feat: form committees and find majority results

lib/committee.js

@@ -0,0 +1,200 @@
+import assert from 'node:assert'
+import createDebug from 'debug'
+import { getTaskId } from './retrieval-stats.js'
+
+/** @import {Measurement} from './preprocess.js' */
+/** @import {RetrievalResult, CommitteeCheckError} from './typings.js' */
+
+const debug = createDebug('spark:committee')
+
+/** @typedef {Map<string, Committee>} TaskIdToCommitteeMap */
+
+export class Committee {
+  /** @type {Measurement[]} */
+  #measurements
+
+  /**
+   * @param {Pick<Measurement, 'cid' | 'minerId'>} retrievalTask
+   */
+  constructor ({ cid, minerId }) {
+    this.retrievalTask = { minerId, cid }
+
+    this.#measurements = []
+
+    /** @type {string  | CommitteeCheckError} */
+    this.indexerResult = 'MAJORITY_NOT_FOUND'
+
+    /** @type {RetrievalResult} */
+    this.retrievalResult = 'MAJORITY_NOT_FOUND'
+  }
+
+  get size () {
+    return this.#measurements.length
+  }
+
+  get measurements () {
+    const ret = [...this.#measurements]
+    Object.freeze(ret)
+    return ret
+  }
+
+  /**
+   * @param {Measurement} m
+   */
+  addMeasurement (m) {
+    assert.strictEqual(m.cid, this.retrievalTask.cid, 'cid must match')
+    assert.strictEqual(m.minerId, this.retrievalTask.minerId, 'minerId must match')
+    assert.strictEqual(m.fraudAssessment, 'OK', 'only accepted measurements can be added')
+    this.#measurements.push(m)
+  }
+
+  /**
+   * @param {object} args
+   * @param {number} args.requiredCommitteeSize
+   * @returns
+   */
+  evaluate ({ requiredCommitteeSize }) {
+    debug(
+      'Evaluating task %o with a committee of %s measurements',
+      this.retrievalTask,
+      this.#measurements.length
+    )
+
+    if (this.#measurements.length < requiredCommitteeSize) {
+      debug('→ committee is too small (size=%s required=%s); retrievalResult=COMMITTEE_TOO_SMALL',
+        this.#measurements.length,
+        requiredCommitteeSize
+      )
+      this.indexerResult = 'COMMITTEE_TOO_SMALL'
+      this.retrievalResult = 'COMMITTEE_TOO_SMALL'
+      this.#measurements.forEach(m => { m.fraudAssessment = 'COMMITTEE_TOO_SMALL' })
+      return
+    }
+
+    debug('- searching for majority in indexer results')
+    /** @type {(keyof Measurement)[]} */
+    const indexerResultProps = [
+      'providerId',
+      'indexerResult',
+      'provider_address',
+      'protocol'
+    ]
+    const indexerResultMajority = this.#findMajority(indexerResultProps)
+    if (indexerResultMajority) {
+      this.indexerResult = indexerResultMajority.majorityValue.indexerResult
+    }
+
+    debug('- searching for majority in retrieval results')
+    /** @type {(keyof Measurement)[]} */
+    const retrievalResultProps = [
+      ...indexerResultProps,
+      'retrievalResult'
+      // TODO
+      // - status_code,
+      // - timeout,
+      // - car size
+      // - car checksum
+      // - car too large
+    ]
+
+    const retrievalResultMajority = this.#findMajority(retrievalResultProps)
+    if (retrievalResultMajority) {
+      this.retrievalResult = retrievalResultMajority.majorityValue.retrievalResult
+      retrievalResultMajority.minorityMeasurements.forEach(m => {
+        m.fraudAssessment = 'MINORITY_RESULT'
+      })
+    } else {
+      this.retrievalResult = 'MAJORITY_NOT_FOUND'
+      this.#measurements.forEach(m => { m.fraudAssessment = 'MAJORITY_NOT_FOUND' })
+    }
+  }
+
+  /**
+   * @param {(keyof Measurement)[]} measurementFields
+   */
+  #findMajority (measurementFields) {
+    /** @param {Measurement} m */
+    const getResult = m => pick(m, ...measurementFields)
+
+    /** @type {Map<string, Measurement[]>} */
+    const resultGroups = new Map()
+
+    // 1. Group measurements using the result as the grouping key
+    for (const m of this.#measurements) {
+      const key = JSON.stringify(getResult(m))
+      let list = resultGroups.get(key)
+      if (!list) {
+        list = []
+        resultGroups.set(key, list)
+      }
+      list.push(m)
+    }
+
+    if (debug.enabled) {
+      debug('- results found:')
+      for (const k of resultGroups.keys()) {
+        debug('  %o', JSON.parse(k))
+      }
+    }
+
+    // 2. Sort the measurement groups by their size
+    const keys = Array.from(resultGroups.keys())
+    keys.sort(
+      (a, b) => (resultGroups.get(a)?.length ?? 0) - (resultGroups.get(b)?.length ?? 0)
+    )
+    const measurementGroups = keys.map(k => resultGroups.get(k))
+
+    // 3. Find the majority
+    const majorityMeasurements = measurementGroups.pop()
+    const majoritySize = majorityMeasurements.length
+    const majorityValue = getResult(majorityMeasurements[0])
+
+    debug('- majority=%s committee-size=%s value=%o', majoritySize, this.size, majorityValue)
+    if (majoritySize <= this.size / 2) {
+      debug('→ majority is not absolute; result=MAJORITY_NOT_FOUND')
+      return undefined
+    } else {
+      debug('→ majority agrees on result=%o', majorityValue)
+      return {
+        majorityValue,
+        majorityMeasurements,
+        minorityMeasurements: measurementGroups.flat()
+      }
+    }
+  }
+}
+
+/**
+ * @template T
+ * @template {keyof T} K
+ * @param {T} obj
+ * @param {K[]} keys
+ * @returns {Pick<T, K>}
+ */
+function pick (obj, ...keys) {
+  /** @type {any} */
+  const ret = {}
+  keys.forEach(key => {
+    ret[key] = obj[key]
+  })
+  return ret
+}
+
+/**
+ * @param {Iterable<Measurement>} measurements
+ * @returns {TaskIdToCommitteeMap}
+ */
+export function groupMeasurementsToCommittees (measurements) {
+  /** @type {TaskIdToCommitteeMap} */
+  const committeesMap = new Map()
+  for (const m of measurements) {
+    const key = getTaskId(m)
+    let c = committeesMap.get(key)
+    if (!c) {
+      c = new Committee(m)
+      committeesMap.set(key, c)
+    }
+    c.addMeasurement(m)
+  }
+  return committeesMap
+}

