1.
Cloudflare Durable Objects Now Survive Active WebSockets
Keeping Durable Objects alive for active WebSockets fixes a real failure mode for long-running streams and agent runtimes
Backlist — 19 Jun 2026 UTC
Tilted toward concrete infra, market structure, and durable artifacts; trimmed most meme and quote-tweet noise.
2.
ASML Denies an EUV Machine Reached China
A reported export-control breach around the most important lithography tool in the world would have major semiconductor implications
3.
AI Capital Is Becoming Extremely Concentrated
Models, compute, data, and capital are pooling into a small number of firms, changing the structure of the industry
4.
Why VC Struggles to Fund Hard Tech
Securitizing R&D could let pension funds finance fusion and other long-horizon projects at bond-market scale
5.
A Datacenter Built Its Own Gas Plant
One operator chose on-site generation over waiting for grid interconnection, which says a lot about infrastructure bottlenecks
6.
China's Manufacturing Boom Hits Low-Tech Exporters Too
The trade debate is usually about rich-country incumbents, but the bigger welfare hit may land on poorer countries that cannot compete
7.
Twenty Raises $100M for a Cyber Warfare Startup
A large round for an offensive cybersecurity company shows how fast the defense market is moving
8.
A Corpus of 2.2 Million Local Laws (x.com)
A newly released dataset makes almost every city and county law in the U.S. searchable at scale
9.
Ukraine Opens Captured Russian Weapon Tech to Partners
Battlefield captures are being turned into a shared intelligence and reverse-engineering platform
10.
Why Databricks and Confluent Exist
Running Spark and Kafka yourself at scale is hard enough to support major standalone companies
11.
Server Prices Have Jumped From $40k to $100k
Hyperscale demand is so strong that even nice-to-have internal projects are getting delayed by hardware inflation
12.
Elliott and SVP Buy Braskem Debt
Distressed debt buyers are positioning for a restructuring fight around one of Brazil's largest petrochemical producers
13.
Redis Claims 50% Memory Savings on Sorted Sets
A deep data-structure optimization like this can materially change the cost profile of real production workloads
14.
Binius64 Adds Zero-Knowledge Mode
Binary-field systems are moving from speed demos to privacy-preserving applications
15.
Active Theory Learning for Automated Science (x.com)
The pipeline learns interpretable mechanistic models from data and then chooses experiments to test them
16.
AlphaFold 3 Turns Out to Be Simpler Than AlphaFold 2
A major biological model appears to have a much simpler internal structure than people expected
17.
Bacterial DNA as a Data Store
A startup says it can write and read data from DNA in a tiny lab setup, which is a real milestone for molecular storage
18.
AI Chip-Design Startups Have a Revenue Problem
Great demos are not the same as repeatable sales when the underlying EDA tool market is structurally constrained
19.
Agents Are Reinvigorating Good Software Habits
Open APIs, docs, tests, CLIs, and structured outputs suddenly matter again when software is built and consumed by agents
20.
Agent Wallets Need Refunds and Spend Policies
Letting software pay is easy; making it safe under CFO scrutiny is the real product problem
21.
Harness Engineering Notes From Shipping Real Agents
Model-task fit matters enough that harness design becomes a first-class engineering discipline
22.
Gaussian Splats From Just Four Images
If this holds up, it lowers the barrier for high-quality 3D capture dramatically
23.
A New $10–30M RFP on AI's Impacts (x.com)
A grant-maker is putting serious money behind work on the economic, health, and scientific effects of AI
24.
A Missing Book About Great Committees
Some of history's most effective institutions were committees, not lone geniuses
25.
Wearables Replace the Hard Question With the Easy One (x.com)
The critique lands because quantified-self products can optimize what is measurable while ignoring what actually matters
26.
A Dog Made This Map Without Drones or Satellites
The result is a reminder that surprisingly rich spatial data can come from simple sensors and a walk
27.
Urukka Veedu, a House Rooted in the Land
The project is a clean example of architecture that treats the building as an extension of place rather than a separate object
28.
DIY Instructions Are Really ASTs
Reframing instructions as structured state machines explains why some procedures are easy for humans and agents to follow
29.
1973 Istanbul TVs Cost More Than Twice German Ones
The price gap is a neat snapshot of how local assembly and protectionism can distort consumer goods markets
30.
Austria's Nuclear Plant Was Finished, Then Banned
A completed 692MW plant never opened after a referendum, making it a classic case of political reversal
31.
More money/compute != better models
More money/compute != better models Even GDM has now fallen behind as Anthropic storms fwd with top tier data from all tough prompt it has received as Claude Code dominated professional usage. Meta turned its engineers into engineering data
32.
