[Feature Proposal] Seal Progression Through Active Gameplay (Unique Hunting & Job System)

Ouverte
#3,202 1 commentaire 1 réaction 0 personnes assignées Voir sur GitHub

Les mainteneurs répondent en général sous 1 jour

Personne n'a encore pris cette issue.

Évaluation

Difficulté
5/5
Temps estimé
Plus d'une semaine
Accessibilité débutants
8/100
Type d'issue
Fonctionnalité
Clarté
Plutôt claire
Activité
Active
Stack technique
go, typescript
Domaine
backend, game-dev

Piste de recherche

This is an RFC for design discussion, not an implementation request; the proposal is port-only and gated behind SRO_ flags, and its numbers are proposed starting values. Start by reading CONTRIBUTING §3 and the 'What exists today' table, which points to SRO_TRADE_TIERS, SRO_SEAL_CENSUS and SRO_SEAL_BOARD. Done here means the owner has settled the open decisions (pity mechanic #1690, provenance state, token caps); no code should be written from this thread.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Description

area: server enhancement priority: low status: gated

RFC — design discussion, not an implementation request.
Everything below is port-only, not native. Per CONTRIBUTING §3 it would ship disabled by default behind its own SRO_ flags (off = native), and each gameplay deviation lands only with the owner's approval.
All numbers in this proposal are proposed starting values, not confirmed mechanics, unless they appear in the "What exists today" table.

Summary

Add an optional, deterministic path from Seal of Star +10 → Seal of Moon +0 → Seal of Moon +10 → Seal of Sun +0, paid for with two account-bound currencies earned only through active play:

  • Unique Tokens: from cooperative Unique hunting.
  • Job Tokens: from Trader, Hunter and Thief activity, all three roles.

Natural Seal drops stay as they are. The goal is a predictable path to Seal of Sun that rewards meaningful participation, while multi-account farming gains little from it.

Motivation

  • Seal of Sun currently comes from random drops, which scale with how many characters someone can keep farming, not with how actively they play.
  • Unique hunting and the job system have little influence on endgame gear progression.
  • The gap between casual, active and large-scale farming players keeps growing.

Related context: #1604 (loot and progression epic), #1690 (open, gated: pity for top-tier seals), #2481 (launch Sun scarcity), #2958 / #3162 (5-star caravan stakes). This proposal is an alternative or complement to #1690's pity mechanic, so the two should be decided together.

Design principles

  1. Reward active play, not hours online.
  2. Encourage cooperation and competition between players, PvE and the three jobs.
  3. Keep Seal of Sun hard to get and valuable.
  4. Make progression predictable, but keep the existing alchemy challenge (reaching +10).
  5. Stop multi-account farming from scaling rewards indefinitely.
  6. Don't punish legitimate players through over-aggressive detection.
  7. Server-authoritative, configurable, auditable.

What exists today vs. what is proposed

Topic Confirmed today Proposed here
Seal transformation (Star → Moon → Sun) Not in native v1.150; no such recipe in OpenSRO Two guaranteed recipes (below)
Item binding None: no bound-item concept in the server, no bound-item UI in the v1.150 client Account binding for tokens and crafted Seals
Progression currencies None Unique Tokens, Job Tokens
Caravan stars SRO_TRADE_TIERS (port-only): stars come from the cargo's gold value at reference prices; the price actually paid is not stored. Since #3142 a small transport filled to the brim can reach 5★ (#3162: ~9.4M vs ~29.5M of cargo) Job Token rewards scale with server-verified cargo value, not the star alone
5★ stakes #2958: larger thief cut, hunter bonus (gold/EXP), nameplate star Unchanged; tokens are a separate progression reward
Cargo on transport death Being traced to native in #2563 (reports: goods vanish when the trader dies or a player kills the transport, but drop when an NPC kills it) Provenance follows whatever native rule lands
Seal rates & alchemy SRO_BETA_SEAL_RATES, SRO_SUN_EVENT_ONLY, SRO_ALCHEMY_RATES (#1604). Beta census: 190 degree-9 Sun across ~300 endgame characters (#2481) No change to drops or alchemy rates
Seal telemetry SRO_SEAL_CENSUS, SRO_SEAL_BOARD exist Reused for balance monitoring; new flag names avoid collisions