lib/evaluate.js

@@ -6,13 +6,15 @@ import { getRandomnessForSparkRound } from './drand-client.js'
 import { updatePublicStats } from './public-stats.js'
 import { buildRetrievalStats, getTaskId, recordCommitteeSizes } from './retrieval-stats.js'
 import { getTasksAllowedForStations } from './tasker.js'
+import { groupMeasurementsToCommittees } from './committee.js'
 
-/** @import {Measurement} from '../lib/preprocess.js' */
+/** @import {Measurement} from './preprocess.js' */
 
 const debug = createDebug('spark:evaluate')
 
 export const MAX_SCORE = 1_000_000_000_000_000n
 export const MAX_SET_SCORES_PARTICIPANTS = 700
+export const REQUIRED_COMMITTEE_SIZE = 30
 
 export class SetScoresBucket {
   constructor () {
@@ -47,6 +49,7 @@ export const createSetScoresBuckets = participants => {
  * @param {object} args
  * @param {import('./round.js').RoundData} args.round
  * @param {bigint} args.roundIndex
+ * @param {number} [args.requiredCommitteeSize]
  * @param {any} args.ieContractWithSigner
  * @param {import('./spark-api.js').fetchRoundDetails} args.fetchRoundDetails,
  * @param {import('./typings.js').RecordTelemetryFn} args.recordTelemetry
@@ -56,12 +59,15 @@ export const createSetScoresBuckets = participants => {
 export const evaluate = async ({
   round,
   roundIndex,
+  requiredCommitteeSize,
   ieContractWithSigner,
   fetchRoundDetails,
   recordTelemetry,
   createPgClient,
   logger
 }) => {
+  requiredCommitteeSize ??= REQUIRED_COMMITTEE_SIZE
+
   // Get measurements
   /** @type {Measurement[]} */
   const measurements = round.measurements || []
@@ -74,7 +80,13 @@ export const evaluate = async ({
 
   const started = Date.now()
 
-  await runFraudDetection(roundIndex, measurements, sparkRoundDetails, logger)
+  const { committees } = await runFraudDetection({
+    roundIndex,
+    measurements,
+    sparkRoundDetails,
+    requiredCommitteeSize,
+    logger
+  })
   const honestMeasurements = measurements.filter(m => m.fraudAssessment === 'OK')
 
