1  # ABZU: Mesh Service Credits
  2  >
  3  > A utility token designed for scrutiny
  4  
  5  ## Philosophy: What ABZU Is
  6  
  7  **ABZU is not an investment. It is a service credit.**
  8  
  9  Like cloud computing credits, transit passes, or prepaid phone minutes — you acquire ABZU to **use mesh services**, not to hold and hope for appreciation.
 10  
 11  | ABZU Is | ABZU Is Not |
 12  |---------|-------------|
 13  | Service credits | Investment vehicle |
 14  | Consumed on use | Held for appreciation |
 15  | Utility-priced | Speculation-priced |
 16  | Spent immediately | Staked for yield |
 17  
 18  ---
 19  
 20  ## Anti-Ponzi Design Principles
 21  
 22  ### ❌ Things We Explicitly Do NOT Do
 23  
 24  | Anti-Pattern | Why It's Dangerous | Our Approach |
 25  |--------------|-------------------|--------------|
 26  | Staking yields | Paid from new deposits = Ponzi | No yield, only slashing |
 27  | Emissions to holders | Rewards holding over using | No emissions |
 28  | Early investor discounts | Late joiners subsidize early | Everyone pays market price |
 29  | Team token dumps | Insider enrichment | Team paid in fiat/services |
 30  | Foundation buybacks | Market manipulation | No buybacks ever |
 31  | Lockup bonuses | Creates dump events | No lockups |
 32  | Deflationary burns | Artificial scarcity games | No burns |
 33  
 34  ### ✅ Things We Explicitly DO
 35  
 36  | Pattern | Why It's Safe | Implementation |
 37  |---------|--------------|----------------|
 38  | Fixed supply | Predictable, no inflation tricks | 1T ABZU forever |
 39  | Utility pricing | Value = services received | 1 ABZU = X compute |
 40  | Consumption model | Spent, not held | Credits used for API calls |
 41  | Reputation bonds | Skin in the game, no yield | Lock to advertise, slashable |
 42  | Market price acquisition | Fair for all participants | Buy at prevailing rate |
 43  
 44  ---
 45  
 46  ## Token Economics
 47  
 48  ### Supply: Fixed Forever
 49  
 50  ```
 51  Total supply: 1,000,000,000,000 ABZU (1 trillion)
 52  Decimals: 18
 53  Smallest unit: 1 drop (0.000000000000000001 ABZU)
 54  
 55  All tokens exist at genesis. No inflation. No emissions. Ever.
 56  ```
 57  
 58  ### Genesis Allocation
 59  
 60  | Recipient | Amount | Purpose | Restrictions |
 61  |-----------|--------|---------|--------------|
 62  | Service Reserve | 500B (50%) | Available for users to acquire | Sold via bonding curve Year 1 |
 63  | Foundation Operations | 200B (20%) | Infrastructure, grants, development | Operations only, never sold speculatively |
 64  | Ecosystem Development | 200B (20%) | Developer grants, partnerships | Disbursed for work delivered |
 65  | Team | 100B (10%) | Founding contributors | 4-year vest, used for services |
 66  
 67  **Critical:** No category exists for "investors" or "early buyers at discount." No ICO, no presale.
 68  
 69  ---
 70  
 71  ## Mechanical Solutions
 72  
 73  ### Problem 1: Cold Start Liquidity
 74  
 75  **Challenge:** Who sets the price on Day 1?
