The Sword NFT

some animation tests for pokedbots animated pilot #icp via @hollowpixels

A world first: a demonstration of AI running *on* blockchain as a smart contract 🧠⚡️ The Internet Computer is used – the world's first 3rd gen. blockchain #ICP. AI will become the beating heart of our web3, multi-chain world, and this is only my first demo. Code will follow shortly. This is running on DFINITY's Internet Computer testnet, but you'll be able to take the code and run it on the public network as the NNS is expected to up the per-transaction instruction limit in the coming days. Some important notes. The inference engine used has not been optimized, and we will show vastly greater efficiency in subsequent demos, where the AI runs even faster, and consumes less gas/cycles. We will also propose to the NNS that smart contracts have access to SIMD instructions – which we have determined are deterministic – unlocking vastly more speed and efficiency. Lastly, currently the actor smart contracts ICP hosts run inside a 32 bit environment, which limits their main memory to 4GB. Within the next couple of months, we expect the Internet Computer to move to a 64 bit environment, allowing smart contract memory limits to be raised much higher – allowing for models with far more parameters, and thus power, to be run. Those of you who have followed my posts know that hardware optimization support is planned. Firstly this will involve WASM smart contracts shipping out matrices for processing by the CPUs on existing node machines (another advantage of the Internet Computer running on sovereign node hardware). Secondly, we plan to propose a new public node machine specification, which node providers can build to, for use in subnets specialized for hosting AI smart contracts, in which each machine will incorporate several GPUs. Naturally, this will be packaged with other technological developments that ensure smart contract determinism. This is part of our mission to enable powerful LLMs to run as smart contracts on the Internet Computer – in a forms that are tamperproof, unstoppable, and optionally autonomous (including under the exclusive control of DAOs). Our vision: you will be able to have a chat with a smart contract. A smart contract will be able to coordinate your organization (see my earlier tweet about a "Delphi"). ICP smart contracts will be able to audit Ethereum smart contracts, kitemarking those without backdoors and reentrancy bugs. And smart contracts will be able to do KYC autonmously, matching faces to driving licenses, and more. We will deliver new SDK enhancements, and work with partners, to turbocharge crypto AI developers. Secure and unstoppable AI and third generation blockchain will be two sides of the same coin. Security and AI will be indivisible. AI models will be traded as NFTs. Thanks to Internet Computer's chain key (trustless multi-chain) capabilities, all blockchains will be able to leverage AI smart contracts. The future is beyond exciting. Thanks for watching. Can't wait to give you more demos!

