Zaptam Trust System

Deep dive into trust score calculation, tiers, badges, and penalty system.

Overview

The trust system provides a 0-100 score representing a user's reliability and safety. It influences discovery ranking and provides visual trust indicators to other users.

Trust Score: 0 ─────────────────────────────────────── 100
             │                                         │
             ▼                                         ▼
         Untrusted                              Highly Trusted
         
             │         │         │         │
             0        40        60        80       100
             │         │         │         │
         Very Low    Low     Medium     High

Score Calculation

File: src/services/trust.service.ts

Algorithm

typescript

export async function calculateTrustScore(userId: string): Promise<number> {
  const user = await prisma.user.findUnique({
    where: { id: userId },
    include: {
      reportedBy: {
        where: {
          status: { in: ['RESOLVED'] },
        },
      },
      sentInterests: {
        where: { status: 'ACCEPTED' },
      },
      receivedInterests: {
        where: { status: 'ACCEPTED' },
      },
    },
  });

  const factors = getTrustFactors(user);
  const baseScore = 50;
  
  const totalScore = Math.min(
    100,
    Math.max(
      0,
      baseScore +
        factors.verificationBonus +
        factors.ageBonus +
        factors.reportPenalty +
        factors.interactionBonus
    )
  );

  return Math.round(totalScore);
}

Formula

Trust Score = BASE + Verification + Age + Interactions - Penalties

Where:
  BASE = 50 (starting score)
  
Clamped to range [0, 100]

Trust Factors

1. Verification Bonus

Higher verification levels indicate more thoroughly vetted users.

typescript

let verificationBonus = 0;
if (user.verificationLevel === 'IDENTITY') {
  verificationBonus = 10;
} else if (user.verificationLevel === 'FULL') {
  verificationBonus = 20;
}

Level	Bonus	Description
NONE	+0	No verification
IDENTITY	+10	ID verified
FULL	+20	ID + worth/value verified

2. Account Age Bonus

Older accounts demonstrate commitment and reduce likelihood of fraudulent behavior.

typescript

const monthsOld = Math.floor(
  (Date.now() - user.createdAt.getTime()) / (30 * 24 * 60 * 60 * 1000)
);
const ageBonus = Math.min(10, monthsOld);  // +1 per month, max +10

Age	Bonus
0 months	+0
1 month	+1
2 months	+2
...	...
10+ months	+10 (max)

3. Report Penalty

Resolved reports against the user reduce trust.

typescript

const reportPenalty = -5 * user.reportedBy.length;

Reports	Penalty
0	-0
1	-5
2	-10
3	-15
...	...

Note: Only counts reports with status RESOLVED (not DISMISSED)

4. Interaction Bonus

Successful matches indicate positive community engagement.

typescript

const totalMatches = user.sentInterests.length + user.receivedInterests.length;
const interactionBonus = Math.min(10, Math.floor(totalMatches / 2));

Accepted Interests	Bonus
0-1	+0
2-3	+1
4-5	+2
...	...
20+	+10 (max)

Note: Counts both sent AND received accepted interests

5. Screenshot Penalty (Direct)

Screenshots are detected client-side and reported to the server.

typescript

export async function reportScreenshotDetected(userId: string): Promise<void> {
  await applyTrustPenalty(userId, 10, 'screenshot_detected');
}

Event	Penalty
Screenshot detected	-10

Factor Summary

Factor	Impact	Max
Base Score	+50	50
Verification (IDENTITY)	+10	10
Verification (FULL)	+20	20
Account Age	+1/month	10
Matches	+1/2 matches	10
Reports	-5 each	N/A
Screenshots	-10 each	N/A

Score Ranges

Maximum Possible: 50 + 20 + 10 + 10 = 90 (without penalty deductions) Minimum Possible: 0 (many reports/screenshots)

Trust Tiers

Based on the calculated score, users fall into tiers:

typescript

const getScoreLevel = () => {
  if (score >= 80) return { level: "high", label: "Highly Trusted", color: "text-emerald-500" };
  if (score >= 60) return { level: "medium", label: "Trusted", color: "text-blue-500" };
  if (score >= 40) return { level: "low", label: "Building Trust", color: "text-amber-500" };
  return { level: "very-low", label: "New Member", color: "text-gray-500" };
};

Score Range	Tier	Badge	Color
80-100	High	Highly Trusted	Emerald
60-79	Medium	Trusted	Blue
40-59	Low	Building Trust	Amber
0-39	Very Low	New Member	Gray

Trust Badge Component

File: app/src/components/profile/TrustBadge.tsx

Visual Representation

tsx

const Icon = level === "high" ? ShieldCheck : 
             level === "very-low" ? ShieldAlert : Shield;

return (
  <div className={cn("flex items-center", color)}>
    <Icon className="h-4 w-4" />
    <span className="font-medium">{score}%</span>
  </div>
);

Badge Icons

Tier	Icon	Meaning
High	✓ ShieldCheck	Verified and trusted
Medium/Low	Shield	Standard user
Very Low	⚠ ShieldAlert	New or flagged

Penalty Application

Direct Penalty

Used for immediate trust reduction without recalculation.

typescript

export async function applyTrustPenalty(
  userId: string,
  penalty: number,
  reason: string
): Promise<number> {
  const user = await prisma.user.findUnique({
    where: { id: userId },
  });

  if (!user) {
    throw new NotFoundError('User not found');
  }

  const newScore = Math.max(0, user.trustScore - penalty);

  await prisma.user.update({
    where: { id: userId },
    data: { trustScore: newScore },
  });

  console.log(`Trust penalty applied to ${userId}: -${penalty} for ${reason}`);

  return newScore;
}

Penalty Events

Event	Penalty	Applied By
Screenshot detected	-10	Socket handler
Report resolved (valid)	-5 to -50	Admin action
Blackmail report	-20	Automatic