 76  
 77  **Solution: Time-Limited Bonding Curve**
 78  
 79  ```
 80  Year 1 (Bootstrap):
 81  ┌─────────────────────────────────────────────────────────┐
 82  │  Foundation acts as transparent market-maker            │
 83  │                                                         │
 84  │  Price = f(supply_sold)                                 │
 85  │  Starting: 1 ABZU = $0.0001                            │
 86  │  Curve: Linear or sqrt, TBD                            │
 87  │                                                         │
 88  │  Buy: User → Curve → ABZU                              │
 89  │  Sell: ABZU → Curve → User (at curve price)            │
 90  └─────────────────────────────────────────────────────────┘
 91  
 92  Year 2+ (Mature):
 93  ┌─────────────────────────────────────────────────────────┐
 94  │  Foundation exits market-making completely              │
 95  │  Curve disabled, pure peer-to-peer market              │
 96  │  Price = organic supply/demand on DEXs                 │
 97  └─────────────────────────────────────────────────────────┘
 98  ```
 99  
100  **Why this is acceptable:**
101  
102  - Temporary (sunsets after Year 1)
103  - Transparent (curve formula is public)
104  - Non-manipulative (Foundation cannot deviate from curve)
105  - No profit motive (proceeds fund operations)
106  
107  ---
108  
109  ### Problem 2: Velocity (The "Hot Potato" Effect)
110  
111  **Challenge:** If tokens are only for spending, sell pressure is constant.
112  
113  **Solution: Reputation Bonds + Working Capital**
114  
115  **Reputation Bonds (for Providers):**
116  
117  ```
118  To advertise services on the mesh:
119  ┌────────────────────────────────────────────────────────┐
120  │  Provider must lock ABZU as a trust bond               │
121  │                                                        │
122  │  • Lock amount determines visibility/priority          │
123  │  • Earns ZERO yield (not staking for profit)          │
124  │  • Slashed for misbehavior (bad inference, drops)     │
125  │  • Unlocks after 7-day cooldown                       │
126  └────────────────────────────────────────────────────────┘
127  ```
128  
129  **Working Capital (for Agents):**
130  
131  ```
132  Autonomous agents can't buy on-demand per tx.
133  They need a "float" to operate.
134  
135  Agent wallet holds working capital:
136  • Enough for ~1000 typical transactions
137  • Refilled periodically via fiat or earned services
138  • Creates natural demand for holding (operational, not speculative)
139  ```
140  
141  **Why this solves velocity:**
142  
143  - Providers MUST hold to be visible → buy pressure
144  - Agents MUST hold to function → buy pressure
145  - Neither is "holding for profit" — it's operational necessity
146  
147  ---
148  
149  ### Problem 3: Sybil Resistance
150  
151  **Challenge:** Without stake requirements, what stops malicious nodes?
152  
153  **Solution: Stake-for-Trust, Not Stake-for-Yield**
154  
155  | Mechanism | Ponzi Staking | ABZU Trust Bonds |
156  |-----------|---------------|------------------|
157  | Lock required? | Yes | Yes |
158  | Yields paid? | Yes (Ponzi) | **No** |
159  | Slashing? | Sometimes | **Always** |
160  | Purpose | Passive income | Skin in the game |
161  
162  **How it works:**
163  
164  ```
165  Node wants to offer compute:
166  1. Lock 1000 ABZU as trust bond
167  2. Bond visible on-chain, affects reputation score
168  3. Provide services, get paid directly per job
169  4. If misbehavior detected:
170     - Bond slashed (sent to ecosystem fund)
171     - Reputation damaged
172     - Must re-bond to continue
173  5. To exit: 7-day unlock cooldown (no instant rugs)
174  ```
175  
176  **What counts as misbehavior:**
177  
178  - PoI verification fails (bad inference)
179  - Storage proofs fail (data not available)
180  - Relay dropped packets (Ghost Mode failures)
181  - DHT unavailability (offline when claiming online)
