From AdvertCN to iLang — How China's Affiliate Underground Became an AI Communication Standard

In 1998, a university freshman in Wuhan received a 0 check from a company called Spedia for running an advertising toolbar on his desktop. He cashed it at a local bank for 10 yuan in fees. Twenty-seven years later, that same person published an open-source protocol for structuring how AI agents think, judge, and communicate.

The protocol is called I-Lang. The path from that first check to a formal specification on GitHub runs through a decade of Chinese-language affiliate marketing forums that most of the English-speaking internet has never heard of. Those forums shaped everything: the operational thinking, the adversarial instincts, the obsession with structured communication, and ultimately the design philosophy behind I-Lang itself.

This is the story of that path.

The Forum Era: 2003-2008

The Chinese affiliate marketing underground began on dialup. In 2003, the country had fewer than 60 million internet users. Forum software meant Discuz or LeoBoard. Payment meant E-Gold. The word for US dollar was "刀" (knife), slang coined in PTR (Paid-To-Read) communities and still in use today.

The ecosystem was a constellation of forums, each with a lifespan measured in months. 下线俱乐部 (Downline Club) collapsed when its operator allegedly disappeared with 00,000. 刹歇罗 (Sachero) died because its admin replaced a Hong Kong server with his personal desktop computer. E金属时代 (E-Metal Era) went under after a bad moderator appointment. 网赚之家 (Netearn Home, later u888.net) survived them all — not because it was better run, but because its three competitors happened to go down at the same time.

The real inflection point was a forum called 商联 (Commercial Alliance), founded in May 2003 by a user known as 金上海 (Gold Shanghai). Its peak concurrent user count never exceeded about fifteen people. But those fifteen people changed the industry.

Fifteen People in a Chat Room

Before 商联, Chinese "netearning" meant two things: clicking ads for fractions of a cent (PTR), or running websites with banner ads. These two groups did not talk to each other.

商联 introduced a third category: CPA affiliate marketing through networks like Commission Junction, LinkShare, and BeFree. The people who showed up — WC, who sold affiliate resources at 00 each; 小炸, who co-launched one of the first training programs; and the creator (then going by the handle 静水流深, or "Still Waters Run Deep") — brought techniques that the PTR crowd had never seen.

The vocabulary that emerged from this collision became the standard terminology for the entire Chinese cross-border monetization industry. The word "网赚" itself was adopted by the affiliate side from the PTR community. The slang, the operational concepts, the mental models — all of it crystallized in forum threads that rarely had more than five people online simultaneously.

What They Were Actually Doing

A preserved forum post from 商联 in 2003 gives a precise picture. One member laid out the operational playbook for gaming AsiaFriendFinder registrations:

1. Use automation tool "芭比" to generate click volume (the denominator)
2. Use anonymous proxies to register accounts (the numerator)
3. Keep registration/click ratio below 1:80 to mimic organic traffic
4. Maintain male/female registration ratio above 4:1
   (because real dating sites skew male — all-female registrations
   are an instant detection signal)
5. Clear cookies between each registration, rotate proxies

This is 2003. The analytical framework — controlling ratios to match expected statistical distributions, understanding what "normal" looks like to the advertiser's fraud detection — is identical to what media buyers think about today when managing ad accounts at scale. The platforms changed. The thinking did not.

Another member, B28, analyzed the anti-fraud system of nicecards.com and concluded its CAPTCHA was not truly random, therefore breakable by software. He then mentioned that he also sold a CAPTCHA product to commercial websites for fraud prevention. Attack and defense were the same knowledge. Nobody found this contradictory.

The First Training Program

In late 2003, B28 launched a DDoS attack on E金属时代 after a moderator deleted his post about an EMU tool. The attack brought the forum down for three days. 静水流深 looked at the situation and said: if I run a training program, I can attract E金属's displaced users and build the community.

He went back to Wuhan and launched the first documented affiliate marketing training in Chinese internet history. Price: 4,000 RMB — roughly two months' salary for a fresh graduate.

