DeFi TVL is now sitting at $42B, a 2.5 year low, below even when FTX collapsed.

The naive response is that this is bearish. But that's not necessarily true.

There are two reasons why this TVL downturn is not a big deal IMO. 👇 1. The risk-free rate has skyrocketed, so on-chain yields are now below off-chain yields. Compound is paying ~3% on USDC, while the Fed Funds rate is ~5.5%. This naturally pulls capital off-chain.

This is normal. As rates go down over the coming years, much of that will return.

Yesterday I freaked out about the revelation that @Ledger could spit out your private key with a firmware update.

Yet I noticed the smartest people were not freaking out. Was I missing something?

I spent the evening educating myself, and now I'm in the "nvm it's fine" camp. This was my initial mental model: I thought Ledger's Secure Element was like Apple's Secure Enclave—a box that a private key lives in which can only sign things, but "keys can never leave the device." h/t @roinevirta

But it's not! Firmware can exfiltrate the private key! Oh god!

FTX reminded us of the failures of centralization. Yes, we need regulation, but even moreso we need innovation.

Proud to announce we seeded @renegade_fi, an on-chain dark pool that uses ZKPs and MPC to build a truly anonymized decentralized exchange.

The future is trustless. 👇 1/ In the wake of FTX, traders have been wary of counterparty risk from exchanges. CeFi has entered a new low-trust phase.

With Renegade, counterparty risk is obviated via cryptography and zero-knowledge proofs. Custody is guaranteed on-chain. You don't need to trust anyone.

Just read through the SEC case against Do Kwon/Terraform Labs so you don't have to.

docdroid.net/xC1daxt/kwonzi…

A few takeaways: 1/ Fraud case is rock solid. Chai using Terra was a complete fabrication, with fake on-chain transactions and everything.

I was surprised by how egregious this was. People at Terraform Labs knew it was bullshit.

1/ Had a meta-level shower realization for why MEV auctions are necessary in blockchains.

As I say in this TCB episode—I was taught blockchains are big shared computers that we each take turns running code on.

I always cringed at MEV auctions because they destroy that fantasy.

https://twitter.com/hosseeb/status/1619463203540598785

2/ My instinct has always been "well, exchanges like NYSE use time priority, so shouldn't blockchains *eventually* do the same?"

But there's a key difference:

NYSE is running a continuous process. They are matching orders and executing trades in real time.

Every year at @dragonfly_xyz we do a retro on the biggest lessons of the year. This year we came up with the top 10 lessons that we wanted to take away as venture investors in 2023.

I thought it might be useful to share them. Here are the 10 lessons in order: 👇 1. Diligence actually matters.

In the year of FTX, LUNA, and 3AC, this is the one lesson everyone in the industry learned.

Doesn't matter how impressive the founders seem. If you don't verify what they are telling you, you *will* eventually get burned.

/1 The AI era is upon us. GPT-3 makes it clear: we will achieve AGI in our lifetimes.

Crypto will have a big role to play in this.

Why? Because crypto changes the API of money, decoupling *money* from *people*.

In the age of AI, this will matter more than ever. 🧵 👇 /2 Money has a simple API. It was designed over a thousand years ago.

The API is:

Person => Money
Company => Money
Government => Money

If you are not a person, a company, or a government, you can't have money.

1/ Pouring one out for all those who were impacted by the FTX collapse. If you lost more money than you can afford, I'm so sorry.

I promise it will get better.

I'm exhausted.

Have been thinking a lot about what happens from here, and what it means for this industry. 2/ First, there will be a global regulatory backlash.

It's not just in the US—FTX Int'l collapsing is a global catastrophe. This is much bigger than 3AC or Terra.

It's one thing to defraud lenders or retail degens. But Sam was the golden child of this whole industry.

1/ I’m excited to announce that @dragonfly_xyz has made a strategic investment into @AptosLabs!

Aptos is building a next generation L1 blockchain with a unique take on scalability, safety, and usability—here’s why I’m excited to partner with them: 2/ @AptosLabs's optimistic concurrency, Move VM, account model, and consensus mechanism will let Aptos achieve massive on-chain throughput while retaining decentralization. From day 1, Aptos should support high-throughput use cases across games, social media, NFTs, and DeFi.

OK, I'll summarize @Galois_Capital's arguments:

1) If merge succeeds, ETH1 will continue to exist because of support from miners. DeFi, exchanges, devs need to deal with it. Acknowledging this fork is taboo in Ethereum-land.

2) The merge has a nonzero chance of failure.

Done. (It's that with a lot of shitposting.)

It's important to listen to critics like Kevin. They do an important service to throw ideology into sharp relief.

See also haseebq.com/ethereum-is-no… which I wrote about this possibility ages ago, back when ProgPoW was the fork de jure.

4 Lessons on losing lots of money (from someone who has done so many times) 👇 1. Human beings are resilient. We habituate surprisingly fast.