182  
183  **Why this isn't Ponzi:**
184  
185  - Zero return on locked capital
186  - Only RISK, no passive reward
187  - Honest nodes lose nothing; only bad actors are punished
188  
189  ---
190  
191  ## How Pricing Works
192  
193  **The protocol does not set the price of ABZU.**
194  
195  The protocol defines the **handshake format**. Pricing is peer-to-peer negotiation:
196  
197  ```
198  Node A advertises:
199  {
200    "service": "inference/llama-3-70b",
201    "price": "5 ABZU per 1M tokens",
202    "bond": "5000 ABZU locked",
203    "reputation": 0.98
204  }
205  
206  User B queries:
207  → Sees offers from nodes A, C, D, E
208  → Picks based on price, reputation, bond size
209  → Pays directly to chosen node
210  → Receives PoI-verified inference
211  
212  Market price emerges from aggregate micro-agreements.
213  ```
214  
215  ---
216  
217  ## Why This Isn't a Security (Howey Analysis)
218  
219  | Howey Factor | ABZU |
220  |--------------|------|
221  | Investment of money | Money for services, not equity |
222  | Common enterprise | Decentralized network, not company |
223  | Expectation of profits | Expectation of services |
224  | From efforts of others | Self-service, decentralized |
225  
226  **Primary purpose:** CONSUMPTION (buying services)
227  **No promise:** Price appreciation
228  **Buyers are:** Users, not investors
229  
230  ---
231  
232  ## Why This Isn't a Ponzi
233  
234  | Ponzi Characteristic | ABZU |
235  |---------------------|------|
236  | Early investors paid by later | ❌ Users buy services, not shares |
237  | Requires constant new money | ❌ Works even with zero growth |
238  | Promises returns | ❌ Only promises services |
239  | Collapse without growth | ❌ Fewer users = cheaper services |
240  | Yield from nowhere | ❌ No yield at all |
241  
242  ---
243  
244  ## Long-Term Economics
245  
246  ### Year 1 (Bootstrap)
247  
248  - Bonding curve active for price discovery
249  - Foundation operates core infrastructure
250  - Early providers bond to establish reputation
251  - Developer grants fund ecosystem tools
252  
253  ### Year 2-5 (Maturity)
254  
255  - Bonding curve sunsets
256  - Price floats on open market
257  - Healthy velocity: user → provider → user
258  - Reputation system has meaningful history
259  
260  ### Year 10+ (Steady State)
261  
262  - All 1T ABZU in circulation
263  - Foundation credits depleted (working as intended)
264  - Pure peer-to-peer service market
265  - Token is boring utility, not speculation
266  
267  ---
268  
269  ## Summary
270  
271  ABZU is designed to be **boring and defensible**:
272  
273  | Principle | Implementation |
274  |-----------|----------------|
275  | Simple | Buy credits, use services, pay providers |
276  | Fair | Same terms for everyone, no early advantage |
277  | Stable | Fixed supply, consumption-driven value |
278  | Legal | Utility token, not security |
279  | Honest | No yield, no appreciation promises |
280  | Robust | Reputation bonds, slashing, PoI verification |
281  
282  **The goal is not number-go-up. The goal is a functioning service economy for humans and agents.**
283  
284  ---
285  
286  *Draft v0.3 — 2026-01-31*
287  *Incorporates cold-start, velocity, and Sybil solutions*
288  
289  ---
290  
291  ## Implementation Reference
292  
293  The token system is implemented in the `abzu-token` crate and integrated into the Abzu mesh.
294  
295  ### Core Components (`abzu-token`)
296  
297  - **`Balance`**: 18-decimal fixed-point arithmetic (u128 drops). Safe against overflow/underflow.
298  - **`Ledger`**: Trait for balance storage. Includes `DhtLedger` (distributed) and `MemoryLedger` (testing).
299  - **`ReputationBond`**: State machine for provider bonds (Active -> Cooldown -> Unlocked). Enforces 7-day unbonding period.
300  - **`ServiceOffer`**: Struct for advertising services (type, price, bond size) in the marketplace.
301  - **`BondingCurve`**: Algorithmic price discovery for Year 1 (Square Root curve).
302  
303  ### Integrations
304  
305  - **DHT**: `TokenBalance`, `TokenBond`, and `ServiceOffer` are first-class `ValueType` variants in `abzu-dht`.
306  - **Agent**: The `sovereign-agent` exposes HTTP endpoints (`/api/token/*`) for agents to query balances, place offers, and interact with the curve.