Mikiya Matsuzaka

#FinTech #NetworkProgramming #FinancialInfrastructure In a Financial TCP/IP model, issue and redeem for ASCII tokens should be the minimum required functions of every sovereign, exchange, and provider (basically every closed payment LAN). If all of them support that interface, money can move from anywhere to anywhere.

#FinTech #NetworkProgramming #FinancialInfrastructure Money sounds like the currency itself. Payment can include: • paying • settlement • invoicing • refunds • clearing That means it scales better as a protocol term.

#FinTech #NetworkProgramming #FinancialInfrastructure The Internet already exists. We don't need a new network. We need SMTP for payments. SPTP: Simple Payment Transfer Protocol

#FinTech #NetworkProgramming #FinancialInfrastructure There is criticism of Amazon Gift codes: “Anyone can redeem them. They should only be redeemable by the intended recipient.” I’m not convinced. Both models should exist. Bearer tokens and recipient-bound tokens serve different purposes. An exchange can issue and redeem ASCII tokens, while users route value through exchange wallets.

#FinTech #NetworkProgramming #FinancialInfrastructure In Financial TCP/IP, the “IP” layer is wallet-to-wallet transfer. Money can move from any wallet to any other wallet. The physical layer (“Ethereum”, “Resona Bank”, etc.) only implements deposit and withdrawal to wallets.

#FinTech #NetworkProgramming #FinancialInfrastructure I used to be skeptical of ledgers that preserve every record forever. But if the issuer is a sovereign state guaranteeing all currency it has ever issued, then maybe it makes sense. Old Japanese banknotes can still be redeemed at face value today. Coins and notes remain obligations of the state. If crypto wants to be money, this is the standard it should think about. And once you see that, the use case for other ledgers — like ring-buffer ledgers — also becomes clearer.

#FinTech #NetworkProgramming #FinancialInfrastructure Not one giant ledger. What we need is tens or hundreds of thousands of modest-sized ring-buffer ledgers (exchanges) across the world. Like ISPs on the Internet. Scale through federation, not centralization.

#FinTech #NetworkProgramming #FinancialInfrastructure What is Financial TCP/IP? The ability to move money from anywhere to anywhere. It should be built carefully in a non-capitalist way. Two core principles: • Multi-routing — money can bypass blocked intermediaries through alternate paths. • Content neutrality — rails relay transactions regardless of sender/receiver. Bad money should be handled by auditing, not by breaking the network. In the real world, auditing may deserve to stand beside legislation, judiciary, and administration as a fourth power.

#FinTech #NetworkProgramming #FinancialInfrastructure A ring-buffer ledger “fails” because old tokens remain? That’s only true if you think state must be deleted. Just move the tokens to another ledger. Finance is stuck in a delete mindset.

#FinTech #NetworkProgramming #FinancialInfrastructure If TCP/IP communication could only happen through specific apps, it would be absurd. That’s exactly the situation in finance today. Payments should be infrastructure, not applications.

#FinTech #NetworkProgramming #FinancialInfrastructure We don’t need payment apps. We need a financial TCP/IP. Call it: libpay.a Composable. Programmable. Universal.

#FinTech #NetworkProgramming #FinancialInfrastructure Not all money should be neutral. Imagine: money that can’t be used for gambling money that can’t be used for debt repayment constrained money is powerful.

#FinTech #NetworkProgramming #FinancialInfrastructure Payments as apps are closed systems. Payments as libraries are primitives. From debt collector to central banks— write it as code.

#FinTech #NetworkProgramming #FinancialInfrastructure Amazon gift-style value is weird in a good way: Copyable, but only spendable once. Now imagine: money that expires money with constraints money with intent Not just “better money” — designed money.

#プログラミング #設計思想 #エンジニアの備忘録 何年か前、Requestオブジェクトというものを発見しました。 ある関数呼び出しというのは実はデータ化できます。 関数ポインタ、引数の配列、返り値を入れる変数へのポインタ、 この3つはデータで、それをカプセル化すればいいだけです。 これをキューイングしたり遅延実行したりします。

#PersonalServer #SelfSovereignIdentity #DecentralizedWeb The essence of identity is server capability. An ID is not just a string of characters. What makes it an ID is the server functionality attached to it. That’s why, just as storage services like Google Drive offer free tiers, CPU resources should also have free tiers. Handling something as light as receiving SMTP mail should be trivial. Once everyone can own a small personal server function, many interesting things become possible. For starters, each person should be able to have their own email domain, similar to Google Apps / Workspace. You could register a dedicated domain, or receive a free subdomain. Under that domain, you could create and delete as many mail accounts as you want. Once individuals can own their own mail servers, you build layers on top. Second floor: a language for programs triggered by incoming mail — for example, Scheme. Third floor: encryption and signatures. It may be useful if verifying encryption or signatures can itself be done through command-based mail messages. A key challenge at this layer is making key rotation smooth. Fourth floor: for example, a Signal-compatible layer. Once keys exist and signatures are available, many facts can simply be signed by the appropriate authority. For example: ・over 18 ・resident of Japan ・holds a driver’s license ・holds an IT certification ・is a licensed physician You prove the claim, without revealing the underlying personal data. The freedom to distribute reputation across multiple identities should be recognized. Personal microservers should of course also host wallets and handle small payments. Here’s one idea: Receiving email could be free, while sending email costs money. Spammers probably wouldn’t even pay $1/month just to send mail.

#Plan9 #DistributedSystems #Infrastructure Tips: Plan 9 authentication used a challenge/response model to issue sessions. In other words, it did not hand out semi-persistent access permissions. Another interesting point is why it passed on public-key authentication: cron-style unattended execution was harder to support cleanly.

#AIエージェント #LLM #FutureOfCommunication Email is programmable (thanks to procmail). Voice calls aren’t. Messaging isn’t. Social isn’t. We need: procphone procskype procmeta These correspond respectively to large language models (LLM), large audio models (LAM), large video models (LVM), and large XR models (LXM).

#Trustless #GameTheory #CryptoGovernance This is what I want: 1. I send you an email. 2. You can see whatever public identity data is attached to it — maybe partial, maybe none. 3. But I tell you: “If I scam you, report this address and I can be traced quickly. Trust me on that basis.” The hard part is governance: who can unmask whom, under what conditions? Maybe the clue is bail bonds. Stake enough value that running away is irrational.

#Web3 #DigitalIdentity #FinTech Imagine this: 1. A name inside an HTML or PDF document. 2. The name is a link that reveals an email address. 3. Send a command email there, and the document signature gets verified. 4. The same identity also exposes a wallet. 5. You can send money directly. Documents, identity, messaging, and payments should be one system.