Pretty much everyone I know agrees that IC’s tech is top-tier—I’d even call it close to perfect, offering the strongest Web3 stack around. But when it comes to its tokenomics, you’ll almost never hear praise. In this vibe-post, I want to dig into why that is and get to the root of the problem.
Symmetric games are those where players are effectively interchangeable: they have the same strategy sets, face the same payoff functions, and outcomes depend only on what’s played—not who plays it. Because the incentives are aligned across identities, equilibria tend to be cleaner and easier to reason about; best responses look similar for everyone. In crypto terms, systems where most participants face the same basic trade‑off (e.g., hold vs. stake under similar rules) behave more like symmetric games, which often means fewer opportunities for one group to systematically extract value from another.
Asymmetric games introduce distinct roles—different strategy sets, information, timing, or constraints—so identity matters. Classic examples include leader–follower (Stackelberg) or principal–agent settings, where power, information, or liquidity isn’t shared evenly. These designs can be efficient if the rule‑set (governance) neutralizes imbalances; otherwise they invite rent extraction, reflexivity, and misaligned risk/return. That’s why a protocol’s governance (e.g., an on‑chain system like the NNS) must act as a referee: calibrate rewards, selection, and constraints so no role wins by imposing externalities on everyone else. The posts below apply this lens to compare BTC/ETH’s mostly symmetric setups with ICP’s asymmetric roles and to map the DeFi split and its associated risk loops.
ICP’s Incentive Design Is an Asymmetric Game (ETH/BTC aren’t) — and the NNS’s #1 Job Is to Referee It
TL;DR
-
BTC & ETH: mostly symmetric incentives for those who earn rewards (BTC miners, ETH stakers).
-
ICP: asymmetric incentives across liquid holders, long stakers, and node providers — with node providers enjoying ~“cash‑like,” high‑ROI payouts and zero lockup, while 8‑year stakers carry the risk and the illiquidity.
-
The network is over-provisioned (~1,400 nodes); by most practical estimates you could remove ~⅔ with little impact. Those savings alone could fund ~50 SNSes or ~120 developers.
-
NNS exists to regulate this asymmetry. If it doesn’t, the game tilts and long-term participants exit.
BTC & ETH: Mostly Symmetric
-
Bitcoin: From a reward-creation lens it’s simple: one active role — miners — compete under the same rules (hashrate in, block rewards out). Holders don’t get protocol yield. The “payoff function” among miners is symmetric.
-
Ethereum: Two choices most people face — hold liquid ETH or stake ETH for a modest, floating APR (≈ low single digits). Strategy sets and payoffs are broadly similar across participants; timing differs, but no privileged third role siphons value. This is close to a symmetric game.
ICP: A Deliberately Asymmetric Game — and It’s Unbalanced
Roles & payoffs (summary):
-
Liquid ICP — free to trade.
-
Staked ICP — e.g., ~7% for 6 months; ~14% headline for 8 years. In USD terms, the average outcome for new stakers post‑genesis at average entry prices has been closer to ~4%/yr; some see ~0.5%, a few get ~14% only if they bought the bottom. Pre‑launch holders may realize somewhat higher effective returns because they waited years and didn’t sell.
-
Node providers — roughly $1,500/month per node paid in ICP, translating to ~90% USD ROI/year by common estimates. No lockup. Positions are not typically awarded via open, public processes; they’re selected internally. The slot acts like an “NFT‑style license”: you can resell the position for roughly what you paid while pocketing the high, ongoing ROI.
Lock‑in asymmetry:
-
8‑year neuron holders shoulder multi‑year illiquidity and path risk. Even simplifying the mechanics: you can’t just exit; accessing principal is slow and staged.
-
Node providers face almost none of that. They have liquid, monthly ICP, which many sell to cover costs — a structural sell‑pressure on ICP that long stakers can’t hedge.
Over‑capacity amplifies it:
-
The IC runs on ~1,400 nodes, but the set is largely underutilized. The network would function with ~⅔ fewer nodes.
-
Opportunity cost: trimming the excess could fund on the order of ~50 SNS launches or ~120 full‑time developers. Today, that budget mostly flows to node providers — including nodes that aren’t mission‑critical.
This is textbook role asymmetry + information/power asymmetry: different players have different payoffs and different access to selection and liquidity.
Governance Reality: Who Sets the Payoffs?
-
NNS is supposed to set the rules (rewards, capacity, selection) for fairness.
-
In practice, the community lacks complete, timely information, and Dfinity holds decisive voting power, which means reward schedules and node selection are effectively centralized decisions.