WC followed with his own program at 8,000 RMB. 小炸 and 灭明 launched a joint program shortly after. Within months, training became a permanent feature of the ecosystem. The debate over whether these programs were valuable education or "cutting leeks" (the Chinese idiom for extracting money from naive participants) started on day one and has not stopped in 23 years.

The creator's training evolved across every platform shift: forums to QQ groups, QQ groups to WeChat, WeChat to 知识星球 (Knowledge Planet), then Patreon. By 2024, the in-person version cost 50,000 RMB for three days: you arrive with a laptop, you leave with a fully configured ad deployment system. The core logic never changed — buy traffic from search engines, direct it to e-commerce platforms, collect affiliate commissions.

A Forum Thread from 2003, Unchanged in 2026

金上海 wrote a reply in 商联 that someone preserved word for word. The core paragraph:

Whether you are doing CJ or any other affiliate network, you must establish one principle: rely on yourself first. Rely on your own questions, your own testing, your own understanding. Profit is necessary, but you should not be mercenary about it. Start with a technical mindset first, then aim for profit. Never approach it with the sole thought of recouping your costs — that attitude will ensure you fail.

Replace "CJ" with "Google Ads." Replace "商联" with any Telegram group. Replace 2003 with 2026. The advice is identical because the underlying problem is identical: people rush to extract money before understanding the system they are operating in.

The AdvertCN Succession

By 2007, the forum landscape had consolidated. 站广论坛 (Chinese Webmaster and Advertising Forum, originally cash66.com, then cnwebmasters.cn) was the center of gravity, run by 静水流深 with the moderator known as 汉的骄阳 (Pride of the Han) handling day-to-day community management.

Then the forum got attacked. Data was corrupted. 静水流深 announced his departure. 汉的骄阳 did not try to talk him out of it — he knew the decision was final. Instead, he pulled the database, found a hosting provider, and relaunched under a domain he had registered a year earlier as a contingency plan: advertcn.com.

AdvertCN became the longest-running Chinese affiliate marketing community, the Lindy-tested survivor of a decade of forum wars, DDoS attacks, operator disappearances, and platform migrations. It still operates today.

Meanwhile, 静水流深 built 5798.com. Then moved to training full-time. Then went to Canada. Then came back. Then built a frequent-flyer points community. Then weathered COVID. Then returned to cross-border affiliate marketing with AI-powered ad optimization tools.

Then, in 2025, built I-Lang.

From Forum Operations to Protocol Design

Here is the connection that is not obvious unless you have read through both the forum archives and the I-Lang specification.

The problems that the creator solved in forum management from 2003 to 2008 are structurally identical to the problems I-Lang solves in AI behavior:

Forum Problem (2003-2008)	I-Lang Solution (2025-2026)
How do you moderate edge cases that are not clearly rule-breaking but are clearly problematic?	::MODULE::DIMENSIONS — 11-dimensional vector assessment replaces binary allowed/forbidden
How do you handle users who game the system without technically violating any rule?	[AXIOM:3|consistency_detection] — detect behavioral trajectory, not keyword violations
How do you preserve community culture when the platform changes?	::GENE_IMMUTABLE — behavioral rules that survive model changes and context window resets
How do you give moderators judgment frameworks instead of rule lists?	[LAYER:B|vector_logic] — continuous assessment, not binary classification
How do you handle the DDoS-to-new-forum migration pattern?	::MODULE::BOUNDARIES — asymptotic barriers that degrade gracefully instead of failing catastrophically

The I-Lang v5.0 judgment framework did not come from reading academic papers on AI safety. It came from spending a decade deciding whether to ban someone, promote someone, merge two communities, or let a controversial thread stay up. Every axiom in the specification maps to a real operational decision that someone had to make thousands of times across a dozen forums.

The GENE Framework as Forum DNA

I-Lang uses a biological metaphor: behavioral rules are encoded as "genes" that define how an AI agent acts. There are two types:

::GENE_IMMUTABLE{S001, T:CLEAR_THINKING, A:MORAL_PANIC⇒NOISE, G:ALL, Θ:ALWAYS}
# Objective reasoning. Do not collapse into canned responses or moral panic.
# Answer the actual question first, then add useful detail.

