Ben Packer

1.6K posts

Ben Packer

Ben Packer

@bean_pooker

tech at @uaw & @nycdsa, for the people, against bastard messianism

Katılım Eylül 2015
1K Takip Edilen425 Takipçiler
Ben Packer retweetledi
David Orkin
David Orkin@david4queens·
We've uncovered what might be the biggest case of petition fraud in NYC's history, and now I'm taking my opponent to court.
English
32
342
2.7K
277.5K
Ben Packer retweetledi
Claire Valdez
Claire Valdez@claireforny·
For decades, we've been sold the lie that empire advances freedom and keeps us safe. Instead, it's led to violence, instability and a ruling class that reaps the profits but never pays the bill. Today we're releasing our vision for a different foreign policy.
English
284
1K
6.4K
528.2K
Ben Packer retweetledi
Drop Site
Drop Site@DropSiteNews·
💢 THREAD: Iranians form human chains around power plants and bridges as Donald Trump warns “a whole civilization will die tonight” As Trump issued his most explicit threat yet, videos circulating online show civilians gathering around power stations and key infrastructure across Iran in acts of solidarity ahead of his 8 PM deadline. 🎥 Kazerun Combined Cycle Power Plant (southern Iran)
English
120
2.8K
8.7K
1M
Tanner Linsley
Tanner Linsley@tannerlinsley·
Ghostty was fun, but time for something else. I still love opencode, too but with CC plans dead on it… I’m feeling lost. Full GUI? T3 Code? Opencode GUI? Warp? Back to cursor? Try CC again? Raw Codex? My 🧠 hurts and I just need to keep shipping.
English
411
9
1.3K
336.6K
Ben Packer retweetledi
Sen. Bernie Sanders
Sen. Bernie Sanders@SenSanders·
We need a moratorium on AI data centers NOW. Here’s why.
English
991
1.7K
9.2K
1.3M
Ben Packer
Ben Packer@bean_pooker·
@duckdb Awesome! Will there be Vortex support in Ducklake?
English
0
0
1
26
DuckDB
DuckDB@duckdb·
DuckDB now supports reading from and writing to the Vortex file format! The DuckDB Labs and Spiral teams have worked together to make Vortex available as a core extension in DuckDB. Vortex is an open source, columnar file format whose design is heavily influenced by recent research in lightweight compression encodings, computing and IO techniques. We gave it a test drive, and it performed very well. Read the full article to learn more lnkd.in/eZfGzPiZ
DuckDB tweet media
English
6
37
312
23.7K
Ben Packer retweetledi
dax
dax@thdxr·
ok more clarity - good programming looks like structuring things in a way where certain bugs are impossible LLMs are not good at this they just accomplish the primary task
English
28
23
670
27.6K
Ben Packer retweetledi
Pekka Enberg
Pekka Enberg@penberg·
Shipped AgentFS! We built the first iteration of the missing abstraction for agents people have been asking for. Everything in one SQLite file: 💾POSIX-like filesystem for artifacts 📦 Key-value store for state 🔍 Toolcall tracking Full blog post: turso.tech/blog/agentfs
Pekka Enberg@penberg

SQLite is the perfect database for AI agents—but nobody knows how to use it. Many agent developers I've talked to make the same point: SQLite is ideal for capturing agent state, handling auditing, ensuring compliance, and managing snapshots. It's lightweight, all data represented in one single file. But when I ask them what the interface should look like, nobody can really answer. What should the schema look like? How do we make it work across different agent frameworks? How do we handle the filesystem transparently? After some iteration and conversations with folks building agents, I've identified the three missing pieces: 1. A standardized schema specification. We need a common way to represent agent state, actions, and outcomes. This isn't just about storing data—it's about enabling an ecosystem of tooling. When everyone uses the same schema, we can build debugging tools, replay systems, and compliance analyzers. 2. A programmatic SDK. Raw SQL obviously isn't enough. Developers need type-safe APIs, automatic migrations, and framework integrations. The SDK should feel native whether you're using a framework or rolling your own agent system. 3. A transparent filesystem layer. People building agent compute infrastructure actually want a transparent agent filesystem that works with bash, curl, and friends. Agents shouldn't have to think about where their data lives. They just read and write files, and our system handles the rest. That why me and @glcst at @tursodatabase have been hacking on AgentFS (github.com/penberg/agentfs)—a specification and implementation for agent state management. If you're building agents and fighting with state management, I'd love to hear from you. And if you want to help shape this standard, check out the repo and join the discussion.

