John Jared Foy

@HowToAI_ Probing the Middle is an information theoretic reading of the context window. It works in practice but that doesn't make it true; just less wrong than poking your agent with a stick until it works. drive.google.com/file/d/1S9JISM…

@_Felipe A real forest for the trees moment. Memory safety is an abstraction that affords the requisite semanticity for articulation of an otherwise latent property. Unsafe just means it’s potentially “unsafe”, not that it’s not.

Rust programmers’ reaction when they see a memory-safe program written in a memory-unsafe language, translated into Rust.

@ikechi0x1 @round Build them, my guy!

@jaredef @round I think having conversations about anything, especially those which are seemingly inevitable is great. But things are still fuzzy. We don't even know who is going to own the application layer or on what sort of hardware the everyday person will experience these interfaces.

why isn’t anybody talking about standards for generative interfaces

@ikechi0x1 @round Its time to start having the conversation! x.com/jaredef/status…

@jaredef @round Maybe I'm wrong, maybe what I sense is that we're not at the point where the dust has settled enough to have a coherent conversation about the standardization of generated user interfaces

The point is you are right and wrong at the same time. The ceiling is site by the bandwidth of the information channel itself. CS has an informational channel constraint called binary. Everything else is just an abstraction of this binary constraint. AI generation is fundamentally constrained by this binary information channel itself. So the problem is not the "standard" per se, but an insufficiently abstracted standard. "Standards suck!" = "stop thinking!" and "better standard!" = "think bigger!"

@ikechi0x1 @round Specs are standards for implementation layer. If you want the meta abstraction, it is called constraints.

@ikechi0x1 @round You are missing the point, a standard is a constraint that allows you to clear up a noisy information channel. That's why standards / specs are a thing in CS, period.

@round Standards are unreasonable for things like this. The literal meaning of a standard is that it is a ceiling! Tech needs to stop talking about standards. This is why everything sucks. May the best innovator win.

@round Its not just standards, its an entire apparatus.

@millerman Here’s the reason why philosophical inputs create indistinguishable-from-human outputs, maybe. jaredfoy.com/resolve/doc/08…

Can you tell Heidegger from AI? I built a test. 12 passages. Some real, some generated. The AI is good. Average score: 7/12. turing.millermanschool.com

@artillain @tonyennis @htmx_org Hey! You might be able to apply some of this to your methodology. I built a backend inspired by htmx, and you might be vibing in this direction! github.com/jaredef/presto

I'm building the PHP backend equivalent of HTMX, and here's my theories on why that happens: 1. In corporate software environments, which are the most common today, there are strong pressures that cause programmers to chose job security enabling technologies over simple ones that save work 2. Making hard things easy is one of the hardest things a programmer/designer could do, and is also a dangerously anti-job-security move in large corporate environments 3. What Neal Ford said

An idea that's been occupying me a lot lately: "*Low Status* tech/software practices". There are a lot of places where older or "dumber" approaches are actually better or faster, but where 99% of engineers won't acknowledge or even engage with them. cc @htmx_org

@a_westgate I see he’s moved on from string theory 😂

Yeah we already know dark matter is fake and gay thank you Chinese science man

@uncreated_light Haha, this is awesome

@HowToAI_ Probing the Middle is an information theoretic reading of the context window. It works in practice but that doesn't make it true; just less wrong than poking your agent with a stick until it works. drive.google.com/file/d/1S9JISM…

@HowToAI_ You can actually get this same perf out of any model by Probing the Middle. You don’t need to get Lost in the Middle if you are using hierarchical nested constraints as a prompt discipline. jaredfoy.com/resolve/doc/68…

Google has quietly dropped what researchers are calling "Attention Is All You Need V2." And it signals the end of the Transformer era as we know it. In 2017, the original "Attention Is All You Need" paper changed the world by proving that AI doesn't need recurrence, it just needs to pay attention. But today, even the most advanced models like GPT and Gemini suffer from a massive, structural flaw: Catastrophic Forgetting. The moment an AI learns something new, it starts losing what it learned before. It’s why AI "hallucinates" or loses the thread in long conversations. This paper, titled "Nested Learning: The Illusion of Deep Learning Architectures," completely replaces the way AI stores information. The researchers have introduced a paradigm shift called Nested Learning (NL). Here is why this is "V2": For the last decade, we treated AI models as one giant, flat mathematical function. NL proves that a model is actually a set of thousands of smaller, "nested" optimization problems running in parallel. Instead of one giant "memory," each layer has its own internal "context flow." This allows the model to learn new tasks at test-time without overwriting its core intelligence. It moves us past the static Transformer. The new architecture (HOPE) demonstrated 100% stability in long-context memory and "post-training adaptation" that was previously impossible. The technical takeaway is brutal for the competition: Existing deep learning works by compressing information until it breaks. Nested Learning works by organizing information so it can grow forever. We’ve spent 7 years trying to make Transformers bigger. Google figured out how to make them "Nested." The Transformer replaced the RNN in 2017. Nested Learning is here to replace the Transformer in 2026.