Blackmail Report Handling

Blackmail reports receive special treatment due to severity:

typescript

export async function createBlackmailReport(
  reporterId: string,
  reportedId: string,
  description: string,
  evidence?: any
): Promise<Report> {
  const report = await prisma.report.create({
    data: {
      reporterId,
      reportedId,
      type: 'BLACKMAIL',
      description,
      evidence: evidence || null,
      status: 'INVESTIGATING',  // Skip PENDING
    },
  });

  // IMMEDIATE -20 penalty
  await prisma.user.update({
    where: { id: reportedId },
    data: {
      trustScore: {
        decrement: 20,
      },
    },
  });

  return report;
}

Score Recalculation

When Triggered

Trust scores are recalculated:

Admin manually triggers recalculation
After verification status changes
Periodically via background job (not implemented)

Admin Recalculation

typescript

export async function updateTrustScore(userId: string): Promise<number> {
  const newScore = await calculateTrustScore(userId);

  await prisma.user.update({
    where: { id: userId },
    data: { trustScore: newScore },
  });

  return newScore;
}

Admin Endpoint:

POST /api/admin/users/:id/recalculate-trust

Discovery Impact

Trust score affects profile visibility in discovery:

typescript

const profiles = await prisma.user.findMany({
  where: whereClause,
  orderBy: [
    { verificationLevel: 'desc' },  // 1st priority
    { trustScore: 'desc' },          // 2nd priority
    { lastActiveAt: 'desc' },        // 3rd priority
  ],
  ...
});

Higher trust = Higher visibility:

Users with high trust scores appear earlier in discovery
Combined with verification level for overall ranking

Score Examples

New User (Unverified)

Base:          50
Verification:   0 (NONE)
Age:            0 (new account)
Matches:        0 (none yet)
Penalties:      0 (clean record)
─────────────────
TOTAL:         50 (Building Trust)

Established Verified User

Base:          50
Verification: +20 (FULL)
Age:          +10 (1 year+)
Matches:      +10 (20+ matches)
Penalties:      0 (clean record)
─────────────────
TOTAL:         90 (Highly Trusted) - capped at 100

Problematic User

Base:          50
Verification: +10 (IDENTITY)
Age:           +5 (5 months)
Matches:       +5 (10 matches)
Penalties:    -20 (4 resolved reports)
─────────────────
TOTAL:         50 (Building Trust)

Severely Penalized User

Base:          50
Verification: +10 (IDENTITY)
Age:           +3 (3 months)
Matches:       +2 (4 matches)
Penalties:    -50 (blackmail + reports + screenshots)
─────────────────
TOTAL:         15 (New Member)

Admin Controls

View Trust Score

GET /api/admin/users/:id

Response includes trustScore field.

Manual Update

PATCH /api/admin/users/:id
{
  "trustScore": 75
}

Recalculate

POST /api/admin/users/:id/recalculate-trust

Response:

json

{
  "success": true,
  "data": {
    "trustScore": 72
  }
}

Apply Penalty via Report

PATCH /api/admin/reports/:id
{
  "status": "RESOLVED",
  "applyPenalty": true,
  "penaltyAmount": 15
}

Future Improvements

Potential Additional Factors

Response rate: Reward users who reply to messages
Profile completeness: Bonus for filled-out profiles
Photo quality: AI-assessed photo quality
Activity consistency: Regular engagement patterns
Positive feedback: Let users rate interactions

Decay Over Time

Consider:

Slowly restoring trust after periods without incidents
Penalties that decrease over time
"Probation" periods after violations

Verification Weight

Consider increasing verification bonus:

Full verification is the gold standard
Make it worth even more (+30 or +40)

API

Backend API

Okia AI Service

Frontend

Student Dashboard

Teacher Dashboard

Parent Dashboard

School Admin Dashboard

Zaptam Trust System ​

Overview ​

Score Calculation ​

Algorithm ​

Formula ​

Trust Factors ​

1. Verification Bonus ​

2. Account Age Bonus ​

3. Report Penalty ​

4. Interaction Bonus ​

5. Screenshot Penalty (Direct) ​

Factor Summary ​

Score Ranges ​

Trust Tiers ​

Trust Badge Component ​

Visual Representation ​

Badge Icons ​

Penalty Application ​

Direct Penalty ​

Penalty Events ​

Blackmail Report Handling ​

Score Recalculation ​

When Triggered ​

Admin Recalculation ​

Discovery Impact ​

Score Examples ​

New User (Unverified) ​

Established Verified User ​

Problematic User ​

Severely Penalized User ​

Admin Controls ​

View Trust Score ​

Manual Update ​

Recalculate ​

Apply Penalty via Report ​

Future Improvements ​

Potential Additional Factors ​

Decay Over Time ​

Verification Weight ​

Zaptam Trust System

Overview

Score Calculation

Algorithm

Formula

Trust Factors

1. Verification Bonus

2. Account Age Bonus

3. Report Penalty

4. Interaction Bonus

5. Screenshot Penalty (Direct)

Factor Summary

Score Ranges

Trust Tiers

Trust Badge Component

Visual Representation

Badge Icons

Penalty Application

Direct Penalty

Penalty Events

Blackmail Report Handling

Score Recalculation

When Triggered

Admin Recalculation

Discovery Impact

Score Examples

New User (Unverified)

Established Verified User

Problematic User

Severely Penalized User

Admin Controls

View Trust Score

Manual Update

Recalculate

Apply Penalty via Report

Future Improvements

Potential Additional Factors

Decay Over Time

Verification Weight