1M TPS, merklized, EVM, single node with state persisted. One really needs to do a lot of things right to make it possible.

Suppose we have 100 uniswap pools with high contention. FAFO can make each pool run at 10K TPS.

Just realized that the word “fast” is very serial. When you can do one thing at a time, the latency (how fast you are) determines the throughput. But with parallelism, throughput is decoupled from latency. For example, a business that sells 12-year-old whiskey has high latency

Turns out you can just stack MMRs (and MMR-shaped Merkle trees) on top of each other to add functionality (amortizing proof cost in the process) 👀

An overly simplified version of work-incentive system in the future has three pools: no job, real job, Steve jobs. They correspond to proof-of-no-work, proof-of-work and proof-of-change, respectively.

Append-only (authenticated) databases are inevitable. Huge thanks to raz, IsaacNotNewton, and Thomas Kim for taking the time to unpack the insights behind the pioneering QMDB and explore how its scalability can drive SOTA TPS via FAFO.

The days of hiring for overly specialized roles are over. A smart person with AI can become adequate in another field over night. The exception requires deep insight that are not yet compressed by AI. Hiring smart and hungry people is the right way. Even more so these days.

Bring the Bridge Home. Every day decisions get made internally at LayerZero on products to build, partnerships made, integrations Every day small micro decisions of do we do it ourselves, send them to Stargate, who writes the code. All of this is overhead All of this

A preliminary draft of “Minimmit: Fast Finality with Even Faster Blocks” is now available on arXiv! TL;DR trading the “slow path fallback” for a ~19.7% reduction in transaction confirmation latency 🏎️

A famous quote from House of Cards "everything is about sex, except sex. sex is about power" The version for startups is "everything is about growth, except growth. growth is about mission" Growth is a silver bullet for startup's problems.

Musk’s “Idiot Index” = cost of finished thing ÷ cost of raw stuff. In rockets, it was 50x (ouch). But it’s everywhere—products, software, housing—things cost way more than the basics. Time to rethink complexity and price

Decision making is a search process. Clarifying questions reduce the search spaces effectively. Clarifying questions in the most precise and effective form would be a simple yes/no question. Acuity of decisions depend on the qualities of the clarifying questions

I think this puts us in the number eight spot now 👀

Reading about TCP BBR this morning makes me realize how much it mirrors the effective leadership principles. Unlike traditional loss-based algorithms, such as TCP Reno, which reactively slow down when packet losses signal congestion, BBR takes a bolder approach. It periodically

The best team/org works like a truth machine

A question came to me "can a deterministic system like a computer exhibit emergent properties?". put another way, does a computer provide any property that was not predicted? The ability to perform universal compute could be considered an emergent property of the hardware: a

The nature of over claims is Fools think they’re cool; the cool know they’re fools.

Grits, wits and rizz. Formula for success.