When you lose it all, the first night feels hopeless. The next day feels gloomy. The next week feels manageable. The next month, you forget that you even lost it.

You'll be OK, I promise.

Today we released the AMM Test: a no BS Look at L1 Performance.

Research piece benchmarking L1 performance across the EVMs and Solana. If you want to know how fast the L1s *really* are, read on for the (surprising) details 👇 🧵

medium.com/dragonfly-rese…

(By me and @0xsudogm) 2/ First, there's no universal way to benchmark performance. But most internal benchmarks are bullshit (just xfers).

So we decided to create an objective benchmark: AMM trades per second. It's easy to measure, it reflects real-world usage—it's not perfect, but it's informative.

1/ Blockchains are not networks—they're cities. 🏙️

This simple mental model will help you understand why the L1 wars are evolving the way they are.

New blog post + short 🧵

medium.com/dragonfly-rese… 2/ Blockchains are usually described as networks, which implies they're endlessly extensible like the Internet, or Facebook. If blockchains are networks, then network effects will dominate, and one blockchain will win.

But networks are the wrong analogy for blockchains.

1/ The MEV crisis could completely disappear all on its own after Ethereum transitions to PoS. Maybe.

Shower thought 🧵: 2/ Recap: MEV refers to Miner-Extractable Value. It denotes the $$ that miners can extract by opportunistically inserting and reordering transactions in blocks. Miners have extracted tons of value from Ethereum doing this. (Flashbots estimates about $750M so far).

1/ With $AVAX Rush program just announced ($180M in DeFi incentives), we're now going to see a true cross-chain explosion in DeFi. Not just "EVM sidechains", but fully fledged blockchains like AVA.

That's why I'm really excited about @BenqiFinance👇 2/ BENQI is the first Avalanche native lending protocol to launch on Avalanche and will be one of the core financial primitives for the Avalanche DeFi ecosystem. There are two things reliably working in DeFi today: AMMs and lending. BENQI will be the lending platform of AVA.

1/ Only two things in DeFi today work at scale: spot trading and lending/borrowing. But in CeFi, derivatives are way bigger than either of those two. So why aren't DeFi derivatives taking off?

Here's a short thread, featuring the soon-to-launch @SynFuturesDefi. 👇 2/ In TradFi, futures volumes eclipses spot markets, accounting for 70%+ of all trading.

In DeFi, legacy derivatives DEXs like Synthetix still haven't hit product-market fit. What gives? Is the market not ready, or is there something deeper wrong with the model?

1/ Algorithmic stablecoins have exploded over the past year: Dai, FEI, ESD, Basis Cash.

But fewer and fewer people actually understand how these stablecoins work.

So I made a simple visual explanation of the major algorithmic stablecoins. Check👇

medium.com/dragonfly-rese… 2/ There are four basic types of stablecoins. First, full-reserve banks, a la Tether. These are centralized, hold all their assets in fiat, and their liabilities should always match their assets.

Think USDT, USDC, BUSD, and all the others.

🦄 Uniswap v3 just dropped.

It contains some huge ideas—I remember hearing about them from Hayden back in early 2020. Now they’re finally public!

These are my first reactions for the effects v3 will have on Uniswap's future. Rough takes right now, but here's a thread. 👇 In current Uniswap (v2), all liquidity is provided across an constant product curve from 0 to ∞. This makes v2 really capital inefficient.

You can see this in practice—Uniswap has $5B locked in capital but only does about $1B in volume a day (20% turnover). This is pretty low.

In @BrianBrooksOCC FT piece "Get Ready for Self-Driving Banks" he articulates a vision in which regulators, transformed into software auditors, could examine DeFi-based algorithmic banks for compliance with banking regulations, or even issue them bank charters.

But... (1/6) ... it begs the question: why would a DeFi bank *care* whether it complies with a US banking regulations?

If a normal US bank is not compliant with banking regulations, a judge will eventually order them to cease operations. Implicitly, regulators wield a threat of force. (2/6)

1/ I’ve been informed by multiple sources that China's BSN is going to be integrating with public blockchains, the first of which are @ethereum and @NervosNetwork. This is kind of a big deal. Thread. 👇 2/ To catch you up: BSN stands for the Blockchain Service Network; it’s a China-wide SOE-backed blockchain network, mostly permissioned. There's been a lot of speculation about it in the Chinese blockchain community.

1/3 @gauntletnetwork just released an in-depth stress test of Compound. It’s worth a read.

Result: Compound is extremely robust, and could safely scale to 10x the volume with minimal default risk.

gauntlet.network/reports/Compou… 2/3 They analyze this using agent-based simulations of lenders, borrowers, of liquidators, simulating actual EVM transactions that use a re-creation of the Compound contracts. They repeatedly simulate the worst ever day of Ethereum trading and observe happens to the system.