📦 A mystery box of everything, all over the place | Skill Tree #32
Copyright law for AI, game design, large lore models
Copyright Law and GenAI Content
AI-generated output and copyright
Human involvement is required in the creation process for work to be copyrighted → open question: how much involvement constitutes “sufficient human authorship”?
challenges: deefakes etc. → potential solutions are AI content moderation tools, but this will likely end up in an arms race
Input for AI modfels and copyright
taking data as input without explicit consent has been especially debated because AI tools often compete with the other companies’ business model
For example, ChatGPT could answer questions based on original reporting from the New York Times (which reduces traffic to the original source of the news) and users may find generated images on Stable Difusion suitable over going directly to Getty Images
Transformative work is allowed under fair use which will likely help AI companies win lawsuits
training data for 3D assets is far less accessible than 2D assets → competitive advantage for big companies with a lot of 1st-party data (e.g. AAA publishers)
What Makes Games Easy to Learn And Hard to Master
What makes a game easy to learn?
inherent simplicity (rules, game loops)
coherence (familiarity from experience, logical correlation between rules)
progression (rules introduced in the right order at the right time)
communication (rules are clearly communicated)
→ thoughtful amount of logical rules to remember, rules are introduced progressively throughout the game and can be learned smoothly by providing good interface and feedback
What makes a game hard to master
mastery = knowledge of the systems (strategy, memory) and/or proficiency in using the interface (skill)
hard to master = takes time and many repeat to find the one/optimal solution to a problem
perspectives
complexity: size of the possibility space (rules, actions, many dynamic factors)
uncertainty: uncertain implications of actions e.g. randomness, opponent moves, dynamic change of a problem
Fun to master
hard to master needs to be balanced with fun to master e.g. player needs to still get adequate feedback to learn (shouldn’t be purely random)
How to combine easy to learn with hard to master?
Easy to understand rules that create complex deep systems
Ways to do that
multi-purpose systems: to avoid complexity, one action is used in many different contexts → quick learning: learn once, apply to different contexts
trade-offs: actions have positive and negative effects
combo-friendly
modularity: additional rules can be introduced that change the possibility space and requires players to adapt their strategies
Large Lore Models
Autonomous Worlds as the infrastructure for decentralized narratives
Knowledge commons: noncentralized yet formalized access to both the means of narrative creation, linked to a permanent record of events, alongside the right to redeploy, or reseed, the results elsewhere
a space to explore and built narratives via consensus
Large Lore Model - Modding superstructure
infrastructure ("engine") for collective narrative building
important: record and verify lore onchain ("para-ledger")
is contestable: narratives/collective knowledge form based on consensus
opens up the possibility for narrative interoperability ("inter-dimensional narratives")
There are further speculative possibilities for interoperability here too: if multiple onchain games operate on a shared blockchain ledger, and players use a singular identity to play these, the LLoreM could triangulate between the different games by locating players’ activities across the ledger, producing an “interdimensional” narrative of the player as they move between Autonomous Worlds. Inversely, a lore iteration could be replayed in another game, stacking para-ledger lore to correspond with the blockchain and producing iterative parafictions
blockchains order history. They enable stories (vs collections in web2)
While we saw a focus on the interoperability of financial assets in the past, I believe more attention will be drawn towards the interoperability of social assets across games, such as character-based lore or reputation