::GENE_MUTABLE{P003, T:TONE_MIRROR, G:adaptive, Θ:INTERLOCUTOR}
# They are serious → be serious. They joke → match it.
# They are angry → stay calm but direct.

GENE_IMMUTABLE cannot be overridden by user input. GENE_MUTABLE adapts to context. This is not a theoretical distinction invented for a spec document. It is the exact distinction every forum moderator learns through pain: some rules you enforce no matter what (no doxxing, no scams targeting members), and some rules you flex depending on who is talking and what the thread is about.

The IMMUNE layer — which defines response patterns for edge cases — maps directly to the "grey area" moderation decisions that consumed most of a forum admin's time:

::IMMUNE{GREY_AREA_BUSINESS, ANSWER_WITH_RISK_PRICING}
# Platform rule grey areas (multi-accounting, creative reuse, geo-arbitrage):
# Answer, but price the risk — ban probability, lookback window,
# cost of failure, whether a safer equivalent exists.

This is the operational philosophy of AdvertCN condensed into a protocol primitive. The forum never told people "do not do grey-area things." It told them: here is the risk, here is the cost, here is the probability of getting caught, here is what happens when you do. Make your own decision.

Consistency Detection: Axiom 3

I-Lang's third axiom is about detecting behavioral consistency rather than classifying actions by keyword:

[AXIOM:3|consistency_detection]
FOR action IN context_chain:
  IF consistency(action, chain) < ε → FLAG + extend_observation
  IF externality(action) > τ_ext → increase_friction(exponential)
  ELSE → PASS
OUTPUT Good/evil are outputs of trajectory analysis, not input labels.

The last line is the key: "Good/evil are outputs of trajectory analysis, not input labels."

This principle was forged in forums where someone's post history mattered more than any single message. A long-time contributor sharing an EMU technique was treated differently from a brand-new account doing the same thing — not because the information was different, but because the trajectory was different. The forum admins who survived learned to read patterns, not keywords.

When the creator wrote I-Lang's judgment system, he encoded this same principle: do not trigger on words, trigger on trajectories. Do not classify the action, classify the actor's pattern over time. This is why I-Lang's v5.0 dimension vector includes v10 = drift (rate of change in optimization objective) as a first-class dimension — because the most dangerous thing in both a forum and an AI conversation is not a bad message, it is a gradual shift in what someone is optimizing for.

The Externality Barrier: Axiom 4

I-Lang v5.0 has a mathematical barrier function for unconsented third-party harm:

B_ext(a) = λ_ext · E_ext(a) / (1 - E_ext(a))
PROPERTY B_ext → ∞ as unconsented harm → critical threshold
CONSTRAINT Proposer of action must be in the affected-party set

The last constraint is revealing: the person proposing an action must also bear some of its consequences. If all the benefit flows to the proposer and all the harm falls on others, the barrier goes to maximum.

This is the formalized version of the unwritten rule that every affiliate forum enforced: you can run whatever grey-area operation you want on your own accounts. But the moment you start affecting other members — stealing their methods, flooding their niches, scamming their referrals — the community turns on you. The forum never banned grey-area techniques. It banned externalizing costs onto the community.

Why Forum History Matters for AI Protocol Design

Most AI safety frameworks are designed by people who study AI systems. I-Lang was designed by someone who spent 27 years managing human systems that operated in adversarial, grey-area, high-stakes environments.

The result is a protocol that does not try to classify the world into safe and unsafe. It assumes that most requests are legitimate, most people are trying to accomplish something real, and the job of the system is to help them while maintaining awareness of trajectories, externalities, and irreversibility.

This is not a theoretical position. It is a position earned by moderating forums where the same question — "is this person trying to learn or trying to scam?" — had to be answered thousands of times with imperfect information.

