One of the trends for Cypherpunks will be local-first software. Local-first does not need a backend, or if it needs it, it's a dumb storage for encrypted data, often only just for synchronization/backup. Everything needed is stored locally. Think accounting, personal photo albums, family chats, ... One of the developers building tooling for these paradigms is @Daniel Steigerwald . I've recently spoke to him and he explained to me who local-first scales better. No data centers, no AWS, etc.

Scaling local-first software - Evolu

Building local-first apps is already a challenge, and making them scalable is an even greater one. It's not just about keeping data local; scalabil...

Many people will launch useful vibe coded apps. It's a superpower that is accessible to all. And some of them will want to have some kind of monetary function, a direct payment, value4value, etc. Being a merchant in a legacy system is very difficult, you need to go through bureaucratic hell, start a company, get a merchant account with the big players. Or you can tell your favorite model to import a nice library that handles payments, such as Cashu, which will allow you to integrate lightning payments without even needing to run a node. This will enable a whole new economy of micro apps. Accessible to those who can pay a Bitcoin lightning payment. This will be beneficial to both creators and the consumers of these apps. They will be mostly web apps, no app store bullshit, no companies, no banks, just a simple domain and hosting for a few bucks per month. Or a home computer and some sort of exposing it, for example thorough cloudflare tunnel. Lowering barriers to creativity is what the world needs and what we observe with Bitcoin and LLMs.

Statist thinking: We will make the rules and everyone will comply. Entrepreneur thinking: We go where we're treated best.

Why Tether Refuses To Comply With MiCA | ZeroHedge

ZeroHedge - On a long enough timeline, the survival rate for everyone drops to zero

PSA: Buy Bitcoin if you don't have any. Yes, it's peaking, but you missed all lower prices. The reason is peaking is that everyone else in the fiat economy is falling. You don't want to be caught empty handed even shit hits the fan. Bitcoin is insurance and insurance is expensive when the probability you might need it gets up. Buy it without KYC to your own cold storage. No ETFs, no MSTR, no "investment broker". Hard cold storage. If you made up some story in your mind about morality that's stopping you, read https://cryptoanarchy.institute/nocoiner-syndrome-2024.pdf You don't need to tell anyone, if you publicly hate bitcoin, we don't care. But when shtf, you don't want to hold government paper, or other fiat assets (stocks, real estate, bonds, ...). You might add gold for good measure. No, we're not pumping our bags with this, just look at the daily volume of btc/usd. You are not rich enough to move the price with your purchase. I fully know most of you nocoiners will just ignore me, one of the reasons I post this is to be able to show it to you later. It's up to you, if we'll be both laughing. I'd rather not laugh alone. Be smart. A quick intro (no sign up needed, free, I'm not making money on this):

A quick introduction to Bitcoin – wallet setup, buying, payments – Hack Yourself!

Tell your nocoiner friends. You don't want your friends to be poor, even if it means you were right and they were wrong.

I recommend this podcast (you can find it in podcasting apps as Lazarus Heist). It's about the North Korean Lazarus Group and everything they've pulled off and how they're being investigated. I knew many of the technical aspects, but how the group operates, where they come from, what kind of people they are, how many there are, and how they organize with other dark web people (like how they coordinated withdrawing tens of millions from ATMs in 28 countries after hacking a bank) is brutal. I recommend both seasons. I think it ends with that Bybit hack, but I haven't gotten there yet.

BBC World Service - Cyber Hack - Episode guide

Looking forward to @BTC Prague in June! I'll be joining the 'Practicing Cypherpunk' panel on the main stage at 10:20 on Friday 20th & the fireside chat 'Bitcoin Ninja' on the Czech Stage at 16:40 on Saturday 21st. Join me with 10% off tickets using code JURAJ

Another repo drop. This one I wanted to release "when it's ready", I still don't think it's perfect, but it is useful (helps with Liberation Travel Newsletter for example). It summarizes Signal groups - themes discussed, links shared, etc. It is a bit clever, read the description - it understands images, transcribe voice messages, extracts and describes links shared in the group, etc. Many of us are members of numerous Signal groups with many messages and it is nice to reflect (for example once a month) on what was discussed, what we learned (or what others learned because we did not have time to read them). This one took a lot of work and a lot of compute, so I would appreciate value4value - contribute back in time, talent or treasure, if you find it valuable. If someone wants to help, we can even turn it into a service for others, add other messengers, etc. Let me know if you are interested. Please, please, let me know if you try it and your experience.