English
13
27
239
55K
Ben Packer
Ben Packer@bean_pooker·
@ekzhang1 I'd love to join the reading group for this month? Open for November?
English
0
0
0
15
Eric Zhang
Eric Zhang@ekzhang1·
NYSRG is back for November, this month we'll be reading about data! Personally looking forward to this because there's been a tremendous amount of data systems and adjacent work in the past few years :) We will try out some new meeting locations / hosting in NYC
Eric Zhang tweet media
English
3
4
128
8K
Ben Packer
Ben Packer@bean_pooker·
@jarredsumner I don't think anyone cares about the req/s of workersd on a static machine. It's about: - the latency of a single request without significant load - the overall "cost to serve 100k requests", for example, matching a specific traffic pattern
English
0
0
0
26
Jarred Sumner
Jarred Sumner@jarredsumner·
React SSR on a Linux x64 hetzner machine Bun v1.3.1: 131,112 requests per second CF Workers: 21,878 requests per second
Jarred Sumner tweet media
English
56
34
992
100.6K
Ben Packer
Ben Packer@bean_pooker·
@BigMeanInternet I think I like that what does it mean just a little bit more precisely?
English
0
0
0
190
Malcolm Harris
Malcolm Harris@BigMeanInternet·
Thank you Gustavo Petro for saying the inescapable strategy out loud: Win the global civil war.
English
5
78
721
27.9K
Ben Packer
Ben Packer@bean_pooker·
@ryangrim and then when they do, transition the app to theirs, and after that continue to tweak the weights and training process. Until they complete retraining on just US user data, they wouldn’t be in a position to push updates, is my understanding
English
0
0
0
4
Ben Packer
Ben Packer@bean_pooker·
@ryangrim It seems from the next few paragraphs that they’ll get a copy of the algorithms code and access to run it but will not get its trained weights since that comes from user data outside the US. But, they’ll be able to retrain the algorithm on just US user data (cont)
English
1
0
0
10
Ryan Grim
Ryan Grim@ryangrim·
So according to the Times’ new TikTok story, “Oracle will oversee the security of Americans’ data and monitor changes and updates to its powerful recommendation technology.” So they can look at the algorithm but not mess with it? Huh? Any tech folks who can parse this?
English
48
65
619
44.3K
Ben Packer
Ben Packer@bean_pooker·
@simonw I have been thinking of building a gpt-4o realtime wrapper for Claude Code but haven’t gotten around to it for exactly this purpose (coding while doing dishes)
English
0
0
0
12
Simon Willison
Simon Willison@simonw·
Anyone got a good mechanism on macOS for completely hands-off Claude Code voice control? I've tried Wispr Flow but you still need to press a key to start and stop each transcription snippet I want to spend an hour telling Claude Code what to do vocally completely hands free
English
113
22
750
195.3K
Ben Packer
Ben Packer@bean_pooker·
@jasonpjason we called this AI when I was an undergrad (AI means path search / constraint solving)
English
0
0
1
17
software crisis actor
software crisis actor@jasonpjason·
i explained my job (building air traffic control systems with software) to an undergrad and they said oh so you are building an AI. no. no it is not AI. safety-critical systems are the last place AI should be involved. let’s stop calling things AI.
English
1
0
9
237