Inside Qubic's Selfish Mining Campaign on Monero: Evidence, Tactics, and Limits

Technical Summary: Inside Qubic's Selfish Mining Campaign on Monero

Problem Statement
In August 2025, the blockchain network Qubic launched a high-profile mining campaign on Monero (XMR), publicly framing it as a "51% takeover" and a demonstration of selfish mining. Selfish mining involves withholding blocks to orphan competing blocks and waste other miners' resources, potentially undermining the incentive compatibility of Proof-of-Work (PoW) systems. While theoretically studied, large-scale selfish mining deployments on live networks are rarely documented, particularly in privacy-preserving networks like Monero where pool attribution and private release decisions are inherently difficult to observe. The central problem addressed is determining the operational reality of Qubic's campaign: Did Qubic sustain majority control? Did it execute a profit-maximizing selfish mining strategy? And what was the actual economic impact on the Monero network?

Methodology
The authors employed a multi-source attribution approach to reconstruct Qubic's activity, overcoming Monero's privacy design which obscures pool identifiers.

1. Data Collection: The study combined three distinct data sources:
   • Monero Node Measurements: A pruned full node operated from September to October 2025 to capture main-chain and locally observable orphan blocks.
   • Qubic Pool Observations: RPC API queries (5-second intervals) to the Qubic mining pool to track job notifications, block heights, and previous block hashes.
   • Community-Shared Artifacts: Datasets provided by Monero community observers (DataHoarder and Sergei Chernykh) monitoring Qubic network traffic, including orphan blocks missed by standard nodes.
2. Block Attribution: The authors validated block ownership using a combination of:
   • View Keys: Qubic disclosed view keys after weekly epochs, allowing retrospective verification of coinbase rewards.
   • Heuristic Patterns: Structural regex patterns in the extra-nonce area of coinbase transactions were used to identify candidate blocks before view keys were available.
   • Cross-Validation: Initial heuristic labels were cross-referenced with community datasets, achieving a 99.92% confirmation rate for positive labels.
3. Strategy Analysis: The study identified ten candidate periods (P1–P10) of sustained selfish mining activity using a heuristic based on orphan block counts. These periods were analyzed against two analytical models:
   • Standard Selfish Mining: The Eyal–Sirer model.
   • Conservative Strategy: A variant where the attacker fully publishes a private chain at lead 3 (rather than holding it) to avoid tie-breaking risks, motivated by observed Qubic behavior.
4. Economic Evaluation: The authors calculated Qubic's revenue share relative to an honest mining baseline, accounting for immediate block rewards and delayed difficulty-adjustment (DAA) spillovers in the gaps following active periods.

Key Contributions

• Multi-Source Attribution: The paper establishes a robust methodology for attributing blocks in a privacy-focused PoW network by synthesizing node data, pool telemetry, and community observations, validated against disclosed view keys.
• Empirical Campaign Measurement: It provides the first detailed reconstruction of a real-world selfish mining campaign, quantifying Qubic's mining share, orphaning rates, and reorganization depths.
• Strategy and Economic Evaluation: The study compares observed behavior against theoretical models, revealing that Qubic did not follow a single optimized strategy. It also introduces a method to account for delayed DAA spillovers when evaluating mining profitability.
• Operational Response Analysis: The paper documents the "arms race" between the Monero community's monitoring efforts and Qubic's countermeasures (encryption, routing changes, transaction withholding), highlighting how observability limits strategy inference.

Key Results

• No Sustained Majority: Contrary to Qubic's public claims of a "51% takeover," the study found no evidence of sustained majority control. Qubic's average mining share was approximately 23.38%, with brief hourly spikes near 50% but no stable majority in daily or weekly aggregates.
• Disruption without Profit: While Qubic did not achieve majority control, its activity caused elevated orphaning and deeper chain reorganizations. However, the campaign did not yield a reward advantage. Across the identified periods and subsequent gaps, Qubic's total block count was 4.0% lower than the target-rate honest mining baseline.
• Variable Strategy: Qubic's behavior varied between standard selfish mining (releasing at lead 1) and a more conservative approach (releasing at lead 2 or 3 to avoid ties). The observed behavior did not consistently align with a profit-maximizing strategy.
• Tie-Breaking Dynamics: In specific periods, Qubic won state 0' tie-breaking events at rates higher than its hash rate, but these local advantages were insufficient to offset aggregate losses.
• DAA Spillovers: While delayed difficulty adjustments in the gaps following active periods provided a slight reward boost (+6.3% relative to replay estimates), this was insufficient to recover the losses incurred during the active selfish mining periods.
• Community Impact: The Monero community successfully monitored and relayed reconstructed blocks, forcing Qubic to implement countermeasures (encryption, withholding transactions) that reduced the observability of its internal operations over time.

Significance and Claims
The paper claims that this incident offers a rare empirical test of selfish mining claims in a privacy-preserving environment. The authors conclude that while Qubic's campaign successfully disrupted Monero (increasing orphaning and reorganization depth), it failed to execute a stable, profit-maximizing selfish mining strategy. The campaign did not result in a "Goldfinger" attack that devalued Monero in the short term (prices remained relatively stable), nor did it generate excess mining rewards for Qubic.

The significance lies in demonstrating that real-world mining campaigns are shaped by ecosystem responses. The interaction between community monitoring and attacker countermeasures creates a dynamic environment where the "intended" strategy may be obscured, and where disruptive incentives may exist beyond direct mining rewards (e.g., promotional value for Qubic's native token), even if the mining operation itself is economically suboptimal. The study underscores the limitations of static models in capturing the complexity of live network interactions and the importance of operational data in validating blockchain security claims.