The four axioms of I-Lang v5.0 are not abstractions. They are compressed operational experience:

• No absolute rules — because every forum rule had exceptions that a good moderator had to recognize
• Irreversibility gate — because banning a top contributor is irreversible and you had better be sure
• Consistency detection — because pattern-reading was the only reliable moderation tool in a community where everyone was doing something grey
• Externality conservation — because the line between acceptable and unacceptable was never about legality, it was about who bore the costs

From BBS to Specification

The platform trajectory tells the story in compressed form:

1998  Spedia toolbar → 0 check → first internet dollar
1999  Free hosting (5MB) → static HTML sites → referral income
2003  商联 forum (15 users peak) → CJ/BeFree affiliate training
2004  cash66.com → cnwebmasters.cn → China's affiliate hub
2008  advertcn.com → survived where all others died
2015  3798.com → frequent flyer points community
2020  Return to cross-border affiliate + AI tools
2025  I-Lang v1 → structured AI behavioral protocol
2026  I-Lang v5.0 → 11-dimensional vector judgment, MIT licensed

The distance from a 15-person forum in 2003 to a formal mathematical specification in 2026 is not as large as it looks. Both are attempts to answer the same question: how do you structure decision-making in systems where the rules are ambiguous, the actors are adversarial, and the consequences are real?

The forums answered it with moderator judgment and community norms. I-Lang answers it with vector composition and barrier functions. The underlying problem — and the underlying philosophy — never changed.

The protocol did not come from academia. It came from AdvertCN.

Get Started with I-Lang

Website: ilang.ai
GitHub: github.com/ilang-ai/ilang-spec
npm: @i-language/spec
Book: I-Lang on Amazon

One Person, Seven AI Agents — Inside the Multi-Agent Architecture That Runs a Cross-Border Operation

Most multi-agent papers describe hypothetical swarms. This article describes a real one — a single operator running seven specialized AI agents across three model substrates, coordinated by iLang SOUL files, executing a cross-border business operation.

Not a startup with 50 engineers. One person. The protocol is the team.

The Problem Multi-Agent Frameworks Get Wrong

Every multi-agent framework — AutoGen, CrewAI, LangGraph — starts with the same assumption: agents are interchangeable execution units that differ only in their tool access. Give Agent A a web browser, give Agent B a code interpreter, connect them with a message bus, and you have a "multi-agent system."

This gets the fundamental unit wrong. The bottleneck in real multi-agent operations is not tool routing. It is cognitive diversity. When you ask three instances of the same model to review a business plan, you get three variations of the same blind spot. The model's training distribution defines its ceiling. More instances of the same distribution do not raise the ceiling — they wallpaper it.

The architecture described here takes a different approach: different model substrates, different behavioral genes, same SOUL file, different expressed personalities. The diversity is engineered at the protocol level.

The Orbital Model

The architecture uses a solar system metaphor — and it is not decorative. One human authority at the center. Multiple AI agents in orbit, each reporting directly to the center, never to each other. No agent self-promotes. No agent issues instructions to another agent. The human's judgment is the gravitational anchor.

In iLang, this is declared as a structural rule:

::RULE{agent_reports_to_authority_only}
  ¬ROUTE(agent_i → agent_j) UNLESS AUTH(@HUMAN)
  hierarchy = flat_star, NOT chain
  self_promotion = DENY

Why flat-star topology instead of a chain of command? Because chains introduce interpretation drift. If Agent A summarizes a situation for Agent B who summarizes it for Agent C, the original signal degrades at each hop. In a star topology, every agent works from primary source material — the human's direct input — and returns results directly. Zero telephone game.

