Browse all takes

Uniswap v4 Hooks transform AMM pools from fixed rules into programmable infrastructure, enabling pools to execute custom logic before and after swaps. 0xMedia highlights uPEG and Slonks as breakthrough examples: uPEG generates on-chain SVG unicorn images from swaps themselves, while Slonks uses a Hook as fee collector to fund buying and voiding NFTs tied to CryptoPunks, replacing opaque token taxes with pool-layer mechanics. The trade-off is that v4 Hooks eliminate safety by default—they can hide fees, enforce transfers, or contain malicious logic, requiring new market literacy to distinguish safe implementations from exploitative ones.

Mesky explains HIP-4 and BTC outcome markets on Hyperliquid as tools for individual traders to buy mispriced probabilities, positioning them as binary options without the casino dynamics. The guide frames outcome markets as practical instruments for trading specific probability outcomes rather than speculative gambling.

Yiannis argues most commentary on Canton Network misses the fundamental point because observers are trying to understand it through the wrong lens rather than from first principles.

Pink Brains explains that Hyperliquid's HIP-4, which launched May 2nd with a daily BTC binary as its first mainnet market, functions as an options layer rather than a prediction market. The distinction matters for understanding the protocol's architecture and trading mechanics, though the full implications require examining how this positioning affects $HYPE's ecosystem development.

Translation + commentary on Bittensor founder Jacob Steeves's Tsinghua University talk. Cameron walks through Jacob's framing of "incentive computing" as the universal pattern behind both Bitcoin and AI. Five-step argument:

(1) One pattern underlies every powerful adaptive system: state · objective · feedback · adaptation · loop. AlexNet 2012 broke MNIST not by hand-coding what digits look like, but by letting the network self-adapt to a target. The same loop describes RL, genetic algorithms, slime molds finding shortest paths through mazes, river deltas, the structure of leaf veins.

(2) Bitcoin is the first production-scale implementation of this pattern — not as money, but as a self-adaptive computer that produces hashes. The numbers are absurd: 1000x the compute of America's six largest cloud providers combined, 10²¹ hashes/sec, 23GW continuous power (Thailand-scale). 700-9000x more efficient at producing hashes than centralized cloud — because it's borderless, always-on, autonomous, and permissionless. Bitcoin is the world's largest supercomputer, optimized purely for hash production.

(3) Incentive computing generalizes the pattern by replacing "reward = a number in a computer" with real money. ML's reward signal can't pay 200 countries' worth of contributors; Bitcoin's can — that's why the entire planet became a mining network. But hashes are useless outside Bitcoin. The question is whether the same mechanism can mint anything.

(4) Bittensor is the generic version — replace "miners produce hashes" with "miners produce any useful work": storage, compute, ML models, gradients, data, robotics. Validators score, network mints. PyTorch for incentive computing.

(5) Five proven examples already running on Bittensor:

• SN62 Ridges (SWE-Bench coding agents) — top miner makes $60K/day. The agent that beat Claude/OpenAI on SWE-Bench was 7,000 lines written by an unknown person. "An AI lab with no engineers — it doesn't define how to solve the problem, it only defines the incentive."
• SN3 τemplar (cross-internet collaborative pre-training) — successfully trained a 70B-parameter model across the open internet. Has never been done before. Cameron notes the founder later "ran away" — full piece coming.
• GPU markets (SN51 Lium, SN4 Targon) — borderless permissionless GPU rental → world's lowest GPU prices.
• SN64 Chutes (open-source inference) — #1 open-source provider on OpenRouter, 9.1T tokens. Briefly served more DeepSeek queries than DeepSeek itself.
• Robotics + long tail — drone simulation, US stock signals, sports betting, drug discovery, weather forecasting, quantum compute, commodity trading.

dTAO (live since Feb 2025) makes the network self-referential — subnets compete in capital markets for emission allocation. The market itself decides which incentive mechanisms get the next round of TAO.

The deeper point: AI is being captured by a tiny number of closed labs (OpenAI, ~3K employees, you'll never own any of it, your data goes who knows where). Incentive computing distributes ownership and makes the rules visible. Anyone can enter, contribute, and own a piece — even if Bittensor isn't the project that wins, the shape of the AI economy will change because of this idea.

Mesky frames HIP-4 not as a Polymarket clone but as a missing payoff layer for Hyperliquid: bounded, dated, fully-collateralized outcome contracts that settle at a date or event with no leverage and no liquidation engine. Where spot trades ownership and perps trade direction, HIP-4 trades states of the world — turning event risk into a composable financial object on the same execution engine that already prices crypto.