GitHub

GitHub - jooray/signal-summarizer: AI summaries of Signal group content - text, images, links, voices. Runnable locally through ollama.

AI summaries of Signal group content - text, images, links, voices. Runnable locally through ollama. - jooray/signal-summarizer

V 80. rokoch, za železnou oponou, krajiny východného bloku organizovali svoje ekonomiky prostredníctvom Rady vzájomnej hospodárskej pomoci (RVHP), rozdeľujúc výrobné úlohy medzi členské štáty. Počítače a ich súčasti neboli výnimkou. Bulharsko dostalo za úlohu vyrábať počítačové pevné disky—čo viedlo k výstavbe úplne novej továrne na zelenej lúke. Továreň si rýchlo vybudovala povesť naprieč východnou Európou: jej pevné disky boli notoricky nepredvídateľné, citlivé dokonca na zmeny teploty v miestnosti. V tom čase na Matfyze v Bratislave dostal mladý slovenský študent menom Róbert Szelepcsényi za úlohu napísať softvérové ovládače práve pre tieto bulharské disky. Zakaždým, keď sa pokúsil čítať prvý sektor dát, chybný hardvér mohol umiestniť magnetickú hlavu nepredvídateľne—na správny sektor, alebo možno o jeden či dva sektory vedľa. Szelepcsényi vymyslel elegantné riešenie: vždy keď sa jeho softvér spustil, najprv prečítal špeciálny marker z miesta, kde pristála hlavička disku, čím určil jej skutočnú pozíciu. Potom zodpovedajúcim spôsobom opravil posun, čo umožnilo zvyšku systému správať sa, akoby bol disk celý čas spoľahlivý. O roky neskôr táto praktická skúsenosť poskytla Szelepcsényimu pohľad, ktorý priniesol do teoretickej informatiky, inšpirujúc prelomový dôkaz známy dnes ako Immerman–Szelepcsényiho teorém. Tento dôkaz vyriešil dlhodobú hádanku: či problémy riešiteľné nedeterministicky (uhádnutím riešení a ich overením pomocou obmedzenej pamäte) mohli tiež riešiť svoje komplementárne problémy—problémy, kde sú odpovede "áno" a "nie" vymenené—v rovnakom (lineárne) obmedzenom pamäťovom priestore. Szelepcsényiho pohľad odrážal jeho techniku ovládača: aj keď výpočet okamžite nepoznal svoju presnú situáciu, mohol nedeterministicky "uhádnuť" stav a potom metodicky overovať správnosť systematickou kontrolou a počítaním. Rovnako ako kedysi premenil chybný hardvér na niečo spoľahlivé, demonštroval, ako zdanlivo obmedzené výpočtové modely mohli systematicky overovať svoje komplementy. Je to trochu "don't trust" (bulharskému* pevnému disku), "verify" (že som na sektore, kde chcem byť). Aj keď sú produkty nekvalitné, existujú šikovné spôsoby, ako využiť túto myšlienku. Používam to často pri programovaní, bola to jedna z najviac inšpirujúcich lekcií a je mimoriadne užitočná v oblasti umelej inteligencie. Nechajte AI vygenerovať odpoveď, overte, či to správne pochopila. Robím to často. Napríklad môj projekt rss2podcasts prevádza článok do formy vhodnej na prečítanie, potom overuje, či LLM nezmenila význam, potom vykoná prevod textu na reč pre daný úsek a následne spätný prevod reči na text, aby overil, či bol správne vygenerovaný (čo je zvláštne, často nie je!). (*) Bulharsko prešlo dlhú cestu od sovietskych čias, produkty boli v tom čase nekvalitné v celom sovietskom bloku.