laser-eye maxis are mad at @brian_armstrong, because he tweeted that paying merchants with L1 BTC has a poor UX, so coinbase no longer supports that. guess what? he’s right! how do i know? because at @TaprootWizards, we’re one of the largest BTC merchants on the planet. this month, we processed over 300 BTC in bitcoin L1 payments (over $15m USD). on top of that we also processed over 20 BTC (over $1m USD) in payments that had to be refunded due to inefficiencies of L1 BTC as a payment protocol. my understanding is that this probably makes us the largest BTC-only merchant in the world for the month of february, or at least one of the largest. (BTC merchant = selling goods directly to consumers for BTC) i’m going to describe some of the issues we encountered, but first, some background. these payments were processed during the @quantumcats sale. the buyers were buying Quantum Cats ordinals from us. bitcoin jpegs. i know that some of the readers don’t like (or don’t understand) jpegs. so allow me to introduce another, very similar type of sale: ticket sale for a high-demand concert. imagine a taylor swift concert. the demand is much higher than the supply. tickets are not cheap, but in an attempt to reward loyal fans, the organizers are pricing them below what the market perceives as “fair value”. organizers attempt to prioritize “true fans” and to limit the amount of tickets each fan can purchase, but no matter what organizers do, it is clear that many people will try to buy as many tickets as they can just to turn around and resell them for a higher price as soon as all tickets sell out, probably within minutes. that is the exact situation that popular jpeg mints find themselves in, and so even if the reader really hates on-chain jpegs, bitcoin payment solutions will still have to improve to the point that they can support the high-demand ticket sale usecase, if we ever hope for bitcoin to become a successful payment method. ok. so here are some of the issues we encountered: 1. LONG CONFIRMATION TIMES for some usecases, long confirmation times don’t really matter. for example, if you’re selling a bitcoin tshirt, with unlimited supply, and low demand, and the tshirt takes 2 weeks to ship, then it doesn’t really matter if the transaction takes 30 minutes to confirm. as soon as a transaction hits the mempool, you tell the buyer that their payment was received and they can close the tab and move on with their lives. once the payment confirms onchain, you can start the shipping process. and if it fails to confirm, you send them an email and tell them something went wrong. cool! but this isn’t really that interesting of a usecae. it works for products in low demand. in other words, it works when no one’s using it. what happens if there are only 2,000 shirts, and they’re really sought-after so you know that 50,000 people are going to be racing to buy them before supply runs out? suddenly, fast confirmation is really important, because the buyers NEED TO KNOW ASAP if their payment was successful or not. if they’re not sure, they might send you multiple payments, hoping one of them goes through! they REALLY want to make sure they’re getting one of those tshirts, you see… and if the order only confirms once the payment confirms, you might end up having to process A LOT of refunds, because 50,000 people would make payments hoping to get in, but you can only fulfill orders for 2,000 people! so for us, in order to minimize refunds, we had to approve orders based on UNCONFIRMED TRANSACTIONS, as soon as they hit the mempool. this meant that buyers were reserved a Qunatum Cat before having their payment confirmed in a block, which is a better experience for the buyers, but risky for the merchant: before a payment is confirmed, there are multiple ways for the buyer to reverse it. to mitigate that risk, we used a bunch of proprietary heuristics to verify that a buyer really *intends* to complete the purchase, and won’t try to reverse it, even though they technically could. the good news: those heuristics worked, and all of the unconfirmed payments we approved actually got confirmed eventually. the bad news: this required our engineers to review incoming payments manually and sometimes take action to mitigate risk. it was a very labor-intensive process. it was worth it for us, to ensure a smooth experience for our customers, but most merchants won’t be able to afford doing this. 2. PAYMENT AMOUNTS AREN’T ENFORCEABLE the way L1 bitcoin payments work is different from most modern payment methods. the buyer gets an address to send BTC to, and while the merchant can tell them which amount to send, the customer can send any amount they want. it can be below or above the amount that the merchant is asking for. this issue doesn’t exist with other payment systems. when you pay with a credit card for example, you can only pay the amount that the merchant dictates, not more, not less. smart-contract-powered payment processing solutions, like the one coinbase provides, can also reject transactions that don’t match the merchant’s requested amount. so why is this a problem? for most merchants, the main issue is that some customers might send an amount below the invoice amount, either in error, or to see if they can get away with it. in that case the merchant will probably have to refund the transaction. but for HIGH-DEMAND products, like Quantum Cats, or like taylor swift tickets, there’s another concern: some customers might try to pay MORE, hoping they can fool the system and buy more than their allotted allocation. that’s the issue we had, and fortunately we were prepared for it, assuming that some customers would attempt this. so we kept track of overpayments, and ended up processing 20 BTC in refunds - OVER $1 MILLION DOLLARS - for people who intentionally overpaid, just trying their luck. but again, this required manual reconciliation, and many hours of manual work. it’s feasible in our case, but not feasible for most merchants. 3. ONCHAIN FEES onchain fees are a real issue for payment processors. essentially it’s another one of those “it only works when no one’s using it” scenarios. if a high-demand taylor swift concert was selling tickets only for L1 bitcoin payments, and tickets were only confirmed once the onchain payment ids confirmed, you’d see transaction fees rising to unprecedented levels, as buyers attempt to overpay miners to ensure they get their tickets before anyone else. we wanted to minimize our impact on the network, so we decided to let buyers reserve their Quantum Cats as soon as their transaction hit the memppol, as opposed to when the transaction is confirmed. this also helped mitigate long confirmation times, as described in (1). but as explained above, this required us to manually review incoming payments to ensure we only reserve Quantum Cats for payments that are likely to confirm. it’s a manual labor-intensive process and isn’t feasible for most online merchants. if we hadn’t done that, we’d have to deal with people setting fees that are too low and getting their transactions “stuck”, but also with people paying high fees, making the network more expensive to use for everyone else! 4. “JUST USE LIGHTNING BRO” the usual laser-eye response to these issues is telling people to “just use lightning”. well, we tried. we looked into it, and while LN does structurally resolve most of the issues above, our conclusion was that the lightning network cannot support the volume of payments we were expecting. why? well, 300 BTC is almost 10% of the entire capacity of the lightning network. because of the way lightning works, we’d have to source so-called “incoming liquidity”, which means that other lightning network service providers would have to lock up at least 300 BTC to connect to us and to allow their users to pay us. and since we can’t tell ahead of time which service providers our customers would be using, we’d have to have MULTIPLE providers lock up 300 BTC each to be safe. we’d probably have had to have providers lock up somewhere around 1000 BTC or moreto make this work smoothly. 25%+ of the entire network capacity. it wouldn’t have worked. btw we did use the lightning network for another campaign last year. we had 10,000 people make payments of $2 each over one weekend, and it created so much stress on the network that multiple CEOs of wallets and service providers contacted us and asked us to stop. imagine what would’ve happened with a $15m sale. lightning is simply NOT FEASIBLE for high-demand sales. it is not a serious solution. 🐱🐱🐱 i’m not writing this to FUD bitcoin. i’m obsessed with bitcoin and have been for a decade, and am very bullish on its future. we insisted on using only BTC for our sale even though it was hard, and even though our customers begged us to accept other currencies. Many of our customers got a BTC wallet and made a BTC payment for the first time ever to get a Quantum Cat. we’re gonna continue to push this forward and come up with innovative solutions to improve the experience, but if we care about bitcoin’s success, we need to be honest with ourselves: bitcoin is a BAD payment solution right now. we know because we have the experience that most bitcoiners don’t. when brian armstrong and coinbase say that the payment UX with bitcoin is bad, they know what they’re talking about. and no, lightning doesn’t fix this. 🐱🐱🐱 plot twist: bitcoin’s OP_CAT scripting function, if revived, would allow building better payment networks on top of bitcoin that resolve all of the issues above. we’re intent on bringing OP_CAT back. if you want to learn more and stay up to date, follow @quantumcats and turn on notifications 🔔