@BryanHirsch

Present: https://xitter.cf/t.co/5radCJsI4z. Past: Chief Digital Officer & Founder of MA Digital Service, @Whitehouse44. LinkedIn: https://xitter.cf/t.co/XUsgB5Imys RT ≠ +

 Boston, MA
bryanhirsch.substack.com
Joined April 2009
  • Tweets1.8K
  • Following1.8K
  • Followers863
64Media
Bryan Hirsch retweeted
Simon Willison
@simonwOct 5
I re-ran an experiment @colin_fraser ran against GPT-4o a while back to see how good it was at adding long numbers, only this time I tried Qwen 3.8 27B running locally in both reasoning and non-reasoning modes simonwillison.net/2026/Oct/4…
Heatmap in the same layout as the previous chart, using an orange (low) to white to blue (high) color scale, showing much lower accuracy overall. Title: Addition in words — Qwen3.8 27B Q4_K_M. Subtitle: Reasoning disabled · 30 fixed pairs per ordered digit-length cell (n = 5,070). Overall numeric accuracy: 1,195 / 5,070 (23.57%). X axis: Number of digits in a, 1 to 13. Y axis: Number of digits in b, 1 to 13. Legend: Accuracy, 100%, 75%, 50%, 25%, 0%. Values by row, listed for a = 1 to 13. b = 13: 17%, 13%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%. b = 12: 53%, 20%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%. b = 11: 47%, 10%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%. b = 10: 70%, 27%, 3%, 0%, 0%, 0%, 0%, 0%, 0%, 13%, 0%, 0%, 0%. b = 9: 77%, 47%, 3%, 0%, 0%, 0%, 0%, 3%, 7%, 0%, 0%, 0%, 0%. b = 8: 53%, 20%, 0%, 0%, 0%, 0%, 7%, 13%, 0%, 0%, 0%, 0%, 0%. b = 7: 53%, 23%, 17%, 10%, 3%, 3%, 13%, 3%, 0%, 0%, 0%, 0%, 0%. b = 6: 60%, 60%, 33%, 10%, 53%, 47%, 7%, 3%, 0%, 0%, 0%, 0%, 0%

ALT Heatmap in the same layout as the previous chart, using an orange (low) to white to blue (high) color scale, showing much lower accuracy overall. Title: Addition in words — Qwen3.8 27B Q4_K_M. Subtitle: Reasoning disabled · 30 fixed pairs per ordered digit-length cell (n = 5,070). Overall numeric accuracy: 1,195 / 5,070 (23.57%). X axis: Number of digits in a, 1 to 13. Y axis: Number of digits in b, 1 to 13. Legend: Accuracy, 100%, 75%, 50%, 25%, 0%. Values by row, listed for a = 1 to 13. b = 13: 17%, 13%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%. b = 12: 53%, 20%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%. b = 11: 47%, 10%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%, 0%. b = 10: 70%, 27%, 3%, 0%, 0%, 0%, 0%, 0%, 0%, 13%, 0%, 0%, 0%. b = 9: 77%, 47%, 3%, 0%, 0%, 0%, 0%, 3%, 7%, 0%, 0%, 0%, 0%. b = 8: 53%, 20%, 0%, 0%, 0%, 0%, 7%, 13%, 0%, 0%, 0%, 0%, 0%. b = 7: 53%, 23%, 17%, 10%, 3%, 3%, 13%, 3%, 0%, 0%, 0%, 0%, 0%. b = 6: 60%, 60%, 33%, 10%, 53%, 47%, 7%, 3%, 0%, 0%, 0%, 0%, 0%

Heatmap in the same layout as the previous charts, almost entirely blue. Title: Addition in words — Qwen3.8 27B — medium reasoning pilot. Subtitle: 1 fixed pair per ordered digit-length cell · easiest first (n = 169). X axis: Number of digits in a, 1 to 13. Y axis: Number of digits in b, 1 to 13. Legend: Accuracy, 1.00, 0.75, 0.50, 0.25, 0.00. Every cell shows 100% except two orange cells showing 0%: a = 2 with b = 8, and a = 12 with b = 9.

ALT Heatmap in the same layout as the previous charts, almost entirely blue. Title: Addition in words — Qwen3.8 27B — medium reasoning pilot. Subtitle: 1 fixed pair per ordered digit-length cell · easiest first (n = 169). X axis: Number of digits in a, 1 to 13. Y axis: Number of digits in b, 1 to 13. Legend: Accuracy, 1.00, 0.75, 0.50, 0.25, 0.00. Every cell shows 100% except two orange cells showing 0%: a = 2 with b = 8, and a = 12 with b = 9.

116
29
10
478
53,267
Bryan Hirsch retweeted
DHH
@dhhOct 4
Every developer needs an AI shed: An always-on machine running on their tailscale network where the majority of their herdr agents are running.
627
357
206
7,761
654,594
Bryan Hirsch retweeted
Gergely Orosz
@GergelyOroszSep 29
Oh my god, Firebase’s SDK started to crash ALL iOS apps that were using it, for ALL sessions, just like that, no way for any of these apps to do anything… How amateur is all of this from any SDK, but especially one from a company with as high of an engineering bar like Google…
Vijay Tholpadi@VijayTholpadiSep 29
github.com/firebase/firebase… incase anyone has been wondering why their iOS apps are intermittently crashing since today morning 🙃 reminds me of the Facebook swizzling crash incident. @Firebase pls help

[Resolved] Crash in Analytics after sdk-exp response: -[__NSDictionaryM setObject:forKeyedSubscri...