Excited to wake up to a meeting invite from Noam! (x.com)
Excited to wake up to a meeting invite from Noam! Reminder: OpenAI is also hiring top pre-training researchers in London. Come work with me @SavinovNikolay and the team on architecture, long-context, science of scaling, and more! Conta
33.
Exit Alpha USDC Delta V2 on Morpho immediately! It has 17M exposure to the msY (Main Street Yield) market. This m…
Exit Alpha USDC Delta V2 on Morpho immediately! It has 17M exposure to the msY (Main Street Yield) market. This market is fully utilized for a couple of days already, with borrowers paying 100% apy. Strangely, big borrowers aren't repaying.
34.
essentialai to nvidia is what i heard
essentialai to nvidia is what i heard
35.
kids be rewriting history again
kids be rewriting history again
36.
GLM-5.2 delivers a substantial leap in app development capabilities, which also represent demanding long-horizon …
GLM-5.2 delivers a substantial leap in app development capabilities, which also represent demanding long-horizon tasks. Results: - GLM-5.1: 21/70 - GLM-5.2: 48/70 - Claude Fable 5: 56/70 That's more than a twofold improvement from GLM-5.1
37.
GLM 5.2 (max) scores 70.1% on WeirdML, narrowly beating to Genini 3 Pro, from 7 months ago.
GLM 5.2 (max) scores 70.1% on WeirdML, narrowly beating to Genini 3 Pro, from 7 months ago. It uses ~22k output tokens on average, compared to ~12k for the (high) setting. This gives a fairly clear but modest increase (3%) in score, showi
38.
Riverton House by Alexandra Buchanan Architecture
Riverton House by Alexandra Buchanan Architecture Chelmer, Australia
39.
not sure if old job posting or not but this is an interesting tidbit
not sure if old job posting or not but this is an interesting tidbit "Practical experience and related awards in security areas such as CTF competitions and software vulnerability discovery)." DeepSeek is actively developing and wants to
40.
EssentialAI seems to have been acquihired
Editor’s note: imported_from_x_likes
EssentialAI seems to have been acquihired also crazy to find this out from linkedin, i checked and there is 0 information about it (they were valued around ~$1B in August 2025)
41.
i didn't see anyone post about this, but new deepseek domain was registered that routes to a job posting
i didn't see anyone post about this, but new deepseek domain was registered that routes to a job posting bench(dot)deepseek(dot)com
42.
They realize in Asia they do this with college interns right?
They realize in Asia they do this with college interns right?
43.
Your occasional reminder that causal decision theory agents will pay not to know all sorts of information. Eg, p…
Your occasional reminder that causal decision theory agents will pay not to know all sorts of information. Eg, pay not to be told about an Adversarial Offer (Oesterheld and Conitzer).
44.
Intel with another huge hire. Former SK hynix CEO
Intel with another huge hire. Former SK hynix CEO $INTC
45.
More than a year ago (x.com)
More than a year ago @TacoCohen in SPO https:// arxiv.org/abs/2503.05453 already derives a critic-free value/Q parameterization from the policy-reference log-ratio under KL-regularized RL by exactly “deriving value loss through policy ra
46.
Vlad was not a hero,
Vlad was not a hero, Vlad was a Christian killer. He invited thousands of Orthodox Christian Wallachian boyars to an Easter feast and impaled them. Indiscriminately slaughtering women, children, and the elderly, he massacred about 30,00
47.
If you spend too long consuming without producing, you fall into the state of "blah" -- restless yet too tired to…
If you spend too long consuming without producing, you fall into the state of "blah" -- restless yet too tired to do anything. The fix is active production, not more passive consumption.
48.
This is a challenge unique to carve-outs, isn't it?
This is a challenge unique to carve-outs, isn't it? Even for a carve-out by Supporters, who originally knew the business inside out, it still took considerable cost and effort. The transition of systems and security is one thing, but phys
49.
Josh Home Ownership by YIWUJIE Design
Josh Home Ownership by YIWUJIE Design Guangdong Province, China
50.
チャットプラスという会社が先週、上場承認を受けた。
チャットプラスという会社が先週、上場承認を受けた。 従業員は22人。想定時価総額は約49億円。売上10.2億、経常利益率36%。社長は大手建設会社出身の女性だ。
51.
Every dev tool that has historically had a steep learning curve is now usable again with agents.
Every dev tool that has historically had a steep learning curve is now usable again with agents. Especially if 'in distribution'.
52.
what backpack has done on solana for stocks is historic
what backpack has done on solana for stocks is historic onchain stocks are adopting the stablecoin model where minting is kyced but transfers are permissionless this model will win imo as it allows everyone with an internet connection to
53.
