Together with researchers at Boston Children’s Hospital and Harvard, we published a study in NEJM AI showing how o3 Deep Research helped clinicians revisit previously unsolved rare pediatric disease cases, and find answers for families who had waited years.
Work-in-progress: a new split layout framework for nearly all Apple platforms (macOS, iOS, etc.). A mix of SwiftUI and raw CoreAnimation. The goal is "every frame perfect" animation. Still very wip, but getting there! Goal is to bring it back to Ghostty eventually.
I posted on X about AI in the Finnish public sector and got a pretty revealing debate going.
My take is that Finland has roughly 700K public sector employees plus state-owned companies, civil society organizations running on government transfers and the church. The majority of that work is form processing, decision preparation, report writing, and moving data between systems. All automatable today.
Someone replied that this would mean "machines handle thinking and human relationships" and compared it to something dystopian of course. But none of these tasks require thinking or human relationships. Decision preparation is collecting data, running analyses, making comparisons, and writing summaries, which is data processing. AI does it in seconds, without errors, without fatigue.
The next argument was "what about discretion and risk management." Risk management is literally one of the best AI use cases that exists. AI goes through thousands of data points and calculates risk scenarios faster and more comprehensively than any person. Discretion means the final decision a human makes. Nobody said remove the human from the final decision. But everything that happens before that discretionary decision is data processing and AI is better at it.
The real problem is not whether AI can do the work because it absolutely can. The problem is that the public sector is simultaneously the largest employer and the largest voting bloc in Finland. No political party will push automation that eliminates hundreds of thousands of jobs from people who vote so we will keep paying for manual labor on tasks that machines already do better. Not because we have to but because no one has the political courage to just do it.
Minun viestini kaikille miehille on: me emme saa vaieta.
Tehdään Suomesta yhdessä turvallinen maa naisille.
Allekirjoita sitoumus:
https://t.co/C5jFr8054s
👨🎨 We're sharing a formula for making vibe-coded apps look professional.
It's a deconstruction of the "science of design" and we think it will help you coach your agents into mobile app design that brings polish to your apps.
The formula and examples are in the post below ↓
This is a new paradigm for interacting with Claude that is significantly more "inline" with all the other human activity org-wide. Once you do all of the under the hood engineering work to make this "just work" (e.g. across tools, integrations, compute environments, memory, security, etc.), Claude basically joins the team in a seamless way - you can talk to it as you would talk to a person and it can help with a very large variety of workloads.
Imo this is the 3rd major redesign of LLM UIUX. The first paradigm was that the LLM is a website you go to, the second was that it is an app you download to your computer. This third one is that it is a self-contained, persistent, asynchronous entity with org-wide tools and context, working alongside teams of humans. It really takes a while to wrap your head around it, but it works and it is awesome.
This year, the EF is decreasing its budget by roughly 40%, which entails some difficult decisions. The goal of the decreases was set out in the Treasury Management Policy last year: the EF is transitioning into being a long-term-oriented endowment-based organization, shifting from its pre-2026 average of spending ~15% of its remaining funds each year, toward a post-2030 target of ~5% per year.
Often, when an organization goes through something like this, people try to pretend that nothing of great value was lost, that it is an efficiency increase, that the only people cut are unproductive dead weight, and everyone else stopped partying, studied the blade, entered cracked S-tier beast mode, and this was sufficient to make up for the downside. I will not try to pretend this. I respect my EF colleagues far too much to pretend that there was not much that is lost. They are brilliant people. They are dedicated engineers of whom some have worked on the Ethereum protocol for nearly a decade. They have brought a bright light to the Ethereum ecosystem with their code, their words, their warmth as human beings and their actions. My dearest hope is that they find a path that brings them fulfillment and happiness whether inside Ethereum or outside. Hopefully many will be able to bring their excellent talents and mindset to the wider Ethereum ecosystem, or the even wider CROPS world.
Instead, I will try to explain what *are* some of the grand sacrifices being made. The Ethereum Strawmap is no small thing. It is an extremely ambitious undertaking seeking to replace and augment almost every part of the protocol - consensus, proofs, privacy, account model, state, and more. This is the third iteration of Ethereum, in the same way that the Merge was the second, even if the shipping style is less Big Bang and more one-piece-at-a-time. On top of this, the EF is increasing its role in the Access Layer. We are not compromising on Ethereum being a Deeply Impressive protocol, something worthy of its place in a world with quantum computing, rockets to Mars and powerful biotech and AI, and capable of meeting the challenges that this era will bring.