Description Since 2026-09-29 00:41 UTC, our production iOS app has been crashing at a growing rate (56 crashes / 56 users in the first 18 minutes). We have not shipped an update. The crash started ...

github.com
143
200
98
2,811
608,491
This TUI for browsing markdown files is soooo useful! github.com/sanford/lsmd

GitHub - sanford/lsmd: lsmd - a terminal-friendly Markdown (.md) reader built for speed and easy...

lsmd - a terminal-friendly Markdown (.md) reader built for speed and easy navigation in large projects - sanford/lsmd

github.com
29
Bryan Hirsch retweeted
Gergely Orosz
@GergelyOroszSep 24
Bury your head in the ground at your own risk. I aim to not jump on any hype trains, but since Opus 4.6 and GPT-5.2 + the harnesses it was clear that these things can write code nearly as good as I can in my best language; better in other languages. But we won't see non-devs push production code for a long time; possibly forever, if you ask me. Software still needs to be built in robust ways, and it's our profession to do this well, and use the new tools we have, which create new and interesting (+tough!) challenges
76
83
23
1,576
339,590
Bryan Hirsch retweeted
Simon Willison
@simonwSep 25
The more time I spend working with coding agents, the more convinced I am that they make software engineering even harder We can do amazing things with them, but unlocking their full potential requires extraordinary discipline and knowledge
Gergely Orosz
@GergelyOroszSep 24
Replying to @GergelyOrosz
Bury your head in the ground at your own risk. I aim to not jump on any hype trains, but since Opus 4.6 and GPT-5.2 + the harnesses it was clear that these things can write code nearly as good as I can in my best language; better in other languages. But we won't see non-devs push production code for a long time; possibly forever, if you ask me. Software still needs to be built in robust ways, and it's our profession to do this well, and use the new tools we have, which create new and interesting (+tough!) challenges
379
390
109
4,476
624,718
Bryan Hirsch retweeted
DHH
@dhhSep 14
Omarchy is a perfect fit for the iPad, because tiling already makes the most sense. What an awesome setup!
Seeking Gradient
@seekinggradientSep 14
Replying to @seekinggradient
Okay, here's what I've been working on: Cloud Window 🙂 getcloudwindow.com After the last demo, a bunch of you asked how to try it. I've been making the setup easier, and the first DIY release is now available. An Omarchy desktop in your own AWS account, streamed to your iPad, phone, or laptop. Same files, same apps, same session. I really love using this on the iPad. Here's a look at the experience, with the setup walkthrough below.
65
52
2
1,343
169,905
Bryan Hirsch retweeted
Rohit
@HeyRohhitSep 14
In 1997, a developer publicly told Steve Jobs he had no idea what he was talking about. Jobs paused and answered with the most important business lesson ever delivered under pressure. Bookmark & watch today, no matter what.
164
1,417
83
14,277
1,026,623
Bryan Hirsch retweeted
Simon Willison
@simonwSep 10
Here's a Blender model of a Pluribus themed Fabergé egg I had GPT-6 Astra build - I generated the concept image using ChatGPT Images 2.5, then pasted the image into Codex and told Astra to turn that into a Blender file Details here: simonwillison.net/2026/Sep/9…
Screenshot of a web-based 3D Blender file viewer showing a rendered gold Fabergé-style jeweled egg. The top of the egg is lifted open, revealing a miniature desert diorama inside with green cacti, brown rocks, and a small black-suited figure standing in the center. The egg's shell is decorated with green gems, pearls, ornate scrollwork, a ring of smiling white egg-shaped faces, and oval medallions painted with a cactus and desert scene. The egg stands on four gold legs over a black marbled base with a gold plaque reading PLURIBUS. The URL bar reads https://github.com/simonw/vibe-coded-blender-projects/blob/main/pluribus-faberge-egg/deliverables/Pluribus_Jeweled_Egg_v1.blend with a View button; top-right buttons read Fit, Wireframe, Clear. A bottom-left info panel reads: Loaded Blender 05.01 · 7.2 MB / 387 meshes · 1099 curves · 1 text · 783,764 vertices · 1,446,560 triangles · 17 materials / Preview: approximate materials and text; unapplied modifiers are omitted. / GitHub URL resolved

