IIIT Work Summary Presentation June 2024

Players and Games: A Retrospective

• This is a look back at several different research directions pursued as part of the IIIT-H Master's by Research (Exact Humanities) program requirements.

• The program in question no longer exists in its original form.

• Thesis Statement: A game can be imagined as a dynamic entity whose nature is shaped by its players.

Philosophy

Heterotopias

Prof. PRK Rao originally suggested to me the research topic of heterotopia, and offered The Heterotopia of Facebook as a starting point.

• A heterotopia is an unreal space that changes as it interacts with whatever is constructing it.

• A utopia is a special case of heterotopia.

  • built from the imagination of a moral ideal

  • brittle, sensitive to change in the generating ideal

  • contradictions inside the ideal break the utopia.

• We can imagine a heterotopia derived from a more heterodox moral paradigm, a la Isiah Berlin

  • By nature dynamic.

  • If inhabitants of an imagined society are modeled as having the power to change that society, how would we imagine the changes we allow them to make, and the consequent space they inhabit?

Philosophy

Rhetoric of Science

• McCloskey explores the idea of rhetorical work as central to economics, as a heuristic precursor to evidence-supported theory in a field that experiences evidence scarcity.

  • When many theories can fit an evidence base, it's usually best to have them all active and in use, and treat them as languages - media of the work - rather than outcomes.

  • The rhetorical thesis: Good sentences are useful sentences, which are not necessarily provably true sentences.

• A language-game Wittgenstein is the idea that language as a tool built to win a collaborative game.

  • A game is most minimally characterised not by the rules, but by the winning strategies.

  • Imagine economics as a language-game.

  • The practice thesis: Economics is better characterised as a practice than as the associated theory.

• Assertion: it is better to think about science in general as a practice, and as rhetorical.

Logic

The project here was to formally represent player-editable games - games whose structure depended on players (i.e. on player strategic action, either via their preferences or their available moves). We can conceive of this as strategic selection among alternative games - , i.e. playing a metagame. There are several ways to approach this idea. I've detailed my attempted approaches chronologically.

Logic

2016: Player-editable games

• formal representation of games as collections of rules, with rules describing rule editing. This is effectively a formalization of a Nomic-like game Suber. We quickly run into Suber's original issue with Nomic: either a game simplifies to something that can't be called editing, or it's too underspecified to be able to make positive claims about.

• motivated mechanism design - formally representing mechanism design and turning it into game moves. This proved intractable in the general case. Prior art exists in subproblems that take a similar approach: Chen & Micali, Ausubel….

Logic

2017-18 Game semantics based approaches

Game semantics: the study of alternate formal models of games, and their behaviour.

• Gluing of disparate game trees, or game tree generators, with "meta-moves", a la Ramanujam & Simon

• Variations of alternating-time temporal logic Alur… - i.e. coalitional strategies that represent narrowings of a game tree.

  • Easy to construct a dominance relation over games

  • Hard to build a notion of dynamics.

  • A single good example would have been sufficient to complete this original work.

• Dynamics driven by changes in the information available to players over time: temporal epistemic games.

  • This proved unfruitful because clunky.

Logic

2022 - Alternate models of preference.

Following Dietrich & List, representing players as preferring properties of game states - i.e. preferences between rules. We tried to treat the properties as statements about the reachability of terminally desired states.

Confluence: Preference meronomy

Two central questions can now be asked, in several different spheres of concern:

• Given a collection of agents - purposive, complex, mutually aware actors - what can we say about the world they construct in the pursuit of their goals? Can we infer form from function?

• Given a world, inhabited and constructed by agents - can we say the world as a whole wants something? Can we infer function from form?

Post mortem

What went wrong here? What could have gone better?

• Spread too thin. I had too many ideas and did not follow through on any of them enough in the time I had. - Sparse specialised guidance. I did not have enough guidance a priori, and did not reach out to suitable guides.

• Insufficient feedback. I didn't finish or share anything until well far too late.

• Inadequate general guidance. How to write a paper? How often should one request/receive feedback? What is the appropriate way to reach out to other academics? I didn't know.

Thank you.

Despite obstacles, I enjoyed my time here and the work of answering questions I cared about.

Thank you for your time and consideration.

Questions / comments / feedback?