ZK Basics | Ep. 11: Trustless doesn't mean anonymous A question from the replies this week: if blockchains are trustless, why is it so hard to track an address that stole money? The answer is that two different things got packed into one word. Trustless means you don't have to trust anyone to validate transactions. Everyone checks the rules themselves. That's it. It never meant private, and it never meant reversible. Tracking the address is actually the easy part — every hop is public, and analytics firms follow stolen funds live. Two other things are hard: linking an address to a person, and doing anything about it. Thieves split funds across mixers, bridges and chains. And the same immutability that stops anyone from faking your balance stops anyone from clawing the money back. So a public ledger gives you verifiability, not identity, and permanence, not justice. ZK is what lets you add the missing piece deliberately: prove the specific thing that needs proving — this user is verified, this transfer is compliant — without putting everyone's life on a public ledger forever. Keep the questions coming. This one became an episode 💜

Three things worth your weekend. 1. Quanta Magazine on a new kind of zero-knowledge proof, where the secrecy comes from the limits of math itself. It also retells how the idea was born in 1985. A long read, worth it if you liked the history behind Ep. 10. 2. CryptoSlate's beginner guide to ZK proofs, updated this month. Rollups, identity and shielded payments in one 10-minute read. Send it to the friend who keeps asking what we're on about. 3. Our own Ep. 10, if you missed it — five common myths about zero-knowledge, from "it means anonymous" to "it's too slow". No thread, no takes. Just things worth your time 💜

Crypto News in 3 Lines | #5 1. BNB Chain published plans for a new Layer 1 alongside BSC — 200ms blocks, ~20ms pre-confirmation, and no separate token. BNB stays the only native coin 2. At the Nasdaq bell ceremony, CZ said crypto could become the native currency for AI, and that he wants BNB Chain to be the main network for it 3. The Jenner hard fork hits BSC testnet tomorrow, October 28 The common thread: If AI agents are going to hold and move money, speed is only half the problem. The other half is proving the agent did what it claims before the money moves. Fast rails plus verifiable agents. One without the other is just a faster way to be wrong. Which one matters most to you: 1, 2 or 3? 💜 #BNBChain #AI #ZK

ZK Basics | Ep. 8: Why ZK proofs cost so much Verifying a proof takes milliseconds. Creating one is the expensive part. Here's where the money goes: 1. The computation gets rewritten. Your program has to be turned into a form math can prove — a circuit. Nothing about this step is free. 2. Proving is heavy. Building the proof can take thousands of times more compute than just running the program normally. 3. Specialized hardware. Serious proving runs on GPUs and custom chips, which is why there's a whole industry around proving infrastructure. 4. It's dropping fast. Costs have fallen by orders of magnitude in a few years, which is exactly why ZK moved from papers to production. The tradeoff: pay once to prove, let everyone verify cheaply forever. Save this one💜

Three things worth your weekend. 1. A long-form explainer on zkML, if you want the whole picture in one read. Best fact in it: the largest model proven end-to-end in zero-knowledge so far is around GPT-2 scale. Useful reality check on where this actually is today. 2. A practical walkthrough of running AI inference off-chain with proofs. It makes a point we keep repeating: "zero-knowledge" doesn't mean every input, output and piece of metadata is hidden by default. You choose what stays private. 3. Our own Ep. 8, if you missed it — why creating a proof is expensive, why verifying is nearly free, and why that gap is the whole business model. No thread, no takes. Just things we found good this week💜

You asked. Here's what came up in the replies this week. "Can you prove a specific model ran on a specific input while keeping the prompt private?" Yes. The prompt goes in as a private input, only its commitment is published, and the proof binds that commitment to the run. Nobody sees the prompt. Everybody can check it was the one used. "Does that proof say anything about answer quality?" No, and this is the part worth remembering. The proof covers faithful execution: this model, this input, this output. A bad prompt still produces a perfectly valid proof of a bad answer. "If blockchain is trustless, why is it so hard to track an address that stole funds?" Tracking it isn't hard — every hop is public. Linking it to a person is. Trustless means nobody has to be trusted to validate transactions. It never meant anonymous, and it never meant reversible. Keep them coming. Every question in the replies gets read💜

Crypto News in 3 Lines | #4 1. BNB Chain now holds 42.59% of the tokenized stock market — $1.47B, more than Ethereum and Solana combined. Two weeks ago that share was 30% 2. The 37th quarterly BNB burn is done: 1,646,196 BNB removed, roughly $1.2B at the time of the burn 3. The Cysic bridge was halted after an attacker reached the database behind it and minted 10M CYS on BNB Smart Chain The common thread: Story 3 is the one to sit with. The bridge itself wasn't broken — the database it trusted was. Every system has a part nobody checks. ZK is how you shrink that part until there's nothing left to take on faith. Which one are you watching: 1, 2 or 3? 💜 #BNBChain #RWA #zk

ZK Basics — six episodes in one post If you're just finding the series, here's everything so far: Ep. 1 — Proving you know a password without typing it Ep. 2 — The three rules every ZK proof must pass Ep. 3 — Two balls, one color-blind friend, zero leaked information Ep. 4 — Why 20 rounds of checking became a single proof Ep. 5 — Where ZK already runs today: rollups, payments, identity, AI Ep. 6 — The one thing ZK can't do: prove your input was honest One idea runs through all six: prove the claim, reveal nothing else. Save this as your starting point 🔖 Ep. 7 drops this week 💜

SATURDAY.EXE Loading weekend ...... ██████████ 100% Loading brain ........ ███░░░░░░░ 30% Loading coffee ....... ██████████ 100% Six episodes of ZK Basics done. Ep. 7 lands next week. What's your weekend mode? Building / Learning / Recharging 💜

ZK Basics | Ep. 6: The one thing ZK can't do A zero-knowledge proof can show that a computation was done correctly. It cannot show that the input was honest. Prove you're over 18 — the proof is solid. But if the ID behind it was fake, ZK proves the fake just as cleanly. Prove an AI ran the right model — solid. But if the training data was garbage, the proof says nothing about that. ZK removes the need to trust the process. It doesn't remove the need to source good inputs. Anyone selling you ZK as magic is skipping this part. Worth knowing before you believe the next thread about it 💜