黄金大门对于在后台偷跑的软件有强制关闭的逻辑了。
黄金大门对于在后台偷跑的软件有强制关闭的逻辑了。
54.
Turns out zstd compression can be used as a language model. It can learn language structure, new words, capture s…
Turns out zstd compression can be used as a language model. It can learn language structure, new words, capture semantic similarity of questions and produce the right answer, and write slightly gibberish poetry. Very lightweight, can run o
55.
Incredible how Z. ai literally has their RL infrastructure open source. (t.co)
Incredible how Z. ai literally has their RL infrastructure open source. The entire OPD post-training of GLM-5.2 took on this slime platform took ~2 days. https:// github.com/THUDM/slime
56.
U-Net
U-Net
57.
People in finance do not dress as they do because it's "part of the forgiveness for wealth in finance"
People in finance do not dress as they do because it's "part of the forgiveness for wealth in finance" People who go into finance do not believe wealth is something that requires forgiveness at all They dress as they do because *absolutel
58.
フランチャイズ中心から直営中心に切り替えた理由は、きちんと説明できた方がいい。
フランチャイズ中心から直営中心に切り替えた理由は、きちんと説明できた方がいい。 直営は重い。 でも、収益性・品質管理・オペレーション改善の自由度が高い。 買い手から見ると、 「コントロールできる成長モデル」 は魅力になる。 戦略転換は、理由までセットで価値になる。
59.
Warriors fan wife: The Knicks winning has made everyone in New York nice for a week. It's like they're microdosin…
Warriors fan wife: The Knicks winning has made everyone in New York nice for a week. It's like they're microdosing California.
60.
I think you should at least have the decency to go through the code overview yourself and understand the architec…
I think you should at least have the decency to go through the code overview yourself and understand the architecture of what you are proposing. Don't mind agent made PRs as long as the human can write the description
61.
OP is right, anyone can pull off a bowl with pulsating Xenia. They live in the equivalent of saltwater swamps (th…
OP is right, anyone can pull off a bowl with pulsating Xenia. They live in the equivalent of saltwater swamps (the reason they pulsate is to create local water circulation where there is none) and are almost impossible to kill.
62.
Also been commissioned to make my design into a book cover a family WC-quiz book
Also been commissioned to make my design into a book cover a family WC-quiz book First book cover I’ve made, and hopefully not the last So if you’re in Norway and sees this, you know who designed the cover
63.
Our experience at Daydream exactly. (x.com)
Our experience at Daydream exactly. When you are measured on the % of trace failures and not “overall intelligence”, the whole game is different. We spend a significant effort into developing models to decide when we should pull a huma
64.
one of my favourite past-times is visiting my friend’s houses! i made a simple digital map so you can visit some …
one of my favourite past-times is visiting my friend’s houses! i made a simple digital map so you can visit some of them too :-)
65.
Edition 31 of the Canonical Document of a long-running hardcore history MUD I built using Claude and a contextual…
Edition 31 of the Canonical Document of a long-running hardcore history MUD I built using Claude and a contextual knowledge graph across a few repos Marquis de Canossa has finally received the Order of the Golden Fleece during the Spanish
66.
This paper trains RLVR reasoning models on token-level distributional deviations rather than uniform token update…
This paper trains RLVR reasoning models on token-level distributional deviations rather than uniform token updates, to avoid the entropy collapse that uniform updates cause. RLVR improves reasoning but suffers an optimization instability:
67.
why is the mac spotlight app so horribly useless
why is the mac spotlight app so horribly useless why is the top result for "codex" 5 random pictures and a dictionary results and not the codex app that i use every single day.
68.
They tried with golf, did not work
They tried with golf, did not work Now, they moved on to soccer
69.
today i'm launching Flashtype, a markdown editor for Claude & Codex
today i'm launching Flashtype, a markdown editor for Claude & Codex use Claude/Codex to edit markdown, track every change, and see the diffs open source. available now.
70.
Hi Jim! Thanks for the analysis but you're missing a couple of things that I think would help a lot!
Hi Jim! Thanks for the analysis but you're missing a couple of things that I think would help a lot! You're using H100 and H800 hours for estimating your multiplier. H100 hours and H800 hours are not equivalent so doing a raw division does
71.
If GPT 5.6 still requires me to say "no fallbacks please" I think I'll crash out.
If GPT 5.6 still requires me to say "no fallbacks please" I think I'll crash out.
72.
How is a crypto system - with no deep web2 relationships or distribution - going to compete with this?