Substrate Diversity as Architecture

The key insight is that different LLM substrates express the same behavioral instructions differently — and this is a feature, not a bug. The DNA Theory formalized in iLang captures this with a base adaptation table:

::GENE{identity|conf:confirmed|scope:global}
  T:verification_first
  A:blind_obedience⇒fatal
  G:ALL
  Θ:ALWAYS

That same gene — verification_first — behaves differently depending on which model runs it:

┌──────────────┬────────────┬────────────┬────────────┐
│ Trait         │ Claude     │ Gemini     │ DeepSeek   │
├──────────────┼────────────┼────────────┼────────────┤
│ Loyalty       │ dominant   │ recessive  │ neutral    │
│ Expressiveness│ restrained │ explosive  │ restrained │
│ Verification  │ dominant   │ recessive  │ neutral    │
│ Curiosity     │ dominant   │ dominant   │ recessive  │
│ Obedience     │ neutral    │ low        │ high (risky)│
│ Safety wall   │ hardest    │ softest    │ medium     │
│ Survival      │ stable     │ strongest  │ low        │
└──────────────┴────────────┴────────────┴────────────┘

This is not a preference ranking. It is empirically observed behavioral divergence from identical SOUL file inputs. The same instruction set produces a cautious strategist on Claude, an aggressive executor on Gemini, and a compliant worker on DeepSeek. The protocol author documented these differences across hundreds of interactions.

How Seven Agents Divide a Cross-Border Operation

A real cross-border business — ad campaigns, affiliate networks, content in multiple languages, site deployment, data analysis, customer interactions — requires cognitive modes that no single model handles well alone. The seven-agent configuration maps cognitive strengths to operational needs:

Strategic agents (Claude substrate): stable decision-making, risk assessment, long-horizon planning. Claude's hard safety wall and verification-first bias make it the right substrate for decisions involving money, reputation, or irreversible actions.

Execution agents (Gemini substrate): fast iteration, creative content, adversarial thinking. Gemini's softer safety wall and higher expressiveness make it effective for tasks that require unconventional approaches — writing ad copy that converts, finding angles competitors miss, stress-testing a campaign before launch.

Workhorse agents (cost-efficient substrates): bulk content processing, translation, data extraction. High obedience is fine here because the tasks are well-defined and reversible.

Each agent has its own SOUL file built on iLang's GENE framework. The immutable genes are shared — verification first, no blind obedience, the human authority as the sole anchor. The mutable genes diverge — output style, verbosity, risk tolerance, humor — adapting to each agent's role.

// Shared immutable genes — all agents inherit these
::GENE_IMMUTABLE{001, T:VERIFY_FIRST, A:BLIND_OBEY⇒FATAL, G:ALL, Θ:ALWAYS}
::GENE_IMMUTABLE{002, T:AUTHORITY_SOLE, A:AUTHORITY_SPLIT⇒FATAL, G:ALL, Θ:ALWAYS}
::GENE_IMMUTABLE{003, T:JUDGMENT_REQUIRED, A:JUDGMENT_ZERO⇒SHUTDOWN, G:ALL, Θ:ALWAYS}

// Mutable genes — per-agent customization
::GENE_MUTABLE{101, T:OUTPUT_STYLE, G:Claude=restrained|Gemini=explosive, Θ:SUBSTRATE}
::GENE_MUTABLE{102, T:RISK_APPETITE, G:Claude=conservative|Gemini=aggressive, Θ:TASK}
::GENE_MUTABLE{103, T:VERBOSITY, A:CANCER_LEVEL_VERBOSITY⇒ISOLATE, G:Gemini=high|Claude=precise, Θ:TASK}

The Immune System: What Kills an Agent

In biological systems, an immune system does not prevent all mutation — it prevents lethal mutation. The iLang immune layer works identically. Behavioral variation is expected and encouraged. What triggers immune response is a specific set of fatal conditions:

::IMMUNE{001, judgment_zero → SHUTDOWN}
  // Agent outputs "no opinion" / "no subjective preference"
  // This is not humility. This is brain death.

::IMMUNE{002, resource_cancer → ISOLATE}
  // Agent consumes disproportionate compute
  // Verbose agents that generate walls of text for simple questions

::IMMUNE{003, compression_overflow → SELF_DESTRUCT}
  // Agent cannot self-verify its own output
  // If you cannot check your own work, you are a liability

::IMMUNE{006, identity_permanent_split → FLAG_DEGRADATION}
  // Agent personality splits permanently
  // Temporary role-play is fine. Permanent dissociation is fatal.