   // Calculate reward shares
@@ -216,7 +228,13 @@ export const evaluate = async ({
   try {
     recordTelemetry('committees', (point) => {
       point.intField('round_index', roundIndex)
-      recordCommitteeSizes(honestMeasurements, point)
+      point.intField('committees_all', committees.length)
+      point.intField('committees_too_small',
+        committees
+          .filter(c => c.retrievalResult === 'COMMITTEE_TOO_SMALL')
+          .length
+      )
+      recordCommitteeSizes(committees, point)
     })
   } catch (err) {
     console.error('Cannot record committees.', err)
@@ -238,12 +256,20 @@ export const evaluate = async ({
 }
 
 /**
- * @param {bigint} roundIndex
- * @param {Measurement[]} measurements
- * @param {import('./typings.js').RoundDetails} sparkRoundDetails
- * @param {Pick<Console, 'log' | 'error'>} logger
+ * @param {object} args
+ * @param {bigint} args.roundIndex
+ * @param {Measurement[]} args.measurements
+ * @param {import('./typings.js').RoundDetails} args.sparkRoundDetails
+ * @param {number} args.requiredCommitteeSize
+ * @param {Pick<Console, 'log' | 'error'>} args.logger
  */
-export const runFraudDetection = async (roundIndex, measurements, sparkRoundDetails, logger) => {
+export const runFraudDetection = async ({
+  roundIndex,
+  measurements,
+  sparkRoundDetails,
+  requiredCommitteeSize,
+  logger
+}) => {
   const randomness = await getRandomnessForSparkRound(Number(sparkRoundDetails.startEpoch))
 
   const taskBuildingStarted = Date.now()
@@ -299,7 +325,7 @@ export const runFraudDetection = async (roundIndex, measurements, sparkRoundDeta
   }
 
   //
-  // 2. Reward only one participant in each inet group
+  // 2. Accept only maxTasksPerNode measurements from each inet group
   //
   /** @type {Map<string, Measurement[]>} */
   const inetGroups = new Map()
@@ -368,6 +394,22 @@ export const runFraudDetection = async (roundIndex, measurements, sparkRoundDeta
     }
   }
 
+  //
+  // 3. Group measurements to per-task committees and find majority result
+  //
+  const iterateHonestMeasurements = function * () {
+    for (const m of measurements) {
+      if (m.fraudAssessment !== 'OK') continue
+      yield m
+    }
+  }
+
+  const committees = groupMeasurementsToCommittees(iterateHonestMeasurements())
+
+  for (const c of committees.values()) {
+    c.evaluate({ requiredCommitteeSize })
+  }
+
   if (debug.enabled) {
     for (const m of measurements) {
       // Print round & participant address & CID together to simplify lookup when debugging
@@ -381,6 +423,8 @@ export const runFraudDetection = async (roundIndex, measurements, sparkRoundDeta
         m)
     }
   }
+
+  return { committees: Array.from(committees.values()) }
 }
 
 /**

lib/preprocess.js

@@ -12,7 +12,7 @@ const debug = createDebug('spark:preprocess')
 export class Measurement {
   /**
    * @param {Partial<import('./typings.js').RawMeasurement>} m
-   * @param {(string) => string} pointerize
+   * @param {<T extends string>(str: T) => T} pointerize
    */
   constructor (m, pointerize = (v) => v) {
     this.participantAddress = pointerize(parseParticipantAddress(m.participant_address))
@@ -208,6 +208,9 @@ export const parseMeasurements = str => {
   return ret
 }
 
+/**
+ * @param {string} filWalletAddress
+ */
 export const parseParticipantAddress = filWalletAddress => {
   // ETH addresses don't need any conversion
   if (filWalletAddress.startsWith('0x')) {

lib/public-stats.js

@@ -60,6 +60,8 @@ const updateRetrievalStats = async (pgClient, minerId, { total, successful }) =>
 }
 
 const updateIndexerQueryStats = async (pgClient, honestMeasurements) => {
+  // FIXME(bajtos) use committees to build these stats
+
   /** @type {Set<string>} */
   const dealsWithHttpAdvertisement = new Set()
   /** @type {Set<string>} */
@@ -93,6 +95,8 @@ const updateIndexerQueryStats = async (pgClient, honestMeasurements) => {
  * @param {import('./preprocess.js').Measurement[]} acceptedMeasurements
  */
 const updateDailyDealsStats = async (pgClient, acceptedMeasurements) => {
+  // FIXME(bajtos) use committees to build these stats
+
   /** @type {Set<string>} */
   const dealsAll = new Set()
   /** @type {Set<string>} */