Some of the deficit will be recovered through more work happening outside the EF. But not all. So what are the grand sacrifices that will enable a leaner effort to accomplish all of this? I will give a few examples (though far from an exhaustive list):
* The multi-client model will shift in the direction of multiple clients existing less for _redundancy_, and more for _specialization_. Up to this point, redundancy has been the main security strategy: if one client has a bug, if it has less than 33%, the chain keeps going and does not even stop finalizing. We are increasingly exploring moving more pieces of the protocol to a different security strategy: AI-assisted formal verification. Some smaller pieces of Ethereum (eg. BLS libraries) have worked this way already for a long time. But soon many more parts of Ethereum will likely function on this model. This may greatly reduce resource requirements of shipping a large number of EIPs. The resources saved by client teams can ideally instead be used to better serve different specialized user needs, including EF Access Layer goals.
* PSE (Privacy and Scaling Explorations) is winding down as a unit. The number of people working on ZKPs for privacy and scaling is probably as high as ever, but they are working less on "exploration" and more on *implementing* ZKP-based privacy and scaling into the Protocol and Access Layer
* Devcon will likely over time become smaller-scale, somewhat more spartan, much lower-deficit than previous years, in addition to other changes in vision in line with the Mandate.
* Fewer beyond-Ethereum megaprojects coming from EF. As I announced earlier this year, I am taking on some of the responsibility of doing projects in this category that I consider valuable with my personal funds.
* EF institutional work is reducing in scope, specializing more specifically on creating replicable test cases of highly CROPS-friendly deployments, even if at smaller scale.
These do not explain all departures; in some cases they do not explain departures at all and rather explain _reduced need for new spending_. But they are a large part of the strategy at play.
In the longer term, I personally favor a "soft lean-and-done" approach to Ethereum: once the Strawmap is completed, generally stick to security fixes and small high-value changes, and have a much higher bar for considering new feature additions to the protocol. This allows Ethereum to remain capture-resistant without demanding very large budgets. Learn less from multimillion-line-of-code behemoth projects, more from bitcoin.
The past years have been a challenging era for Ethereum. However, the ecosystem is adapting, both inside the EF and outside, and I am confident that Ethereum is very well-positioned to succeed and thrive.
https://t.co/iZiOonRYzR
There have recently been claims that AI text analysis will make online anonymity untenable.
So let me cannibalize a piece of my own anonymity to do an experiment.
At some point this decade, I wrote a published document of medium importance to Ethereum - I estimate ~200 to 2000 documents in Ethereum are as or more important - not under my name.
Find it.
(I genuinely have no idea how easy or hard this is, will be very curious what comes out)
𝝠 Expo native modules are 1.6 to 2.3x faster on Apple platforms in SDK 56.
Swift talks to JSI directly. The Objective-C++ layer in the call path is gone, so calls lower straight to C++, matching or beating Turbo Modules.
Deep dive from @tsapeta: https://t.co/hllsHdCVp0
I'm joining OpenAI next week!🥹 The job search turned out to be really challenging but also super rewarding, so I wrote a small blog to share what I learned along the way and hopefully make the process a little less mysterious for the next person. https://t.co/6FigSBdenD
🎉 Ep 1 of our new series, Building for Good with G$, is live!
To kick it off, we welcome Rael & @meriiifdez of @gooddollarorg & @GravenPrest of @flowstatecoop to explore how they're using streaming funds to reimagine public goods funding 🌱
🎧Listen: https://t.co/5swl7jf0q3
Building for Good with G$ is a new series exploring the people, projects & communities powering the GoodDollar ecosystem 🎙️
In episode 1, we discuss:
🌱 How GoodDollar has evolved from a universal basic income (UBI) protocol into a growing public goods ecosystem
💸 How they're partnering with FlowState to pioneer streaming-based funding, where money flows to builders continuously instead of in one-off grants
📊 Highlights and standout stories from GoodBuilders Season 3
🔁 How any ecosystem or community can run their own streaming funding rounds with G$
👀 Updates and alpha on GoodBuilders Season 4!