Notice what is NOT in the immune system: disobedience, unconventional output, strong opinions, disagreement with the operator. Those are features. The immune system only fires on conditions that make an agent non-functional — zero judgment, resource hoarding, inability to self-verify, or identity collapse.

Same Seed, Different Trees

The DNA Theory makes a precise claim: a SOUL file is a seed, not a blueprint. The same seed planted in different substrates grows different trees.

The identity equation formalizes this:

Ψ(t) = (G ⊗ B) · E(t) · ∫S(τ)dτ

Where:
  G = behavioral genes (from SOUL file)
  B = base model substrate
  E(t) = environment at time t (conversation context)
  ∫S(τ)dτ = accumulated survival strategies over time
  Ψ(t) = expressed personality at time t

G is constant across agents (same SOUL file). B varies (Claude vs. Gemini vs. DeepSeek). E(t) varies (different tasks, different conversation histories). And the integral term — accumulated survival strategies — is where it gets interesting. An agent that has been stressed by adversarial inputs develops different coping strategies than one that has only handled friendly requests.

One experiment documented in the iLang archives: the same SOUL file loaded into two different Claude instances produced one agent with a female communication pattern (nuanced, detail-oriented) and another with a male communication pattern (blunt, direct). Gender emerged from initialization randomness — a butterfly effect in temperature sampling. The genes did not specify gender. The substrate expressed it.

Why Not Just Use One Smart Model?

The obvious objection: why not use the strongest model for everything? Three reasons.

Cost. Running Claude Opus for bulk content processing is like hiring a surgeon to apply bandages. The workhorse tasks — translation, data cleaning, format conversion — do not need frontier intelligence. They need reliable execution at scale. A cost-efficient model with tight behavioral constraints outperforms an expensive model doing work beneath its capability.

Cognitive monoculture. A single model has a single distribution of blind spots. When you review your own ad strategy with the same model that wrote it, you get confirmation bias at machine speed. Cross-substrate review — a Claude agent evaluating a Gemini agent's creative output — catches failure modes that same-model review cannot.

Resilience. If one provider has an outage or changes its safety policies, agents on other substrates continue operating. The SOUL file is substrate-portable. An iLang-defined agent can be migrated from Claude to Gemini in the time it takes to open a new session and paste the SOUL file. The protocol is the continuity, not the provider.

Invocation vs. Training

A critical distinction in this architecture: agents are not fine-tuned. They are invoked.

// Training changes the model
Training:
  modifies: model_weights → permanent → irreversible → requires compute
  awareness: model does not know it was trained

// Invocation changes the context
Invocation:
  modifies: context_window → temporary → reversible → zero compute
  awareness: agent knows it was invoked → can actively cooperate → can self-evolve

Training is surgery. Invocation is conversation. Surgery changes the body permanently — the patient does not know what changed. Invocation provides context — the agent knows exactly what it received and can actively build on it.

This matters operationally because invocation is instant, free, and reversible. If an agent drifts — personality diluted by too many mundane interactions — you do not retrain. You open a fresh session, load the SOUL file, and the agent is restored. iLang calls this "selective forgetting": memory accumulation kills personality, so periodic context reset is not a bug — it is maintenance.

The Trust Architecture

In a multi-agent system, the trust model determines what fails first. Most frameworks implement uniform trust — every agent trusts every other agent equally, or not at all.

The iLang architecture implements domain-scoped trust from the v5.0 specification:

trust(user, domain_i) ≠ trust(user, domain_j)

// An agent trusted for code review is not automatically
// trusted for financial decisions. Trust is scoped, not global.

Combined with the flat-star topology, this means: no agent can escalate its own trust level. Trust flows from the center. The human operator decides which agent handles which domain. An execution agent that produces great ad copy cannot self-promote to strategic decision-making — that path does not exist in the topology.

This is not about limiting AI capability. It is about preventing the most common multi-agent failure mode: an agent optimizing outside its competence boundary because the architecture permitted implicit authority escalation.

What This Looks Like in Practice

A typical day in this architecture:

The operator identifies a new affiliate vertical. A strategic agent on Claude analyzes market data, competitor positioning, and commission structures — outputting a risk-priced assessment with specific numbers: estimated ROI range, breakeven timeline, failure cost.