1. Crafting

Proposed recipes
Requirement Star → Moon Moon → Sun
Equipment SoStar +10 SoMoon +10
Unique Tokens 10 20
Job Tokens 15 30
Success rate 100% 100%
Result SoMoon +0 SoSun +0

Full path: 30 Unique Tokens + 45 Job Tokens, plus reaching +10 twice through normal alchemy.

Rules
  1. Crafting consumes the equipment and the tokens and creates the next tier at +0.
  2. The transformation always succeeds. Reaching +10 stays the normal, risky alchemy; adding a second random roll on top would only add frustration.
  3. Item type, degree, race and weapon/armor identity are preserved. If a Star item has no Moon/Sun counterpart in the data, the recipe isn't available for it.
  4. Attribute carry-over must be an explicit, documented rule: magic options (blues), white stats and any other per-item attribute either carry over by a defined conversion or reset, the same way every time.
  5. Crafting is atomic: an interrupted craft never duplicates or destroys an item.
Binding (proposed)
Item Binding
Unique Tokens, Job Tokens Account-bound
Crafted SoMoon / SoSun Account-bound from the first successful transformation
Naturally dropped Seals Unchanged

Binding has to hold on every transfer path: exchange, stalls, guild storage, account storage, pet inventory, dropping, mail and any future path. Otherwise tokens farmed on secondary accounts turn into gear on a main character.


2. Unique Token rewards

Tokens are credited by the server on the Unique's death, not dropped as loot. The usual alternatives all fail:

  • Last hit: rewards kill stealing and burst damage.
  • Anyone who attacked: exploited by many low-effort characters.
  • Winning party only: excludes legitimate support groups.
Proposed approach: the top two contributing parties
Unique type Tokens per eligible player
Standard 1
High-level / difficult 2

Eligibility (all proposed and configurable):

  1. Within a set distance of the Unique when it dies.
  2. Party contribution of at least 10% of total effective contribution.
  3. Only the two highest qualifying parties are rewarded; a solo player counts as a one-member party.
  4. One reward per account per Unique instance.
  5. The player must have meaningfully taken part, not just stood nearby.
  6. 8 Unique Tokens per account per week, across all characters.
Support contribution

Raw damage alone underrates Clerics, Bards and tanks. Possible signals: damage dealt, effective healing on participants, damage absorbed while tanking, useful buffs held on active participants, time engaged. Overhealing, refreshing buffs that aren't needed and absorbing trivial damage must not count. Support credit is capped and normalized so artificial activity can't beat real participation.

Anti-abuse for Uniques

No reward for idling nearby, hit-and-leave, reconnecting or swapping parties, or duplicate claims per account and instance. No client-supplied contribution values. Repeated groups doing minimal, identical actions are flagged.

Activity metrics can't prove a player is human: bots can imitate damage, healing and movement. The protection comes from caps and existing conduct enforcement, not from activity checks alone.


3. Job Token rewards

All three roles must have a fair path. All values are proposed.

3.1 Trader
Caravan tier Trader tokens
1–3★ 0
4★ 1
5★ 2

The tier is computed on the server, and per #3162 the reward should scale with verified cargo value, so that a small full transport can't farm rewards sized for a large one. A delivery qualifies only if:

  1. The cargo was bought through a legitimate trade transaction.
  2. Origin and destination are valid and distinct, with a minimum route requirement.
  3. The cargo value meets the tier.
  4. It hasn't already produced a progression reward.
  5. The account is under its weekly cap.

Partial deliveries may earn proportional credit, and the same cargo is never rewarded twice.

3.2 Hunter

Rewards are tied to a validated Trader delivery. Joining the party isn't enough:

  • Registered as an escort for that caravan, wearing the job suit.
  • Present for a meaningful share of the route (proximity sampled along the way).
  • One claim per account per delivery, whatever the parties or characters.

