cereal_killer

Grateful to have my perspective included in a @CoinDesk article for the first time. 🙏 Shared some thoughts on how L1/L2 DeFi dynamics are evolving. Full piece here: coindesk.com/tech/2025/09/1…

@chameleon_jeff @0xSynthesis1 @HyperliquidPC 🔥

I spent the past few days in Washington with @hyperliquidpc meeting with policymakers during the historic advancement of the Clarity Act. We discussed Hyperliquid, the benefits that it offers to American consumers, and the regulatory path to bring onchain derivatives markets into the United States. Some conversations were technical with an impressive baseline understanding of Hyperliquid. Discussions included how onchain trading is a financial innovation that has clear global user demand. Other conversations focused more on a first principles introduction to defi and the promise of onchain markets. It was encouraging to see bipartisan support for thoughtful regulation of crypto. I look forward to continuing discussions in DC and working hard to make American access to Hyperliquid a reality.

@SuhailKakar oh that's really cool thank you!

@FBitach mute keywords are matched via tokenization (not raw string), so casing, punctuation, and hashtags get normalized

x has made their algorithm open source and it's fucking incredible - one of the best social algos i've seen here's exactly how it decides what shows up in your feed (and how to use it) > it watches your last ~127 engagements and feeds them to a transformer. > your behavior is the main signal. demographics (country, language, age, gender) ride alongside but aren't the driver. > no bio matching. no hand-coded interests. no keyword targeting. > it predicts 18 distinct things you might do with each post. > positive signals: likes, replies, reposts, quotes, dwells, shares, follows. > negative signals: blocks, mutes, reports, "not interested," fast scrolls. > dwelling on a post counts as engagement. you don't need to like it. > blocks and reports have heavy negative weights in the score. > in-network posts (from people you follow) get a built-in boost over out-of-network. > brand new accounts get routed to a different model entirely - more exploration, less personalization. how to land on people's For You > originality wins. retweet dedup means viewers see a tweet ONCE, even if 10 mutuals RT it. > virality = reaching new networks, not your bubble re-amplifying. > don't post 5 tweets in 10 mins. > there's a real author-diversity penalty that knocks down each repeat post from the same author in a viewer's feed. > you're literally competing with yourself for slots. > don't dump 12-tweet threads expecting them all to land in FYP. > conversation dedup: only your highest-scored post in a thread shows per viewer on For You. > threads still work fine for followers reading them natively - just don't expect all 12 to surface to strangers. > pick your strongest tweet. let it carry. > fresh content always wins - there's a hard age cutoff on FYP eligibility. > no real slow burn. it hits within its window or dies. > in videos, what you SAY matters. they transcribe audio via FFmpeg + ASR and embed the transcript. > bad audio = you lose a real signal that feeds into your post's embedding. > quality replies under big accounts get surfaced. > a Grok model literally scores replies 0–3 under high-follower posts. > good reply game = real distribution. > mutual-follow overlap (Jaccard) between you and the viewer is computed per post - building inside a community helps. what to avoid > engagement bait gets flagged. there's a literal "slop_score" classifier (a Grok VLM judging your post on creation). > "RT if you agree" tier = death. > rage bait that triggers blocks/reports is the worst possible play. > a single report carries serious negative weight in the score. > spam replies in low-follower convos get caught by a dedicated spam classifier. > reply-guying with copy-paste lines across the network = flagged. > NSFW, gore, hate, violence get routed through 7-category PTOS safety classifiers. > doesn't matter how viral - flagged content gets dropped, blurred, or age-gated. > posting too fast = author diversity penalty. > posting too often in one thread = conversation dedup. how to fix your own feed > engage with intent. every action you take is a training label for your personal model. > scrolling past fast is itself a negative signal (negative_not_dwelled_count is a real input). > use "not interested" aggressively. it's a real negative weight in future rankings. > mute keywords work. tokenized, so "crypto" catches "cryptocurrency." > snooze topics = hard filter. zero exceptions. > follow new accounts to inject new candidates into your retrieval pool. > there's a built-in diversity reranker (DPP) that prevents samey feeds. > if your feed still feels samey, you trained it to be. > like = lighter touch. reply / repost / quote / dwell = stronger. > if you want more of something, don't just like it. interact with it.

@briwhoze 🔥

job hunting is like dating: market urself, ghosting, mixed signals, rejection repeat and keep going

@0xCryptoSam @recvcx Congrats!!

I’m excited to announce that I’m joining Reciprocal as an investor! (@recvcx) This is certainly an interesting spot to enter crypto venture. For the second time in its relatively brief history, DeFi is at an inflection point. While I have some ideas, I’m not entirely sure what it will look like next. What I do know is that each cycle, infrastructure improves, crypto protocols get better, and conviction hardens. We’ve had more products go mainstream in the last couple years than ever before. That is what matters. I’m excited to take on the mission of finding, investing, and supporting the next era of great crypto companies with an incredibly talented team. Onwards!