@theo Use the Telegram channel for Claude Code i use Termius and tmux to keep my session open when I want more visibility. Other than that Telegram does everything i need.

Just learned it's literally impossible to paste images into Claude Code over SSH. How do you CLI people live like this??

@Abunajohn @JayDyer this is up your alley, you can corroborate this. Fr., Jay is probably one of the few that can actually make your case to the world. Time to lay down the sabers.

In 2005 Patriarch Irenaeos was illegally removed from office and imprisoned in the Patriarchal compound in Jerusalem. He was never given a trial. Theophilos replaced him. I stood by Patriarch Irenaeos until his death in January 10, 2023. I am in exile. I have never been deposed or excommunicated. I have been outspoken about this naming bishops and priests that were involved. All this was to move the US Embassy from Tel Aviv to Jerusalem in 2019. Israel and the US were involved. I have no bishop at this time. I still stand by Irenaeos. According to canon law Jerusalem Patriarchate has no Patriarch. I am not a schismatic. I am in good standing with the Patriarchate of Jerusalem. I was ordained at the Holy Sepulcher in November 1998. I am in exile. I have legal standing for my actions. [ Canon 15 of the 1st/2nd Council, Rudder pg. 470, 471. ] This conspiracy had been planned since the late 90s. Because of my US intelligence background I was asked to help Patriarch Irenaeos with the false criminal complaint against him and his being deposed. I was told what happened and given volumes of files that supported the claim conspiracy and a plot to overthrow the Jerusalem Patriarchate. This was orchestrated by a group of Greek businessmen lead by US Ambassador to Qatar, Patrick Nicholas Theros and CIA Director George Tenet. Theophilos was the Jerusalem Patriarchate Bishop to Qatar and also a relative of Tenet. Theophilos was chosen to secede Irenaeos. The entire Jerusalem Synod of Bishops was involved. Constantinople was involved. Reason for the overthrow: Israel wanted the US Embassy to move from Tel Aviv to Jerusalem, the capital of Israel. It could not do that because it did not own enough land in Jerusalem to support the infrastructure for the embassy. The Jerusalem Patriarchate is the largest land owner in Israel. Theophilos sold Church land to Israeli developers which allowed the US Embassy to be moved. Theophilos continues to sell Church land to Israeli developers. Land that was on 99 year leases to Palestinians. Palestinians never wanted the US Embassy moved to Jerusalem. Jerusalem is their capital. Israel wanted the US Embassy in Jerusalem because having the US Embassy in Jerusalem solidified Israel as a legitimate nation to the world. Patriarch Irenaeos told me he would never sell land to the Israelis. He lived in a monastery in Jerusalem from the age of 13. He was told by Theophilos, if he left Israel, he would be set free. Irenaeos loved the Palestinian Christians he served as Patriarch and refused to leave the Jerusalem Patriarchal compound in which he was imprisoned. He was not allowed to attend liturgy. He was not allowed visitors. Over the years he became very ill. His teeth were so bad he removed them himself. Both his small dogs died. A Muslim friend brought him food every day. No government would help me to help Irenaeos. No bishop would help. I asked Pope Francis for help and he helped free Patriarch Irenaeos. Irenaeos was restored to Patriarch in 2019. In 2021 he was hospitalized in Greece. He died in a monastery in 2023. I visited Patriarch Irenaeos many times in Jerusalem. I spoke to him by phone often during the 15 years I served him. I lost my church. I lost my home due to financial problems associated with supporting Irenaeos. I am 73. I am Palestinian Christian. I have always been Orthodox. My only regret is that I couldn’t help Patriarch Irenaeos more. God bless you+

@stevemur @HowToAI_ This is literally what prompt engineering is 😂

@jaredef @HowToAI_ Wait. Middle out?

If you caught me on the Reversion podcast talking about AI, this is essentially the kind of apparatus that I built for my personal thought, except with way more affordances 🔥

@newolddlowen I was just talking with a friend that wanted to replicate my method, I was struggling to put it into words he could “operationalize” because my entire apparatus is just in my head. You’ve built the apparatus that my friend needs. Looking forward to launch 🚀 🔥

This is not a second brain. Not notes. Not connected .md files. Not Obsidian + Claude. AI that reasons over the most important thinking you do every day — forever. One query away. It tracks conviction, surfaces contradictions, understands your beliefs & decisions, and shows how they evolve and strengthen over time. I’ve been quietly building this for a year. The graph is just the foundation. The app / ui layer is the workspace. Could this be useful? Curious your thoughts? @DoctorYev @jaredef @nathanbenaich

Let AI build your graph. Work smart. Not hard. One keystroke → Any AI chat instantly flows into your private graph. Claude, GPT, Gemini, Grok, Perplexity… all of it. No plugins. No extensions. No APIs. Fully local. Multi-modal uploads (PDFs, images, research papers, etc.).

@8ntmuch Shore break on Big Beach, Maui