Reputation

Shiroi maintains an epoch-based reputation system for searchers to regulate access to low-latency infrastructure under varying network conditions.

Reputation does not influence ordering, inclusion guarantees, or auction outcomes. It is used exclusively to derive rate limits and access parameters during periods of contention and congestion.

Design Goals

The reputation system is designed to:

  • preserve deterministic auction resolution under load

  • prevent infrastructure saturation and spam

  • align infrastructure access with economic intent and execution quality

  • adapt dynamically to overall network conditions

Reputation is non-discretionary, transparent, and epoch-scoped.

Reputation Inputs

Searcher reputation is derived from three independent signals:

  1. Shiroi Tips

  2. Transaction Quality

  3. Network Load Context

All signals are evaluated over fixed epochs and combined deterministically.

Shiroi Tips

Shiroi Tips represent explicit, on-chain economic commitment from a searcher.

Tips are attributed to the searcher over the course of an epoch and contribute positively to reputation.

Reputation weight increases with:

  • total tips paid during the epoch

  • consistency of tip submission

  • sustained participation across the epoch

Tips do not:

  • guarantee inclusion

  • affect ordering

  • bypass auctions

They serve solely as a credible signal of intent used for rate-limit calibration.

Transaction Quality

Transaction quality reflects how reliably a searcher submits executable transactions.

Quality is evaluated using observed execution outcomes, including:

  • successful execution

  • deterministic failure due to auction loss or contention

  • invalid or non-executable transactions

Reputation is penalized for:

  • malformed transactions

  • systematically non-executable instructions

  • spam-like submission patterns

Complex, state-dependent, or competitive transactions are not penalized when failure is consistent with deterministic auction resolution.

Network Load Context

Reputation is evaluated relative to current network-wide load.

Rate limits are dynamically adjusted based on:

  • overall transaction volume

  • congestion levels

  • contention intensity at the BBM

  • infrastructure backpressure signals

Under low network load:

  • rate limits are relaxed

  • reputation differences have minimal impact

Under high network load:

  • rate limits become more selective

  • reputation plays a stronger role in access to low-latency paths

This ensures that scarce execution-adjacent resources are allocated proportionally during congestion.

Epoch-Based Evaluation

Reputation is computed per epoch.

At the end of each epoch:

  • Shiroi Tips are finalized

  • transaction quality metrics are aggregated

  • network load normalization is applied

  • rate-limit parameters for the next epoch are derived

Reputation does not accumulate indefinitely and cannot be permanently locked in.

Rate Limit Derivation

Rate limits are derived as a deterministic function of:

  • epoch-level Shiroi Tips

  • observed transaction quality

  • network load during the evaluation window

Higher effective reputation results in:

  • higher allowed submission rates

  • improved access to low-latency sending paths

  • reduced backpressure under congestion

Rate limits never affect:

  • auction resolution logic

  • ordering rules

  • execution determinism

Abuse Resistance

The reputation system is designed to resist:

  • burst spam during congestion

  • tip-free saturation attacks

  • griefing via non-executable transactions

Epoch-based resets prevent long-term accumulation of advantage through abusive behavior.

Neutrality Guarantees

Reputation does not:

  • grant priority ordering

  • bypass BBM-resolved auctions

  • provide private execution guarantees

  • create persistent privileges

All ordering and inclusion decisions remain deterministic and BBM-enforced.

Summary

Reputation in Shiroi is a load-adaptive access control mechanism.

It aligns low-latency infrastructure access with demonstrated economic intent, execution quality, and prevailing network conditions — while preserving deterministic, BBM-resolved auction behavior.

Last updated