Immerman–Szelepcsényi theorem - Wikipedia

In the 1980s, behind the Iron Curtain, the countries of the Eastern Bloc organized their economies through Council for Mutual Economic Assistance, dividing production tasks among member states. Computers and their parts were no exception. Bulgaria received the assignment to produce computer hard drives—leading to the construction of a brand-new factory right beside a cow farm. The factory quickly earned a reputation across Eastern Europe: its hard drives were notoriously unpredictable, sensitive even to the temperature shifts inside the room. Around this time, at Comenius University in Bratislava, a young Slovak student named Róbert Szelepcsényi found himself tasked with writing software drivers for these very Bulgarian drives. Each time he tried reading the first sector of data, the faulty hardware might land the magnetic head unpredictably—on the correct sector, or perhaps one or two sectors off. Instead of giving up, Szelepcsényi devised an elegant workaround: every time his software started, it first read a special marker from wherever the drive’s head landed, determining its actual position. It then corrected the offset accordingly, allowing the rest of the system to behave as if the drive had been reliable all along. Years later, this practical experience provided Szelepcsényi with an insight he brought into theoretical computer science, inspiring a groundbreaking proof known today as the Immerman–Szelepcsényi theorem. This proof solved a longstanding puzzle: whether problems solvable nondeterministically (by guessing solutions and verifying them using limited memory) could also solve their complementary problems—the problems where "yes" and "no" answers are swapped—in the same constrained memory space. Szelepcsényi’s insight echoed his driver technique: even if a computation didn't immediately know its exact situation, it could nondeterministically "guess" a state and then methodically verify correctness by systematic checking and counting. Just as he had once turned faulty hardware into something reliable, he demonstrated how seemingly limited computational models could systematically verify their complements

Immerman–Szelepcsényi theorem - Wikipedia

PSA: If any project, service, especially anything related to privacy, uses Discord or Telegram for discussions, support or sharing knowledge, avoid them, it's a huge red flag. They don't care about your privacy and often security. They involve third parties and if there are groups, there are often scammers. Encrypted messengers, or even web forms are better. It never ceases to amaze me how people easily fall for "it's user friendly" and put people at risk.

Another repo drop from me: rss2podcast Not everyone has time to read, but reading is essential. This tool enables you to listen to what you want to read instead. It has two modes - you can convert your own blog to a podcast, or you can create a personal podcasts from articles from all around the web. This powers the audio version of my blogs:

Juraj Bednar

Juraj's blog - podcast edition

I have an audio version of my podcast, converted using a state of the art text to speech model. I know TTS is not as cool as the real thing, but yo...

This also powers the website

Loaditfor.me

which allows you to convert the articles on the internet that you want to read. There is a lot of work I put into this, I hope someone finds it useful. Returning value (for example as sats) appreciated if you find it useful. This is actually the last piece, markdown2audio and a better StyleTTS2 inference engine has been already released before. Look at how I generate it, it's quite interesting, it's not just raw reading of the article, I convert it into a form that is better for voice narration using LLMs, I verify outputs with speech to text, etc.

GitHub

GitHub - jooray/rss2podcast: A tool to convert blogs, websites and articles to podcast, with a daemon and a web interface that powers loaditfor.me

A tool to convert blogs, websites and articles to podcast, with a daemon and a web interface that powers loaditfor.me - jooray/rss2podcast

Enjoy!

Inspired by @calle I am pushing some repos that just lay on my laptop and someone might use them. This one is for a very specific use case, probably not many will need it. If you need to hold a shitcoin for something, but don't want to risk being rugpulled, short it to stay in USD. But before doing that, see if you won't pay too much in the funding of the short. You can also calculate any staking or other form of yield and see the net p.a. position. If you know what I am talking about, you know, if you don't you don't need it 😉 Supports Binance and Hyperliquid public APIs (no API keys required, uses public data). More repos to come this week.

GitHub

GitHub - jooray/funding-rate-tools: Fetches funding rates, stores them, calculates p.a. rates, and generates a dashboard

Fetches funding rates, stores them, calculates p.a. rates, and generates a dashboard - jooray/funding-rate-tools