The existing hunter share calculation is a natural integration point, with escort participation verified on top.

A shared pool per caravan stops 40 escort accounts from getting 40 full rewards:

Tier Hunter pool Max per Hunter
4★ 2 1
5★ 4 2

Two eligible Hunters on a 5★ caravan get 2 each, four get 1 each. Unused capacity isn't handed to ineligible participants.

3.3 Thief

Tokens come from stealing legitimate Trader cargo and selling it at the Thief NPC. A PvP kill alone never generates tokens.

Stolen cargo tier Thief pool Max per Thief
4★ 4 2
5★ 6 3

The reward follows the value actually stolen and sold, not the caravan's advertised tier. Robbery participants (fighting, destroying the transport, securing, carrying and selling, meaningful support) are fixed by the server at the time of the incident, so nobody can join afterwards and collect.


4. Cargo provenance (the anti-abuse foundation)

Every qualifying purchase gets a server-side record that follows the cargo through its whole life:

Purchase → transport → drop/theft → recovery → Thief sale or Trader delivery

It must capture at least: the original buyer's account and character, origin, time, server-computed value and quantity, a lineage ID that survives drops, splits, transfers, restarts and disconnects, and the robbery participants with the quantities stolen and sold.

Key rule: reward eligibility is never decided from the goods' current owner alone. The original Trader account and the lineage stay attached until the transaction is final.

Feasibility note: SRO_TRADE_TIERS doesn't store the purchase price today, so provenance is genuinely new persistent state. It's the largest single cost of the job half of this proposal.


5. Hard rules (checked before any reward)

Rule Proposed behaviour
A: No self-farming A Thief earns nothing from cargo bought by another character of the same account: automatic rejection. Admins may also block proven linked accounts. A shared IP alone never proves common ownership.
B: Cargo is rewarded once Each unit of cargo contributes once per qualifying transaction. Destroyed or unsold goods give no Thief reward. Splits keep the original quantity and value. Recovered cargo is reconciled without duplication.
C: Same-pair diminishing returns (rolling 24 h, both accounts, any characters) 1st: 100% · 2nd: 100% · 3rd: 50% · 4th+: 0%. Applies to progression only, not gold, EXP or PvP. It's an economic limit, not an accusation: repeat encounters are normal on small servers.
D: Valid route Valid origin and destination, a minimum distance and plausible travel time, server-verified value, allowed transport and teleport use. Robberies close to a gate get reduced credit rather than being rejected, so no artificial safe zones are created.
E: Weekly account caps 8 Unique / 15 Job Tokens per account per week, across every character and job. Switching job doesn't reset the cap. Fractions are stored as fixed-point credits to avoid rounding exploits.
F: Server authority The client never decides the amount, eligibility, contribution, linkage, cooldowns, craft outcome or transferability.

6. Suspicious behaviour detection (later stage, optional)

The hard rules catch the obvious exploits. Coordinated play (a Trader and a Thief faking a fight, for example) needs pattern detection over time, used to prioritise human review, never to ban automatically.

Signals to consider:

  • Repeated pairs: Thief B gets 85% of their cargo from Trader A, 8 times in a day.
  • Cargo patterns: drops right after purchase, the same isolated coordinates, identical purchase-to-theft intervals.
  • Combat anomalies: no hostile interaction across many handoffs, the Trader stopping at the same spot every time. No combat is a weak signal (a Trader may flee or surrender), and a fight doesn't prove legitimacy either.
  • Account relationships: verified common ownership, plus device or session similarities where lawfully available. A shared IP or device is supporting evidence only: households, universities, VPNs.
  • Networks: rotating Traders and Thieves to dodge pair cooldowns shows up only in the interaction graph (concentrated cargo flows, small clusters taking a disproportionate share of rewards, accounts alternating roles).
Illustrative scoring (not validated thresholds)
Signal Score
4+ robberies by the same pair in 24 h +25
>70% of a Thief's qualifying cargo from one Trader over 7 days (≥5 robberies) +20
Repeated unusually short routes or early handoffs +20
4+ handoffs with no hostile interaction +10
Highly repetitive location and timing +15
Strongly correlated activity in a small account network +25
Shared IP or device characteristics +10
Score Response
0–34 Normal
35–59 Monitor / review
60+ (several independent strong signals) May hold the affected progression reward pending review
Confirmed abuse Revoke illegitimate rewards and apply the published conduct policy

