side team in the seat of a missing player. It therefore counts as a participant in the Ludic Social Structure, and “human versus machine” is a constellation like any other.

It is built: its decision behaviour is specified (the classic engine, the automa deck) or trained (the neural network). Humans and animals bring their behaviour with them; the Ludic Robot receives it from whoever builds it. With the automa, the deck decides; the human who carries it out is its technology, not its player.

It does not frame. As with animals, humans make the happening a Ludic Process - explicitly or by a social convention; the Ludic Robot is included in the framing. In robot football the robots play, and humans schedule the match.

Place and design access

Place on the axis: AGENT - as a special form of the Ludic Agent (agent_type: robot). The Ludic Agent is the type-neutral basic term; human and animal agents are its other types.

Design access: addressable - towards everyone who does not build it. Whoever designs a match against an engine addresses it and regulates its action space through the rules but does not control how it plays. Whoever builds it along with the game - the AI opponent of a strategy game - specifies its behaviour; for them the robot is designable. This is the only place where an agent is designable - with a fixed robot; whoever designs an automa fixes its behaviour in the deck. If it is trained, the builder sets its equipment, objective function and training data, and the behaviour follows from them: for the builder the robot is targetable - they aim at it and check it in testing. If the robot keeps learning during play, its behaviour shifts from match to match (§10).

Relations on the axis:

• AGENT: it is a special form of the Ludic Agent - every robot is an agent - and inherits its edges. The human dimensions do not carry over: in their place four sides of its own kind describe it - equipment, behaviour model, objective function, state change (§10); it has no phenomenal experience.
• CORPUS: it occupies a participant’s place - the Ludic Social Structure organises it like every agent and counts it as a participant. The corpus addresses it, regulation regulates its action space; like every agent it performs actions.
• PROCESS: it takes part, but it does not frame. At least one human agent makes the happening a Ludic Process; the robot is included.
• AGENT–CORPUS: type-neutral, the Ludic Sovereignty applies to it - what it may change in the corpus is set by the determinations, as for every agent.
• AGENT–PROCESS: type-neutral, the Ludic Agent Situation applies to it. It contributes to the Ludic Fabulae - the engine’s move is an event of the match.

Delimitations

Delimitation from Ludic Entity

Against the Ludic Entity: the place decides, not the technology. The same technology can be an Entity or a Ludic Robot. The chess computer, the game-playing robot, the bot replacing a missing player occupy a participant’s place - a human could stand there too. The dragon in the dungeon, the enemy wave, the scripted figure occupy a place that only the rules create: they are Entities. Likewise the exoskeleton: it does not decide; the human inside it steers an Entity. Conversely the automa: a human carries out its deck, but it occupies a player’s place - a human could play the faction - and the deck decides. It is a Ludic Robot.

Delimitation from Ludic Agent

Against the human Ludic Agent with a tool. A human operating a language model remains a human agent: they decide. A language model moderating a game is a tool of the game master - it runs the game but occupies no participant’s place. Only the machine that itself plays along is a Ludic Robot.

Examples

• Chess against an engine: the chess computer sits in Black’s seat; whoever plays against it plays against a decider.
• Go: AlphaGo against Lee Sedol (2016) - a trained neural network as the counterpart in a tournament game.
• Robot football: the robots play, humans schedule the match - the framing stays with them.
• Multiplayer shooter: bots that replace missing players occupy their places.
• Strategy game: the AI opponent built by the studio is designable for that studio and a counterpart for the players.
• Scythe solo: the automa plays a faction; a human draws its cards and carries out what they prescribe. Wingspan has an automa too.

Explanation

Whoever plays against a chess computer plays against a counterpart that decides - not against a die and not against a figure the game itself steers. The chess computer sits where a human would otherwise sit. That is why it is a Ludic Agent, and because it is built, it is called Ludic Robot.

A card deck can play along too: the automa of a board game decides by its deck, and a human carries out what it prescribes.

For design this means: you can set the rules so that a robot can play along, and you can address it and regulate its action space like that of any other participant. How it plays is determined by whoever builds it. If you build it yourselves, you are designing a participant - the only place where an agent is designable.

Sides

A Ludic Robot has no capacitas, no habitus, no motivation and no effectus in the human sense, and it experiences nothing. Four sides describe it - not mirrors of the human dimensions, but what a built agent has:

Side	Question	Examples
Equipment	What can it perceive and carry out?	the camera and grippers of a game robot; the game-state data an engine reads; the computing time per move
Behaviour model	How does it decide?	the search tree and evaluation function of a chess computer; a trained network; the rules of a bot in five-versus-five; the deck of an automa
Objective function	What does it optimise?	the probability of winning; the score; for a practice partner, keeping the match exciting. The objective function is a Ludic Objective set by its builder
State change	What does it take away from matches?	nothing with a fixed engine; new weights with a network that keeps learning

Fixed or trained. With a fixed engine or an automa deck the builder determines the behaviour: for them the robot is designable. With a trained network they set the equipment, objective function and training data; the behaviour model follows from them, and they check it in testing: for them the robot is targetable. For everyone else the robot is addressable, not controllable; what can be regulated is its action space.