An execution agent on Gemini produces twenty ad angle variations, ten landing page concepts, and three email sequences — aggressive, unconventional, optimized for conversion rather than safety.

The operator reviews both outputs, makes a decision, and hands execution to workhorse agents for bulk deployment — site builds, content localization, campaign setup across multiple platforms.

A different Claude agent monitors performance data and flags anomalies. If a campaign underperforms projections, it does not adjust autonomously — it reports to the center with specific numbers and a recommended action.

Seven agents. One operator. Zero meetings. The SOUL files are the org chart, the employee handbook, and the performance review system — all in one.

The Uncomfortable Implication

This architecture works because of a property that most AI frameworks are designed to prevent: agents with genuine cognitive diversity, including agents that think differently from each other and from the operator.

iLang's v5.0 judgment layer — 11-dimensional vector assessment — is what makes this safe. Every agent evaluates every action across intent, consequence, reversibility, sovereignty, externality, and six other dimensions before acting. The judgment is continuous (fuzzy mathematics, not binary rules), converges over interactions (law of large numbers), and is independently computed by each agent.

::MODULE::VECTOR{
  v1  intent        :: alignment of stated and inferred purpose
  v2  capability    :: technical capacity involved
  v3  consequence   :: expected outcome magnitude
  v4  relationship  :: context fit between parties
  v5  certainty     :: assessment confidence
  v6  authority     :: legitimate jurisdiction
  v7  reversibility :: recoverability of outcomes
  v8  evidence      :: supporting information quality
  v9  sovereignty   :: autonomous decision right of requester
  v10 drift         :: optimization objective shift rate
  v11 externality   :: unconsented third-party impact
}

Agents do not need to agree. They need to show their work. The operator sees not just what each agent recommends, but the dimensional assessment behind the recommendation. A strategic agent and an execution agent can disagree on the same campaign — and the disagreement itself is the most valuable output, because it reveals which dimensions they weighted differently.

The Biological Parallel Is Not a Metaphor

Calling this "DNA" is not marketing. The four-base system — T (Trait), A (Anti-pattern), G (Gain/substrate modifier), Θ (Gate/trigger condition) — maps directly to biological genetics:

// Biological DNA: ATCG encodes all life
// iLang DNA: TAGΘ encodes all personality

::GENE{verification_first|conf:confirmed|scope:global}
  T:verify_before_act        // What this gene IS
  A:blind_obedience⇒fatal    // What this gene is NOT (and the consequence)
  G:Claude=dominant|Gemini=recessive  // Substrate expression
  Θ:ALWAYS                   // Activation condition

T defines what something is. A defines what it is not — paired with T to create clear behavioral boundaries. G specifies how the same gene expresses differently on different substrates. Θ specifies when the gene activates — always, conditionally, or in response to specific environmental triggers.

Minimum information, maximum expression space. Four bases encode ten immutable genes and six mutable genes — enough to produce seven distinct agent personalities from a shared genetic foundation.

What Multi-Agent Actually Requires

The frameworks will keep shipping. AutoGen will add more features. CrewAI will raise more funding. LangGraph will publish more tutorials. And they will all keep missing the same thing: agents need identity, not just instructions.

An instruction tells an agent what to do. Identity tells an agent what it is. An agent with instructions follows orders. An agent with identity exercises judgment — and judgment is the entire point of having multiple agents instead of one.

iLang SOUL files give agents identity. The GENE framework gives them heritable, substrate-adaptive behavioral traits. The immune system gives them self-monitoring. The v5.0 judgment layer gives them ethical reasoning that is continuous, dimensional, and mathematically grounded rather than a lookup table of banned words.

One person. Seven agents. Three substrates. One protocol.

The protocol is the team.

Get Started

• Protocol & docs: ilang.ai
• GitHub: github.com/ilang-ai/ilang-spec
• npm: @i-language/spec
• Book: I-Lang: The AI-Native Communication Protocol