Reliability is Blind: Collective Incentives for Decentralized Computing Marketplaces without Individual Behavior Information

TL;DR

This work tackles reliability in decentralized cloud marketplaces where individual actor behavior is not observable. It introduces Collective Incentives, a stake- and reputation-based mechanism that blindly punishes or rewards all assets involved in a task based on its outcome, and derives its operation from ruin theory to target a desired task-success rate $F_t^{\mathit{target}}$. Through Python simulations, the authors demonstrate that the mechanism can meet or exceed the target reliability while pruning failure-prone assets and preserving reliable ones, with the combined Coll-SR configuration offering the best performance. The approach offers a practical pathway to robust Web3 marketplaces under limited observability, while outlining extensions to handle asset combinations and adversarial settings.

Abstract

In decentralized cloud computing marketplaces, ensuring fair and efficient interactions among asset providers and end-users is crucial. A key concern is meeting agreed-upon service-level objectives like the service's reliability. In this decentralized context, traditional mechanisms often fail to address the complexity of task failures, due to limited available and trustworthy insights into these independent actors' individual behavior. This paper proposes a collective incentive mechanism that blindly punishes all involved parties when a task fails. Based on ruin theory, we show that Collective Incentives improve behavior in the marketplace by creating a disincentive for faults and misbehavior even when the parties at fault are unknown, in turn leading to a more robust marketplace. Simulations for small and large pools of marketplace assets show that Collective Incentives enable to meet or exceed a reliability target, i.e., the success-rate of tasks run using marketplace assets, by eventually discarding failure-prone assets while preserving reliable ones.

Figures (5)

• Figure 1: Decentralized cloud computing pipeline example.
• Figure 2: Failure model in decentralized cloud computing.
• Figure 3: Asset ruin probability depending on initial stake $S_{0}=xP$ and experienced task failure rate $F_{t}$, given $F_{t}^{\mathit{target}}=20\%$.
• Figure 4: Median (dark) and $95^{\mathit{th}}$ percentile (light) task failure rates over 100 runs for different incentive mechanisms (using per-run task failure-rates computed as moving averages with a window size of 100 tasks). $N_{t}=10^{4}$ tasks each involving 4 assets are executed per run, with $N_{a}=100$, $\alpha=0.05$ ($\mathit{mean}(F_{a})=4.8\%$), $S_{0}=10$, and $F_{t}^{target}=1\%$.
• Figure 5: Stacked bars of removed and remaining assets using "Coll-Stake" after $N_{t}=10^{4}$ tasks each involving 4 assets, averaged over 100 simulation runs, with $N_{a}=100$, $\alpha=0.1$ ($\mathit{mean}(F_{a})=9.2\%$), $S_{0}=10$, $F_{t}^{target}=5\%$.