Correlated signals (repeat pair + high share from one victim) must not be counted as independent proof. A held reward never blocks the normal job transaction: gold and EXP proceed, only the token is pending, and it shows a visible reason and an audit trail. The target is a decision within 48 h. If the team can't staff timely reviews, the first release should rely on hard caps and passive monitoring, with the hold disabled.

Escalation
  1. First suspicious pattern: log it, monitor more closely, apply the normal pair cooldown. No accusation.
  2. Repeated strong signals: flag the accounts and hold the rewards from those specific transactions.
  3. Confirmed collusion: revoke the tokens, and reverse crafted progression where provenance clearly allows it. Apply the published policy.

If collusive reward farming is to be punishable on its own, separately from the existing botting and alt-farming rules, the prohibition must be published before enforcement. No retroactive punishment.


7. Dynamic job balance (optional)

If almost everyone trades, Thieves need more incentive. The measure is accounts that completed qualifying job activity in the last 7 days: not idle job-suited characters, not duplicate characters, and not proven linked clusters.

Active Thief share Thief token-credit bonus
< 25% +30%
25–34% +15%
≥ 35% 0%

The bonus applies to progression credits before they become whole tokens (and, at least initially, not to gold or job EXP). It's recalculated weekly from a smoothed average, with a minimum sample size and a cap on the change from one week to the next, so job-switching can't game it. If Thieves become heavily over-represented, a smaller bonus could go to Traders and Hunters instead.


8. Feasibility and implementation considerations

Kept at the level of constraints. A concrete plan would follow the owner's decisions.

  • v1.150 client constraints. The client defines which items and UI exist. Open choice: tokens as items (reuse existing item records? new ones?) or as an account currency shown in an Experimental client panel. Likewise the transformation could go through the alchemy window (its v1.150 messages are fixed) or an exchange NPC, the way event exchanges already work. Binding needs new client UI either way.
  • Server authority and reward integrity. Every grant is idempotent (a stable key per source event, account and reward type) and atomic with cap consumption and an auditable ledger entry that records the rule version.
  • Performance. Nothing expensive in the combat or tick hot path: compact event records and indexed counters, with network analysis run asynchronously. Hard eligibility checks still run before a reward is granted.
  • Configurability. Every value in this RFC is a tunable, not a hardcoded constant.
  • Telemetry. The existing SRO_SEAL_CENSUS and the kill and EXP stats from #1690 can feed the balance metrics.
Proposed flags (none of these exist yet; all default off)
Flag Purpose
SRO_SEAL_PROGRESSION Crafting recipes, token balances, binding
SRO_SEAL_UNIQUE_REWARDS Unique Token attribution
SRO_SEAL_JOB_REWARDS Job Token attribution and cargo provenance
SRO_SEAL_JOB_BALANCE Dynamic job bonus
SRO_SEAL_ANTI_ABUSE Behavioural detection
SRO_SEAL_REWARD_HOLD Temporary reward withholding

Both settings get tested, per CONTRIBUTING: flag off behaves exactly like native.

Suggested order, if accepted (for discussion, not a plan)
  1. Crafting, binding, token balances, ledger, weekly caps (test-granted tokens).
  2. Unique attribution.
  3. Cargo provenance, then Trader, Hunter and Thief rewards with the pools and hard rules A–F.
  4. Detection, review queue, reward hold and reversal.
  5. Dynamic job balance, tuning from live data.

The sophisticated scoring and network detection can wait until the core reward path is reliable.


