Maximizing NFT Incentives: References Make You Rich

TL;DR

This work tackles the fragmentation and stagnation of NFT incentives by introducing a DAG-based reference incentive mechanism in which NFTs form a graph of references and rewards accrue across generations. The authors formalize a payoff $\mathbb{U}=\mathcal{I}-\mathcal{O}$ and prove, via Stackelberg-style analysis,that the pricing/game dynamics exhibit finality, non-convex NP-hardness, and the existence of a mixed-strategy equilibrium. Given the impracticality of exact solutions, they develop a decentralised DRL approach (policy-based, e.g., PPO) to approximate optimal utility in a dynamic NFT reference environment, and demonstrate this on an ML resource marketplace with a two-reference constraint. Empirical results show convergent rewards and a tendency toward fair distributions as the number of publishers grows, with reward patterns shaped by initial economics and candidate-set size. The work further discusses broad adoption, standards compatibility, and extensions to other resource marketplaces, highlighting a practical path to sustainable, scalable NFT incentives with real-time revenue dynamics.

Abstract

In this paper, we study how to optimize existing Non-Fungible Token (NFT) incentives. Upon exploring a large number of NFT-related standards and real-world projects, we come across an unexpected finding. That is, the current NFT incentive mechanisms, often organized in an isolated and one-time-use fashion, tend to overlook their potential for scalable organizational structures. We propose, analyze, and implement a novel reference incentive model, which is inherently structured as a Directed Acyclic Graph (DAG)-based NFT network. This model aims to maximize connections (or references) between NFTs, enabling each isolated NFT to expand its network and accumulate rewards derived from subsequent or subscribed ones. We conduct both theoretical and practical analyses of the model, demonstrating its optimal utility.

Figures (9)

• Figure 1: NFT formed network.
• Figure 2: Use Case: An ML Resource Marketplace
• Figure 3: Comparison between different settings of the number of publishers $N$ in terms of normalized reward with a setting of uniform-distributed initial NFTs, $\hat{q}=0.01$, $|\Omega_{\mathcal{R}_{i,h,\Theta}}|=10$, $\hat{d}=10$, $\hat{\mathcal{I}}=2$, $\hat{\mathcal{O}}=0.1$.
• Figure 4: Comparison between different settings of the quality distribution of initial NFTs in terms of normalized reward with a setting of $N=10$, $\hat{q}=0.01$, $|\Omega_{\mathcal{R}_{i,h,\Theta}}|=10$, $\hat{d}=10$, $\hat{\mathcal{I}}=2$, $\hat{\mathcal{O}}=0.1$.
• Figure 5: Comparison between different settings of the initial raw interest $\hat{q}$ in terms of normalized reward with a setting of $N=10$, uniform-distributed, $|\Omega_{\mathcal{R}_{i,h,\Theta}}|=10$, $\hat{d}=10$, $\hat{\mathcal{I}}=2$, $\hat{\mathcal{O}}=0.1$.

Theorems & Definitions (7)

• Definition 1
• Theorem 1
• Proof : Theorem \ref{['lem-distri']}
• Theorem 2
• Proof : Theorem \ref{['lem-convex']}
• Theorem 3
• Proof : Theorem \ref{['lemma-stack']}