LA Token Fee Distribution Logic for High-Frequency Microtasks
Scenario:
The DApp wants to batch submit micro-inference tasks with a latency of 50 milliseconds to the Coprocessor. Each individual task is very light, but the quantity is massive, making the traditional 'pay in full at once, generate a large proof' approach uneconomical.
⸻
1 Batch Microtask Packaging
1. The user sends an array of tasks along with the total LA fee to TaskManager.batchSubmit().
2. The contract generates a Bloom filter based on the task hash, and off-chain nodes fetch and execute in parallel.
3. Every 32 microtasks share a local Trace, aggregated using a ring-buffer; the STARK generator commits directly to the entire Trace block, no longer breaking it into sub-proofs.
⸻
2 Real-time Proportional Distribution of LA
Phase Action LA Allocation
Batch Submission Freeze Total Fee F 2% immediately burned; the rest locked
Segment Completion commitPartial(id) Unlock F/n to the execution node
Aggregation Successful finalizeBatch() An additional 5% reward for the Aggregator, automatically distributed upon success
n is the number of tasks within the batch. Thus, completing each microtask allows for immediate receipt of the corresponding LA, without needing to wait for the entire batch to finish.
⸻
3 Deviation Locking and Penalties
• If the aggregation proof fails, the unlocked amount already received is deducted proportionally from the node's collateral;
• A failure incurs a 30% burn penalty, and 70% is injected into the insurance pool for compensating affected callers;
• If there are three consecutive batch failures, the node is automatically placed on a watchlist, restricting batch order acceptance.
⸻
4 Adjustable Governance Parameters
• Distribution ratios, burn percentages, and failure penalty rates are all recorded in GovStorage;
• Proposals require staking LA and voting, with a Timelock of 36 hours;
• The contract upgrade entry can only be paused and cannot directly modify storage, protecting against hot wallet risks.
⸻
Core Benefits
1. Immediate receipt of microtask rewards reduces cash flow pressure on computing nodes;
2. Burning and penalties continuously shrink circulation, offsetting incentive releases;
3. The governance layer can dynamically adjust fees in response to on-chain congestion and GPU market prices, maintaining economic balance.
After evolving from 'single payment' to 'streaming distribution + risk buffering', LA can support millisecond-level high-frequency calls without sacrificing trustworthiness and deflationary attributes.