I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
There's a serial killer on the loose in Jeju Island, South Korea!!
The discovery of four bodies on Jeju Island in three days stems from police corruption and systemic failure. A local officer under investigation falsely closed roughly 200 missing-person cases over 17 months to inflate closure metrics. Cases were archived as resolved without searches, allowing victims to rot while predators operate unchecked on an isolated island
South Korea has seen this breakdown before with predators like Lee Choon-jae during the Hwaseong murders and Yoo Young-chul in the early 2000s. In those historical cases, lazy police work, lack of cross-jurisdictional communication, and institutional neglect of vulnerable populations enabled prolonged killing sprees. Jeju’s geography amplifies this hazard, as isolated rural terrain, transient resort worker populations, and poor local surveillance create an environment where bodies remain hidden until administrative fraud is exposed
When I visited Jeju, the lack of police presence in rural pockets was glaringly obvious. The primary driver of this systemic failure is perverse institutional incentives. Korean police precincts face intense pressure to show high case clearance rates, creating clear moral hazard where corrupt officers falsify paper records to meet quotas rather than conduct actual field investigations
The first-order effect of this institutional failure is a complete breakdown of public safety and a surge in uninvestigated homicides across the region. When police actively conceal missing persons to pad administrative stats, violent criminals receive free immunity to continue operating without fear of detection
The second-order effect hits the regional economy, threatening major hospitality and travel stocks like $LOTE (Lotte Tour Development), $083540 (Paradise Co), and $035250 (Kangwon Land). Institutional capital will quickly discount Jeju-exposed leisure assets as safety concerns drive away domestic tourists and foreign visitors
Share this analysis with a fellow investor or reply below with your thoughts on how governance failures impact regional leisure assets.
Follow for more deep-dive market insights.
The first autonomous agent cyberattack is an unprecedented event that deserves unprecedented transparency. Today we’re sharing everything we can: a full technical timeline, an interactive replay, and how we used an open model to defend ourselves, so defenders everywhere can learn from it and prepare for what’s next.
https://t.co/uPxIpjW8Xn
/frontend-design, /simplify, /review, and /commit-commands:commit-push-pr are currently my favorite claude skills. Would love to know of any sick ones I'm missing. Also enjoyed the research results of /last30days a few times yesterday.
🧵 THE DEVVE WAKE-UP THREAD — READ THIS AND TRY NOT TO FREAK OUT
Some people still think $Devve is “just another L1.”
Some think it’s “too good to be true.”
Let me be blunt:
If $Devve actually works the way its architecture says, It is not competing with crypto. It is competing with global financial rails.
And nothing else in the industry is even close.
1️⃣ $Devve isn’t an L1. It’s what blockchains were supposed to be.
$ETH? On-chain apps.
$SOL? Speed.
$XRD? Parallelism.
$KTA? Simplicity.
$ATOM? Modular.
$CC? Permissioned interoperability.
$AVAX DAG variants.
$SUI? Parallel execution.
All interesting.
All clever.
All stuck with the same fundamental limit:
> They execute logic on the chain itself.
$Devve does not.
That single design choice is the reason the rest of this thread even exists.
2️⃣ Devve doesn’t execute transactions on-chain. Zero.
Every other chain:
> “When someone swaps, lends, trades, borrows —
our blockchain must execute the logic.”
$Devve:
> “Why would the blockchain execute anything?
Do the work off-chain.
Give me the proof.”
Execution chains run code.
$Devve runs math.
Off-chain execution → on-chain validation.
This instantly explains:
✔ CTS
✔ MIS
✔ T1
✔ global atomicity
✔ deterministic ordering
✔ t=0 finality
✔ insane throughput
When the chain validates proofs instead of running computation, everything else Devve claims stops sounding impossible.
3️⃣ MIS isn’t about being fast — it’s about being ZERO
People say:
“But $SOL is fast.”
“But $SEI is fast.”
“But $KTA settles quick.”
Speed is irrelevant.
MIS isn’t fast — it's literally ZERO.
No pending state.
No exposure.
No reorg risk.
No MEV.
No failed legs.
No rollbacks.
No attack surface.
Because nothing is being executed on the chain.
Markets don’t need “low latency.”
They need no latency — legally, operationally, financially.
That’s the difference between a crypto toy and financial-grade settlement rail.
4️⃣ CTS isn’t an atomic swap — it’s a new primitive
Other chains brag about “atomic swaps.”
Cute.
CTS bundles entire workflows:
Borrow
Swap
Hedge
Transfer
Deliver
Settle
Into ONE contingent group.
Validated as one action.
Traditional chains can’t do this because:
- every step is a contract call
- every step relies on block timing
- failure in one step breaks others
- cross-market execution collapses
$Devve:
“Bundle the whole workflow, prove it, commit it.”
This is not a better atomic swap.
This is a different category of technology.
5️⃣ T1 is what every other sharded chain wishes it had
Sharded chains normally fail because each shard:
has its own timing
its own block schedule
its own execution path
its own ordering
its own failure modes
Devve avoids all of this because:
- shards validate proofs (not execute logic)
- proofs sync through T1
- commit happens at one deterministic instant
This gives Devve global deterministic atomic settlement across independent blockchains.
Nobody else has this.
Nobody else can even fake it.
6️⃣ This is why a major global exchange-tech provider picked Devve
They didn’t pick it for hype.
They didn’t pick it for memes.
They didn’t pick it because it’s “fast.”
They picked it because only one architecture answered their real question:
> “Who can settle across markets, assets, and shards — atomically, instantly, deterministically — at global scale?”
Literally only one architecture qualifies - a blockchain that validates proofs, not executes code.
