New paper alert! 🚨 Weighted Batched Threshold Encryption with Efficient DKG w/ Alexander Frolov, @AditiPartap97, and @ErtemNusretTas In multi proposer consensus, transactions should be hidden until “valency”, the point of no return at which the consensus outcome has been decided. The cleanest way to achieve this property is with threshold encryption. But existing schemes forced difficult tradeoffs between batched, weighted, and efficient schemes. I asked the @a16zcrypto research summer interns if they could come up with a new scheme that was weighted, batched, performant, and didn’t require a trusted setup. Turns out they did!!! With some optimizations, I was able to get decryption for up to 16 simultaneous proposals to run in ~40ms on 12 cores with thousands of virtual shares and decryption shares that are small enough to fit in a single UDP packet along with the rest of the vote data. This leaves enough performance headroom room on my dev box to run a full Solana validator at the tip with the remaining cores!!! These results, along with other recent work in threshold encryption are finally pushing threshold encryption into the realm of practicality for use in the Solana protocol. We still have some more optimizations to try so we might be able to push it even lower!

Don’t understand why stripe can’t build their own token router and needs to pay 10b to acquire one. This software can be built easily and the compute deals / lockups should come naturally to them as well. Also biggest premise is OR won’t narc on you for distillation. Stripe will

How in the eff are we still having the "ETH needs to jack up blob fees" debate? It does not have the pricing power to do this. If you jack up the blob fees, and increase the cost on chains using them (read: those users), they will pivot to EXCEEDINGLY abundant altDA which solves ~all pain points. Users do not care if something is an Ethereum rollup vs a validium vs a 23 validator L1. Robinhood could have launched on Canton and users still would have aped Cashtag. This I can promise.

No technically the new Kimi model is not Fable but it extends the network effects of Fable. High value prompts will still want the security guarantees and trust assumptions of the base model making Anthropic credits more valuable even if they don’t receive that revenue directly

Two ways to generate alpha 1. Look at the data everyone else is looking at and draw new insights. 2. Find new data nobody else is looking at and draw the same conclusions everyone else will draw when they look at it. 1 is a lot harder than 2

DeFi is where: * Lenders get sub-risk-free rates for risky lending * Borrowers are apparently unable to run profitable trades like “borrow at less than tbills and buy a tbill” * Active, multi-strategy credit funds with struggle to provide yield competitive with an FDIC-insured savings account What are we even doing here? Are we all just trapped onchain with no way out? Even if the lenders are stuck, why are the borrowers not able to sustain very low borrow rates by real-world standards? And it’s not like we even built a DeFi that does what it was built for, which might justify the friction and low capacity for competent capital allocation. We were all on the censorship resistant, permissionless finance highway, and then the wannabe hedge fund guys grabbed the wheel and drove us into the ditch because “non-custodial software” didn’t earn them performance fees for underperforming Treasuries. It’s not even that centralized entities or replicating many TradFi structures onchain is sinful or shameful. They’re not. DeFi, CeFi, and TradFi can coexist. It’s that the onchain economy is apparently so unhealthy that the only way to remotely give lenders a reward in line with risk is via massive subsidies. All these gigantic Earn initiatives are money flowing the wrong way, swamping an already overcapitalized DeFi market where we apparently are incapable of scaling any product that’s not minute-by-minute margin or perps, the latter of which is zero-to-negative-sum and closed off from composability, so may as well be offchain from a macro perspective. RWAs were supposed to save us by letting yield flow from offchain markets to investors onchain. But all we got were “tokenized tbills” that were just nosebleed fees slapped onto a money market fund. I’m still waiting for these actual tbills so I can build a ladder of them without paying a middleman or three 60 bps of the 360 bps tbill yield. One gets the impression that onchain markets are only kept from draining into the real world by an invisible dam of CEXs’ and banks’ arbitrary freezing of funds keep people scared to off-ramp. DPRK can get the money out somehow but there’s not enough borrowers able to withstand a sub-5% borrow rate? Either there’s free money on the sidewalk or something is busted.

🚨New Paper with @Weiye_Xi , @ciamac and @Qiaoqiao2001 Here’s two different prediction markets priced at ~6.5%: 1. Will the US confirm the existence of Aliens in 2026? 2. *that* Spurs @ Knicks Game 4, ~4th quarter. They suggest that both these events are the same probability, but intuitively these feel very different: the latter (Knicks won!) feels a lot more uncertain. But prediction markets don’t immediately give us a way to quantify it.