Analyzing Payment Based Question and Answering Service

Steve T. K. Jan, Chun Wang, Qing Zhang, Gang Wang

Abstract

This paper presents an empirical analysis of "supplier-driven" paid Q&A platforms, specifically Fenda and Whale. Unlike traditional altruistic Community Question Answering (CQA) sites, these platforms allow experts to set their own prices and introduce a unique "eavesdropping" revenue-sharing model. Our study reveals that while monetary incentives significantly reduce response times, they also create a highly skewed economy where a tiny fraction of experts captures the vast majority of revenue, and askers transition into speculative investors.

Introduction

Traditional CQA platforms like Quora and StackOverflow rely on voluntary contributions. As these communities scale, expert burnout often leads to increased latency and unanswered queries. This research investigates whether direct monetary incentives can solve these engagement issues. We focus on a "supplier-driven" model where answerers set their own prices and third-party "listeners" pay to access content, creating a complex micro-economy.

The Paid Q&A Ecosystem

The core innovation of platforms like Fenda is the "eavesdropping" mechanism. When an asker pays an expert for an answer, the resulting content is initially private. However, other users can pay a small fixed fee (e.g., $0.14) to "listen" to the answer. This fee is then split between the expert and the original asker, potentially allowing the asker to profit from a high-quality or popular question.

Figure 1: The action and revenue flow in the Fenda platform, highlighting the unique eavesdropping mechanism.

Methodology and Data

The study analyzed data crawled from the mobile APIs of Fenda (China) and Whale (USA) between 2016 and 2017. The Fenda dataset includes over 212,000 QA pairs from 88,540 users, while the Whale dataset comprises approximately 9,200 answers. Together, these transactions represent over $1 million USD in financial activity. The researchers utilized correlation analysis, hierarchical clustering of pricing strategies, and comparative benchmarks against voluntary systems like StackOverflow.

Market Dynamics and User Behavior

The introduction of a financial layer fundamentally alters user roles:

  • The Asker as Investor: By sharing in the "listening" revenue, askers are incentivized to pose questions that will appeal to a broad audience. On Fenda, 40% of questions were profitable for the asker.
  • Revenue Concentration: The market is extremely top-heavy. "Experts" (0.44% of users) generate nearly half of the platform's revenue.
  • Price Rigidity: High-profile users often maintain high prices despite declining engagement, whereas "proactive adjusters" who dynamically price their services tend to achieve better long-term reach.
  • Bounty Hunting: A new class of users emerged who target low-priced experts with sensationalist questions to harvest listening fees.

Key Results

The financial model proved highly effective at reducing latency, with Fenda's average response time of 10.4 hours far outperforming voluntary platforms. However, this efficiency comes with significant income inequality; the top 5% of answerers earn approximately 90% of total profits. Interestingly, there was no strong correlation between higher individual question prices and faster response times, suggesting experts prioritize their own schedules or interest over marginal price increases.

Figure 6: Cumulative Distribution Functions (CDFs) showing the distribution of answers, income, price, and listeners across Fenda (FD) and Whale (WH) for both normal and expert users.

Conclusion and Future Directions

While paid Q&A platforms successfully engage experts and improve response speed, they introduce market manipulations such as "bounty hunting" and collusion. Future work should focus on automated abuse detection and exploring hybrid models that combine supplier-driven pricing with asker-driven bounties to further optimize the efficiency and fairness of CQA ecosystems.

Summary updated

BibTeX
@misc{jan2017analyzing,
  author = {Steve T. K. Jan and Chun Wang and Qing Zhang and Gang Wang},
  title = {{Analyzing Payment Based Question and Answering Service}},
  year = {2017},
  eprint = {1703.01333},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1703.01333},
}
Figure