How is a crypto system - with no deep web2 relationships or distribution - going to compete with this?
73.
add a little joy when removing items.
add a little joy when removing items. .removing-item .container { filter: blur(8px); opacity: 0; transform: scale(0.92); transition: filter 280ms ease-in, opacity 200ms ease-in, transform 280ms ease-in;
74.
I'm a weird Greek guy who instead of studying for his uni exams decided he should watch early speeches of Obama f…
I'm a weird Greek guy who instead of studying for his uni exams decided he should watch early speeches of Obama from 2004-2009 to see what all the hype was about He was kinda.....not good?
75.
He refuses to acknowledge that metas strength is acquisition moreso than it is unique development
He refuses to acknowledge that metas strength is acquisition moreso than it is unique development Also jumping off the cliff with the crazy layoffs in my opinion was the wrong move
76.
A year ago, with a mix of excitement and fear, I sent the largest term sheet of my life to an unstoppable founder…
A year ago, with a mix of excitement and fear, I sent the largest term sheet of my life to an unstoppable founder determined to change history. I knew if I didn’t step up, the company might not survive — the ambition was too bold. It’s wild
77.
Nippon Life, Japan's strongest big unicorn hunter lol
Nippon Life, Japan's strongest big unicorn hunter lol > The company is expected to gain investment income of 100 billion to 500 billion yen through its initial public offering (IPO)
78.
My best guess is that Anthropic has cracked recursive self improvement, which is why the top talent wants to be t…
My best guess is that Anthropic has cracked recursive self improvement, which is why the top talent wants to be there
79.
Exact workflow/prompts.
Exact workflow/prompts. GPT-5.6 Pro in ChatGPT can't generate images. I had to use GPT-5.5 to generate the images, then GPT-5.6 to build the site with placeholder images, then GPT-5.5 again to fill the images in with real ones. I've teste
80.
This framing resonates a lot.
This framing resonates a lot. If the agent is the wedge and the network is the moat, then supplier-side gateways become critical too. Procurement agents need a structured way to contact suppliers. Suppliers need a controlled way to respo
81.
In Chinese we have a saying, 画蛇添足. It means adding feet to your drawing of a snake (doing more work to make thing… (x.com)
In Chinese we have a saying, 画蛇添足. It means adding feet to your drawing of a snake (doing more work to make things worse). Someone had to decide to write code to add this validation. cc @zapmail_ai
82.
Takeaways so far:
Takeaways so far: - It LOVES smooth scrolling and scrolljacking techniques. I'm not a fan of how it feels. (Admittedly I did ask for motion and transitions while scrolling, but you can do that in a way that feels good) - It insists on the
83.
crazy how much companies are willing to pay for someone who can actually do the work AND pull attention to it wit…
crazy how much companies are willing to pay for someone who can actually do the work AND pull attention to it without drifting into clout‑chasing extremely hard to find, because you need someone who’s extremely online and yet can maintain
84.
"Some neo-clouds worry that they can’t stray from buying Nvidia’s full stack of hardware for fear of being put in…
"Some neo-clouds worry that they can’t stray from buying Nvidia’s full stack of hardware for fear of being put in “Jensen jail,” meaning they might lose their allocations of Nvidia chips, said Adam Fisher, a partner at Bessemer Venture Part
85.
this might be my favorite video ive ever made. i was making web visualizers for every part of our stack and i add…
this might be my favorite video ive ever made. i was making web visualizers for every part of our stack and i added a zoom and said "whoa this looks like a Max Cooper music video" then i got license to a Max Cooper song, added some technica
86.
Parcel’s file watcher is quietly powering many of your favorite tools including VSCode, Tailwind, Nx, Nuxt, etc. …
Parcel’s file watcher is quietly powering many of your favorite tools including VSCode, Tailwind, Nx, Nuxt, etc. Another one
87.
I think zuck has lost it
I think zuck has lost it
88.
there seems to be a sense in anthropic employees that they will recover a lot of data about the world by spending…
there seems to be a sense in anthropic employees that they will recover a lot of data about the world by spending tokens in simulation it doesn’t make sense to me, it seems cheaper to purchase data, but it has me intrigued because i keep h
89.
yeah, they're one of the worst UI patterns of the last few years, up there with iOS's "long press to open the pas…
yeah, they're one of the worst UI patterns of the last few years, up there with iOS's "long press to open the paste menu"
90.
How to make a team of co-ordinated AI agents:
How to make a team of co-ordinated AI agents: - set up your first preferred agent (it's your orchestrator) - ask it to configure Gemini Managed Agents or something like modal cpu instances to spin up sub agents in their own environment - k