@ahbeaudry @TokenRelations Congrats!🍾🎈🎉🎊

I'm excited to announce that I’ve joined the team over at @TokenRelations as an Institutional Crypto Researcher! When I got laid off from Messari back in March, I was bummed not only because I was now unemployed, but because I really liked research. I had a passion for writing about developing markets, stablecoin and crypto adoption abroad, and learning how firms were using tokenization in their products. The team at Token Relations is incredible and feels very much like a young Messari; everyone is smart, driven, and hungry to make a name for themselves. I want to also give a shoutout to the new Commander in Chief @jacqmelinek for her tireless energy, clear vision on how to run a team, and constant support for new ideas and innovative research. If anyone is looking for top tier reporting or wants to partner on putting out commissioned research, my DM’s are always open :)

What happens to financial privacy in an AI-native world? Zcash offers one answer: optional privacy, with selective disclosure when needed. Privacy isn’t a feature. It’s infrastructure. Read the $ZEC research report from Grayscale Research: research.grayscale.com/reports/zcash-…

20 crypto projects raised $202 million this week here are the top 5 👇 @arq_finance - $60M > LATAM-based stablecoin financial app serving 2+ million customers > @sequoia & @foundersfund @crossover_mkts - $31M > An electronic communication network (ECN) for crypto trading > @Tradeweb, @wintermute_t, DRW Venture Capital @QFEX - $9.5M > Exchange offering 24/7 trading of RWAs, with up to 50x leverage > @generalcatalyst, @yuris, & @ycombinator @Cyclops_inc - $9.5M > Platform enabling payment service providers (PSPs) to offer crypto and stablecoin capabilities > @CastleIslandVC, @FPrimeCapital, & @Shift4 @utexocom - $7.5M > Bringing $USDT natively to Bitcoin > @BigBrainVC, @tether, & @PortalVentures

The result is programmable confidentiality without sacrificing composability or liquidity. Mainnet is live. Confidential stablecoin transfers have already happened. The ZAMA token powers fees and staking tied to actual protocol usage. Full report here: messari.io/report/underst…

Execution stays on the host chain. Heavy encrypted computation is handled by an offchain coprocessor network. Threshold MPC manages keys so no single party can decrypt unilaterally.

Public blockchains gave us radical transparency. But finance does not run on full transparency. Balances, positions, bids, strategy. Some data cannot be public. That tension has capped real adoption. @zama is tackling this head-on. 🧵

17 crypto projects raised $269 million this week here are the top 5 👇 @whop - $200M at a $1.6B valuation > A marketplace that lets creators sell digital products > @tether @stsdigital_io - $30M > An institutional digital asset trading firm > @CMT_Digital, @strobefund, & @Arrington_Cap @jpyc_official - $11.9M > A regulated Japanese yen stablecoin > @Asteria_ir, QR Investment, & directX Ventures @t54ai - $5M > An AI governance and operational assurance solutions provider > @anagramxyz, @FTDA_US, @plcapital @blupryntco - $4.25M > A compliance tech company streamlining digital asset disclosures > @valorcapgroup, @cbventures, & @QuonaCapital

🔴 The @Optimism Superchain quietly became one of the most used execution layers in crypto. Here’s what actually happened in H2 2025 👇 🔴 The Optimism Superchain scaled faster than most people realized. In H2 alone • 3.6B transactions • 12.7% of all onchain activity • More than 50% L2 market share by transactions • 415M in app revenue This growth did not come from a single chain but from a network of OP Stack chains operating as a shared ecosystem. 🔴 The value capture story also shifted. Apps across the Superchain generated roughly 10x more revenue than sequencers, showing that value is increasingly accruing at the application layer rather than solely at the infrastructure level. 🔴 Adoption also expanded beyond crypto native teams. Major exchanges and institutions (Upbit, OKX) began launching their own chains using the OP Stack, turning it into a default toolkit for customizable L2s tied to real distribution and user bases. This creates shared standards, interoperability across chains, and stronger network effects across the ecosystem. In the H2 report, I broke down • Chain level growth • App revenue leaders • Transaction and usage trends • Institutional OP Stack adoption • Where Superchain network effects are forming next Full H2 2025 report 👇 messari.io/report/state-o…

Free report covering this week's top fundraising deals, DATs, M&As, newly launched funds, VC investor insights, and more: messari.io/report/crypto-…

18 crypto projects raised $62 million this week Here are the top 5👇 @InferenceAI - $20M > A quant trading firm leveraging AI across digital assets > Avenir Group @aveniaio - $17M > Stablecoin payment rails for LATAM and global markets > @QuonaCapital, Big_Bets & @sequoia @LevlFi - $7M > Stablecoin infrastructure for fintechs > @galaxyhq, @CMCC_Global, & @ProtagonistXYZ @SupersetFinance - $4M > Unified liquidity execution layer for stablecoins, tokenized deposits, and onchain FX > @7RIDGE1, @Exponential_Sci, @OrangeDAOxyz @BirchHill_io- $2.5M > Institutional onchain credit infrastructure > @paraficapital, @CastleIslandVC, & @nascent

For more information, read the full report here for FREE: messari.io/report/state-o…

Active loans across Superchain surged 95.8% YoY to $1.95B. @Morpho alone went from $162M to $1.19B. The cbBTC/USDC vault became Base's largest lending market at $1.07B (Double the size of Ethereum's equivalent.)

The Superchain just hit 3.6 BILLION transactions in H2 2025. That's 44% growth since June 2025. But here's what everyone's missing 👇