-
Large, long‑lock stakers (8‑year neurons) have high exposure but low control. That’s not a healthy equilibrium.
Core point: The main job of the NNS is to regulate this asymmetric game so it doesn’t drift into rent extraction. If it fails, the system penalizes patience and rewards privileged roles.
The Return Math (why this feels unfair)
-
Stakers’ USD outcome ≈ (staking APR in ICP) ± (price effect) − (dilution/sell‑pressure).
- At average post‑genesis entries, that’s landing around ~4% USD for many, lower (~0.5%) for top‑tick entrants, higher (~14%) only for bottom buyers.
-
Node providers’ USD outcome ≈ high, near‑cashflow yield (≈ ~90% USD by common estimates), unlocked, resalable slot, with a monthly stream they can sell immediately.
That gap — ~90% vs ~4% — is the imbalance long-term participants feel every day.
What “Fair Refereeing” by the NNS Should Look Like
Capacity & utilization
-
Right‑size the node set to demand (remove roughly ~⅔ if utilization data supports it).
-
Tie NP rewards to utilization (utilization‑indexed formula) so over‑capacity stops being rewarded.
Selection & accountability
-
Open, competitive onboarding for node providers (public criteria, queue, auctions or reverse auctions).
-
Transparent monthly dashboards: node counts, utilization, minted ICP to NPs, realized ROI, geographical diversity, downtime.
-
Bond/lock for NP slots (or vest rewards) to mirror the spirit of staker lockups; slash for downtime or misbehavior.
Payout design
-
Cap NP USD ROI within a reasonable corridor; route the excess to a developer/SNS fund automatically.
-
Consider dual‑indexing NP rewards to (a) capacity utilization and (b) ICP price bands, dampening forced sell‑pressure during downtrends.
-
Cycle‑denominated budgeting: align NP compensation more tightly to real demand (compute consumed), not just provisioned supply.
Staker alignment
-
Publish a clear glidepath: how staker APRs, node rewards, and capacity scale over time. No surprises.
-
Make it explicit that long‑term lockup should earn a premium net of NP sell‑pressure — otherwise the social contract breaks.
Governance checks
-
Supermajority or non‑Dfinity quorum for economic parameter changes.
-
Sunshine rule: no changes to NP rewards/selection without a public impact note (who benefits, who pays, utilization effect).
Why This Matters
If a node slot is a liquid, resellable, high‑yield license while an 8‑year neuron is illiquid, low‑yield exposure, capital will chase licenses, not governance. That’s how ecosystems hollow out.
Fix the asymmetry, and ICP can fund builders instead of idle capacity. Leave it as is, and long-term stakers will do the rational thing: shorten duration or leave.
Open Questions for the Community
-
What utilization threshold should trigger automatic capacity cuts or reward scaling?
-
What’s a reasonable USD ROI cap for NPs that still attracts reliable operators (e.g., 15–25% rather than ~90%)?
-
Which metrics should the NNS publish monthly to rebuild trust?
-
Should NP slots require a bond/lock or a cool‑down before resale?
Bottom line: ICP’s design is an asymmetric game. That’s fine — if the NNS actually referees it. Right now, the incentives look tilted: node providers win; long stakers carry the bag. Let’s re-balance capacity, open up selection, and put excess rewards to work funding SNSes and developers — the people who create demand.
ICP DeFi is splitting into two as well, serving the different asymmetric game roles.
The reflexive loop (high‑level risk model)
Setup:
-
USD‑exposed actor borrows ICP from a lending market (USD/stable used as collateral).
-
Deposits ICP into liquid staking to earn staking yield; uses yield to offset borrow costs.
Flywheel:
3) As long as borrow capacity exists, the actor scales: more borrowed ICP → more LSD yield → covers more fees → builds a bigger staked ICP position. The faster the unlocks - lower minimum dissolve delay, the better it works.
Pivot & pressure:
4) When borrow capacity tightens or funding turns, actor unwinds:
-
Hedge/flip to perp shorts, convert stakedICP to ICP (spot or via dissolving schedule), sell & push price down, and profit on shorts. Perp rails for ICP are readily available.
-
Retail lenders often don’t yank liquidity at −50%; they ride it out.
Second‑order effect:
5) As ICP price falls, NP payouts mint more ICP to reach the same fiat/XDR value → extra sell supply (NPs typically monetize to cover opex). This reinforces downward pressure and transfers value from long stakers/lenders to USD‑exposed operators.
This scenario won’t be as powerful in symmetric systems like ETH and BTC. Lending ICP is dangerous because of the role asymmetry.