9. Edge cases to cover

Scenario Expected
Solo Trader, legitimate 5★ route Trader rewarded
8 Hunters join, 2 actually escort Only the 2 share the pool
Thief robs a character on the same account No progression reward
Same Trader–Thief pair, 5 times in 24 h Diminishing returns apply
Small transport filled to 5★ vs a large one (#3162) Tokens follow verified value; the small cycle doesn't pay more per gold at risk
Cargo split between several Thieves / sold in several batches Total never exceeds the cargo's reward budget, no duplicates
Transport killed by a player vs by an NPC (#2563) Provenance and rewards follow the native outcome
Trader recovers part of the cargo and delivers it Quantities reconciled, nothing paid twice
Two independent players share an IP No automatic rejection
Reconnect or party swap mid-Unique No duplicate reward
Bard / Cleric meaningful support Can qualify, within the cap
Idle character next to a Unique No reward
The same event processed twice Exactly one reward
Server crash during a craft No duplication, no loss
Job change mid-week Weekly cap unchanged
Weekly reset boundary (time zone, a craft exactly at reset) Defined and tested
High suspicion score, nothing confirmed Review, never an automatic ban
Accounts rotating Traders to dodge pair cooldowns Flagged by network analysis
Flag turned off after a season with it on Defined behaviour for balances and crafted items (see open questions)

10. Balance target and monitoring

Proposed target: ~4–6 weeks of consistent active play to gather the tokens for a full Star → Sun path, not counting the time to reach +10 twice. At the proposed caps, the floor is 4 weeks (30 Unique Tokens at 8 a week) and 3 weeks of Job Tokens (45 at 15 a week), so the target assumes play close to the caps.

Track the median, P75 and P90 rather than averages: tokens per active account per week, time to first crafted Moon and Sun, Unique and job participation, Hunter count per caravan, the share of each job, held rewards and the false-positive rate. On the economic side: crafted vs naturally dropped Seals in circulation, the value of SoStar, alchemy material consumption, trade volume, and whether binding chokes off the gear market. Read this alongside #2481's Sun-scarcity target.


11. Open questions for the owner and community

  1. Both token types required? Or a slower single-activity path? (Author's preference: require both, to push players across activities.)
  2. Account or character binding? (Preference: account.)
  3. Token representation: items or an account currency? Alchemy window or exchange NPC?
  4. Relationship with #1690: does this replace the pity mechanic, complement it, or wait for the same live data?
  5. Support contribution: how do we measure Cleric, Bard and tank work in a way that's bounded and can't be farmed?
  6. Thief group pool: does 6 tokens per 5★ encourage organised robberies without letting large parties multiply rewards?
  7. Job bonus scope: progression credits only, or gold and job EXP too? (Preference: credits only at first.)
  8. Moderation capacity: is there staff for a 48 h review queue, or should the first release use caps and passive monitoring only?
  9. Attribute carry-over: keep or reset blues and white stats on transformation?
  10. Turning the flag off: what happens to unspent tokens and crafted bound Seals?
  11. Crafted vs dropped: should a crafted Seal be distinguishable (tooltip, seal board, census)?

Expected outcome

  • Casual players make real progress, and active players get a reliable path to Seal of Sun.
  • Unique hunting becomes more attractive, and all three jobs have a reason to meet on the road.
  • Multi-account farming gets little extra progression, and abuse is traceable.

Seal of Sun should stay hard to get, but getting it should reward meaningful participation rather than the ability to run the most farming accounts. This won't eliminate botting or collusion on its own. It limits what they earn, makes them traceable, and gives active players a real alternative.


References: server rules · player conduct. All amounts, thresholds, scores and cooldowns are suggested starting values, to be tuned from live data.

Langage dominant
Go
Étoiles
97
Forks
61
Merge moyen
4 h 14 min
PR mergées (30 j)
913

Préparer son environnement

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Autres issues de opensro-dev/opensro

Toutes les issues de opensro-dev/opensro

Issues similaires

Plus d'issues Go

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.