That’s it.
7️⃣ FINAL WORD — the part most people aren’t ready for
People underestimate Devve because they’re still thinking in 2015 blockchain mental models.
But when MIS clicks… When CTS clicks… When T1 clicks… When people realise this is the only system that can run the workflows of the global economy…
It’s over.
The re-rating will be violent.
The adoption will be reflexive.
The flywheel will be unstoppable.
This isn’t another chain.
This is a replacement for financial infrastructure.
@DevveEcosystem@DevvExchange
🧵 THE THREAD THAT WILL BLOW UP YOUR INSTITUTIONAL BRAIN
ASX, Canton, and WHY DEVVE’S ARCHITECTURE IS A DIFFERENT SPECIES
@CantonNetwork is being talked about everywhere as the institutional chain.
But the @ASX CHESS saga is the perfect real-world case showing why distributed smart-contract systems hit a wall — and why $Devve is fundamentally different.
🔥 1️⃣ FIRST — CONTEXT
@ASX is one of the world’s major market operators.
They tried to rebuild CHESS using:
⚙️ Digital Asset’s Daml
⚙️ Canton’s distributed smart-contract architecture
After ~7 years and $200M+ spent, the project was halted.
There were many reasons, but a major friction point was the integration burden of the architecture.
Here’s where Devve separates from the pack.
💥 2️⃣ CANTON’S MODEL: “EVERYONE REBUILDS THEMSELVES”
$CC requires every participant — brokers, custodians, registries, clearing firms — to:
🔧 run domain nodes
🔧 host Daml execution
🔧 port business logic
🔧 coordinate multi-party workflows
🔧 sync privacy domains
Elegant in theory.
A nightmare in a 100+ participant ecosystem.
ASX participants faced:
💸 huge integration costs
📚 new execution models
🧪 heavy testing
⚠️ multi-party execution risk
They didn’t walk because “blockchain failed.”
They walked because the architecture demanded they rebuild the plumbing.
⚡ 3️⃣ DEVVE IS THE OPPOSITE: “JUST PROVE THE RESULT”
$Devve does not:
✘ run smart contracts
✘ execute business logic
✘ require every participant to run nodes
✘ force rewrites of legacy systems
$Devve says:
“Keep your systems.
Compute off-chain.
Give me the deterministic proof.
I’ll validate and settle atomically.”
Participants simply:
➡️ compute in their existing stack
➡️ generate a deterministic digest
➡️ submit proofs via API
➡️ $Devve validates + commits
This is why $Devve is institutionally lightweight and $CC is institutionally heavy.
🔥 4️⃣ THE ASX LESSON: ARCHITECTURE DETERMINES SURVIVABILITY
You cannot ask every major participant to rebuild their execution model.
Too costly.
Too complex.
Too hard to coordinate.
$CC decentralizes execution — the hardest thing to decentralize.
$Devve decentralizes validation — the easiest.
This creates massive advantages:
✔ integration drops from $20–50M → ~$1M
✔ testing collapses in complexity
✔ performance is predictable
✔ regulators immediately understand it
✔ existing systems stay intact
🚀 5️⃣ PERFORMANCE: NOT EVEN CLOSE
$CC executes logic across domains, leading to:
⏳ variable latency
⏳ synchronization overhead
⏳ non-deterministic timing
$Devve’s validation model gives:
💠 deterministic ordering
💠 synchronized commit
💠 zero reorgs
💠 zero MEV
💠 atomic multi-shard settlement
💠 near-instant finality
This is market-structure-grade deterministic settlement.
🏛 6️⃣ REGULATORS PREFER THE “MINIMUM DISRUPTION” MODEL
$CC forces regulators to learn:
➜ Daml semantics
➜ distributed execution
➜ workflow logic
$Devve’s model is simple:
“All business logic stays where it already sits.
The chain only validates proofs.”
It’s the least disruptive path to atomic global settlement.
🌋 7️⃣ WHY A MAJOR EXCHANGE-TECH PROVIDER CHOSE DEVVE
Not for hype.
Not for tokenomics.
But because:
💥 distributed execution is too heavy
💥 rewriting business logic is a non-starter
💥 deterministic global settlement is mandatory
💥 cross-shard atomicity is essential
💥 validation > execution
$CC = “rebuild everything.”
$Devve = “keep everything; just settle smarter.”
That’s the architectural difference institutions choose every time.
⭐ FINAL WORD
Canton is brilliant.
But CHESS showed how heavy distributed execution becomes at scale.
Devve is built differently — institutionally, pragmatically, realistically.
A chain that doesn’t ask markets to change themselves.
Only to prove their actions and settle atomically.
That’s why $Devve’s architecture isn’t just “different.”
It’s inevitable.
@DevveEcosystem@DevvExchange
$1M protection sounds good, but there are some details that people miss.
It clearly says:
- coverage for eligible customers
- once active
- subject to maximum policy limits
And @marczeller mentioned that the coverage is provided by Aave Labs via regulated subsidiary.
Because Aave Labs can’t just say “we provide coverage” out of thin air.
To do that legally, they need a company that follows the rules:
- regulated by a financial authority
- allowed to offer protection/insurance-like services
- required to hold reserves and follow compliance rules
We’ll see the full details once it goes live, but even with more limited protection, this could be a game changer.
And don’t forget that there’s a second layer of defense called Umbrella.
If bad debt shows up, this kicks in first to cover losses, giving lenders an extra buffer before they’re affected too.
For the first time, I actually feel confident recommending crypto products to friends.
After years of building quietly behind the scenes, we’re finally reaching a point where we can offer products that truly meet mass demand and adoption.
Bullish $AAVE!