ICGold update: staking rewards end, the 42% becomes ecosystem rewards (daily miners' draw, weekly trading competition), plus ICG games
Hi everyone,
Six days after launch we made the biggest change to ICGold so far. It is live on mainnet since **October 15, 06:11 UTC**. This post explains what changed, why, what we learned, and everything that was built for it.
[Website](
https://icgprotocol.com) · [Whitepaper v1.2](
https://icgprotocol.com/docs/icgold-whitepaper.pdf)
## Short version
* **Staking no longer earns rewards.** Everything still owed to stakers, **49.97 ICP**, was credited at once and can be claimed now.
* **Every lock can be exited at any time, without any burn.** You can still lock ICG if you want to; it simply earns nothing.
* **The 42% share of every deposit is now "ecosystem rewards"**, paid in ICP to the people who keep ICGold active. Today that is a **daily miners' draw** and a **weekly trading competition**.
* **New: ICG games**, player against player, played with ICG.
* **Unchanged:** the emission schedule (50 ICG per block, halving every 210,000 blocks, under 21M), 40% of every deposit to mining, 15% to the buyback, 3% protocol fee.
## Why staking had to go
Staking paid 42% of every deposit to ICG stakers, weighted by lock length. On paper that rewards long-term holders. In practice, in the first week, it did something else:
* **Concentration.** One address held about three quarters of all staked ICG and received about 73% of all staker rewards.
* **A rebate for the biggest miner.** Whoever both mines and stakes gets part of their own deposit back as staking rewards. With 75% of the stake, mining becomes roughly 30% cheaper for one participant than for everyone else. Cheaper mining means more ICG, which means more stake, which means an even bigger rebate. That loop only goes one way.
* **Not what people came for.** Most participants mine with small amounts. For them, a reward stream that mostly flows to one wallet is not an incentive; it is a reason to leave.
We wrote this risk into the whitepaper before launch as a "structural pressure toward concentration". After six days of real data it was no longer theoretical, so we acted early instead of waiting for it to become permanent.
## What happened to existing stakers
Nobody lost anything:
* The upgrade credited **everything still owed**: the part of the 60-day reward stream not yet paid out, plus the reserve. That was **49.97 ICP**, split by the same weights the stream used. It is claimable right now in the Stake tab.
* **Every lock exits at any time without any burn.** Press "Exit now" and the whole position comes back to your wallet.
* Locks still work for anyone who wants one; they just earn nothing.
## The new 42%: ecosystem rewards
The share is the same, 42% of every deposit, in ICP. What changed is who it rewards. The whitepaper deliberately fixes only the share, not the programs, because they will evolve with the ecosystem and are meant to move under DAO governance later. Today there are two programs, each funded with half:
**1. Daily miners' draw (21%)**
* Deposit at least **1 ICP** into mining in a UTC day (pool or your own miner, added up over the day) and you hold **one ticket** for that day.
* At **00:30 UTC** one ticket is drawn with the Internet Computer's random beacon and **wins the whole pot**.
* One ticket per principal: a big miner has the same chance as a small one.
* A day without tickets carries its pot over to the next day.
* The day's ticket list is public on the minter, and the beacon bytes are published with every draw, so anyone can repeat the pick.
**2. Weekly trading competition (21%)**
* Score = ICG you **bought** from the ICPSwap ICG/ICP pool from Monday 00:00 to Monday 00:00 UTC, **minus** ICG you sold there.
* The week is settled **one day after it ends**: ICG sold to the pool in that day also counts against you, and the score is capped by the ICG you still hold (wallet plus locks).
* **Top 5 share the pot: 40 / 25 / 15 / 12 / 8%.**
* Buying and selling the same ICG scores nothing. Liquidity providers and protocol accounts do not take part.
* Scores come straight from the ICG ledger's record of transfers into and out of the pool.
Prizes are paid automatically in ICP to the winner's wallet. The prizes canister has no function that sends ICP anywhere else, and nobody, the owner included, can choose a winner.
## New: ICG games
Two games, **player against player, never against the house**, played with ICG:
* **Jackpot.** Everyone adds ICG to one pot. Your chance equals your share of the pot. When a second player joins, a 60-second countdown starts; at zero one winner takes the pot. It works exactly like winning a block: your chance is your share.
* **Coin flip.** Offer a stake on heads or tails; whoever matches it takes the other side, and the coin is flipped at once. The winner gets 1.94×.
How it works:
* You deposit ICG into a game balance and play from it. **Withdrawals work at any time**, even when betting is paused.
* Every outcome comes from the Internet Computer's random beacon, requested **only after all bets are locked**. The bytes of every game are public.
* Each finished game keeps **3% for the treasury**; the winner takes the rest. The house never bets, so it never needs a bankroll and cannot win against you.
* You can take a break from betting at any time.
The games are gambling. **18+ only.** Only play with ICG you can afford to lose, and check whether this is allowed where you live.
## What we built for this
* A minter upgrade that ends staking rewards once, credits everything owed at once, books the 42% share for the rewards and records each principal's daily deposits for the draw.
* A **prizes canister** for the daily draw and the weekly competition.
* A **games canister** for Jackpot and Coin flip, with ICRC-21 messages so wallets like OISY show in plain words what each call does.
* The watchdog now reports games and prize figures every day and raises an alarm if the games canister ever holds less ICG than it owes its players.
* New **Prizes** and **Games** tabs in the app.
* Whitepaper **v1.2**.
Before anything reached mainnet we audited every change. The audit found and fixed six issues, the most serious a withdrawal path in the games that could have paid twice after a lost reply more than 24 hours old. All 36 integration tests, including an upgrade from the exact code that was running, a full prize draw and a full game session, pass.
## What we learned
1. **Real data beats models.** The concentration risk was in our whitepaper, but only six days of real deposits showed how fast it compounds.
2. **Reward the many, not the biggest.** A reward that grows with what you already have makes the strong stronger. One ticket per principal, and scores capped by what you actually hold, spread rewards much wider.
3. **Fix the formula, keep the promises.** Everything owed was paid, and every lock was freed without a burn. Changing the rules is fine; taking what people already earned is not.
4. **Fix the share, not the programs.** The whitepaper now fixes only that 42% goes back to the ecosystem. The programs can improve without rewriting the protocol again.
## Verify it yourself
| Canister | ID | Module hash |
|---|---|---|
| Minter | `ntx2e-qqaaa-aaaad-qmhha-cai` | `b1fcbe8875afc3155589f612c3dd2bbbe17a5e025e1a8212ca76ab6497e5b5d4` |
| Prizes | `xzf6k-oiaaa-aaaad-qmj4q-cai` | `94a624ba4975bb4d29e405ca35f617aae7ffc0fe2384dfcd8151e6fa806ae387` |
| Games | `xqgvw-yaaaa-aaaad-qmj5a-cai` | `dd7d8ba16c846c232490219fffb6fba1488f8c99b34168e10187166835241f8f` |
The first daily draw runs tonight at 00:30 UTC. If you mine at least 1 ICP today, you are in it.
Questions and criticism welcome, as always.
Jan