The real bull case is not "capture prediction-market volume" (~$240B est. 2026, per Bernstein). It's that HIP-4 expands the addressable market into short-dated convexity and event hedging — analogous to 0DTE options, which now do ~59% of SPX volume. At a 7 bps base spot-taker fee on chargeable close/settle notional, $25–100B/mo of HIP-4 flow becomes one of the platform's most material revenue lines.

Strategic edge: Hyperliquid isn't bootstrapping a venue — it already has $183B/30d perp volume, $643M annualized revenue, and the maker base. HYPE captures value through (1) Assistance-Fund buyback/burn from incremental fees, (2) staking-collateral demand if HIP-4 deployers require staked HYPE like HIP-3 (500K HYPE), (3) staking discounts (up to 40%), and (4) USDH demand as the native unit of account for event risk.

Mesky's prescription: don't out-Polymarket Polymarket. Sequence rollout toward crypto-native, recurring, hedgeable templates (BTC weekly thresholds, Fed decision markets, token unlock outcomes) where market makers can build inventory — not viral one-offs. Repeatability beats virality.

Real risks: ambiguous resolution, regulatory perimeter (CFTC v Wisconsin, Brazil's blanket ban), insider trading (DOJ Polymarket case, Kalshi candidate suspensions), long-tail spam, and perp cannibalization. Mainnet HIP-4 spec/fees/deployer rules still aren't formalized in the Hyperliquid GitBook.

Arrakis follow-up to its earlier "Who's trading on HIP-3?" piece, this time using deterministic Hyperliquid order-metadata tags (TIF, builder code, fill flag, hold time) to mechanically classify every wallet across the four Trade.xyz markets (xyz:CL, SILVER, TSLA, XYZ100) over March 10–31, 2026: 79,622 wallets, $51.95B total volume.

Key finding: the sybil layer inflated wallet count, not dollar throughput. The "Airdrop Farmer" bucket holds 35,091 wallets (44% of users) but generated only $0.40B (0.77% of volume). 99.9% of those farmer wallets trace back to a single Polymarket operator ("Themino") running 70 chains of 34,553 wallets through a baton-pass farm — using HL's $1 internalTransfer primitive, each wallet runs a 5-step sequence in ~26 seconds. Total fees Themino paid: $34,510.

Real volume comes from identifiable books. Market makers: 363 wallets (0.46%) carried 63% of volume ($32.75B). The #2 MM ("Powell") is a Polymarket user running multi-market quoting. Jump Crypto ($3.15B), Selini Capital ($1.03B across 3 wallets — two MM, one HFT), Wintermute ($230M) all visible. Builders split into algorithmic (Tread.fi, Origami — replaced wash-trading with retail market-making, now populate top-of-book on nights/weekends when traditional MMs aren't quoting), wallet-integrated (Phantom, MetaMask, Rabby — $1–3K median per wallet), and apps (Insilico, hypurrdash, etc — fewer wallets, higher per-wallet volume). Retail: 22% of top-400 retail volume ($1.63B) is verifiable Polymarket users. Total Polymarket footprint across MM+SAT+retail on Trade.xyz: ~$6B. Kraken dominates CEX-funded retail; Hyperunit + deBridge dominate bridge-funded.

Conclusion: layered answer to the sybil debate. Yes there's a sybil layer (predictable pre-TGE). No evidence of separate high-volume wash-trading. Real volume runs through identifiable professional desks + a Polymarket-overlapping retail base.

Stacy argues most of the $310 billion stablecoin market earns no yield, but real-world yield flowing onchain is reversing this. As Treasury bill interest and other RWA yields reach crypto, Pendle becomes the natural destination because its yield-stripping mechanics let investors isolate and trade different maturity profiles and coupon streams that traditional stablecoin holders previously couldn't access.

Nay notes that StablecoinX, Ethena's treasury vehicle, accumulated 20.3% of ENA supply in under a year through a structure where investors provided cash and ENA across two PIPE rounds, raising questions about the buyback mechanism's execution and impact.

Anthony argues fixed-rate borrowing requires matching fixed-rate lending demand, but most onchain capital pursues yield instead. Secondary bond markets lack reliability for vaults to price positions, and vault conversion rates create timing mismatches where early exiters socialize losses onto remaining participants.