ALT Screenshot of a web-based 3D Blender file viewer showing a rendered gold Fabergé-style jeweled egg. The top of the egg is lifted open, revealing a miniature desert diorama inside with green cacti, brown rocks, and a small black-suited figure standing in the center. The egg's shell is decorated with green gems, pearls, ornate scrollwork, a ring of smiling white egg-shaped faces, and oval medallions painted with a cactus and desert scene. The egg stands on four gold legs over a black marbled base with a gold plaque reading PLURIBUS. The URL bar reads https://github.com/simonw/vibe-coded-blender-projects/blob/main/pluribus-faberge-egg/deliverables/Pluribus_Jeweled_Egg_v1.blend with a View button; top-right buttons read Fit, Wireframe, Clear. A bottom-left info panel reads: Loaded Blender 05.01 · 7.2 MB / 387 meshes · 1099 curves · 1 text · 783,764 vertices · 1,446,560 triangles · 17 materials / Preview: approximate materials and text; unapplied modifiers are omitted. / GitHub URL resolved

69
15
9
156
26,557
Bryan Hirsch retweeted
BOXABL
@BOXABLAug 25
Imagine a future where we could build homes like cars — one every minute. That's the vision of @BOXABL Co-CEO @GalianoTiramani Homes built on an assembly line, delivered to you. Starting at $60,000. Nasdaq ticker symbol $BXBL* Read disclaimers boxabl.com/ir

1,499
2,012
149
18,756
56,064,446
Bryan Hirsch retweeted
Mitchell Hashimoto
@mitchellhAug 21
Ghostty (and libghostty!) now has a 100% complete implementation of the Kitty Graphics Protocol. We support every feature, every option. We are the first terminal beyond Kitty to do this. Most terminals stop at basic image placements, then give up on the long tail of other features. Ghostty covers all of them: - Unicode placeholders (so it works in tmux) - Relative placements (so images can be positioned relative to other images) - Animation - Scroll region handling (so images scroll properly with terminal scroll regions) - All transport mediums: shared memory, temp files, etc. - etc... All of these features don't impact steady-state use (CPU or memory, besides raw code size). And for libghostty consumers, you can use the compile-time features flags to disable them completely. I read the spec myself a hundred times. I read Kitty's source code to validate behaviors (our implementation is markedly different, so no GPL violation). I also utilized agents, primarily for validation. I spun Fable and Sol in a loop for 3 weeks (pausing only when their dedicated subs exhausted limits) with a prompt to find mismatches between the spec, Kitty's reference implementation, and Ghostty. And now we're done. Ghostty is the most compliant Kitty Graphics implementation (ignoring Kitty itself of course). And, this is all in libghostty, so now everyone can be perfectly compliant.
82
143
16
3,697
269,694
Bryan Hirsch retweeted
Simon Willison
@simonwAug 20
I had Claude Code for web experiment with smolvm as a code execution sandbox Fable 5 spotted that its environment couldn't run that (no /dev/kvm)... so, without asking me first, it wrote a GitHub Actions workflow to run the experiments and pushed that directly to GitHub instead!
141
25
13
295
62,042
Bryan Hirsch retweeted
Nate
@natebjonesAug 17
PSA: If you are tired of Claude-lish or Chat-lish, tell your AI to read the Google Developer Docs Style Guide and build a skill that follows it It’s like magic. Suddenly your AI will make sense again! I’ve tried both and this is much better than the ASD-STE100 Simplified Technical English skill And friends at the labs: please invest more in AI comms skills! A brilliant agentic AI that cannot speak English is dramatically less useful and undermines the utility of RLVR developers.google.com/style

About this guide  |  Google developer documentation style guide  |  Google for Developers

Editorial style guidelines for public-facing Google developer documentation.

developers.google.com
113
218
30
4,175
317,360
Bryan Hirsch retweeted
Kent C. Dodds 🐨
@kentcdoddsAug 6
Really rough day for folks at @github :(
Browser screenshot of a GitHub Status “Incident with Actions” alert. The update says GitHub Actions webhook triggers are throttled, only about 15% of webhooks are being processed, and roughly 65% of queued jobs are succeeding; GitHub-hosted and self-hosted runners, Copilot features, and Enterprise Importer migrations may also be affected. A browser search box highlights “we are continuing to.”

ALT Browser screenshot of a GitHub Status “Incident with Actions” alert. The update says GitHub Actions webhook triggers are throttled, only about 15% of webhooks are being processed, and roughly 65% of queued jobs are succeeding; GitHub-hosted and self-hosted runners, Copilot features, and Enterprise Importer migrations may also be affected. A browser search box highlights “we are continuing to.”

57
30
6
555
95,102
Bryan Hirsch retweeted
Dave Guarino
@allafarceAug 5
It might be more dangerous than it’s ever been previously to be incurious right now.
1
1
10
742
Bryan Hirsch retweeted
Simon Willison
@simonwAug 4
I try not to get excited about models before they've been released, but I gotta admit I'm very much looking forward to the upcoming laptop-sized Qwen 3.8 models
Qwen Developers
@QwenDevsAug 3
Replying to @ddalcu
We promise this 27B comes with a whole new level of capability!
60
68
3
1,844
129,066
Bryan Hirsch retweeted
Matt Pocock
@mattpocockukAug 4
I fucking love the new batch version of /grill-me Makes smashing through easy questions so much faster
56
22
1
1,055
104,844
Claude is a game changer for enjoying free summer Shakespeare! I had @claudeai create this primer before taking my wife to see Midsummer Night’s Dream by @commshakes. Wow! What a different experience it is to have context and understand what’s going on! bryanhirsch.github.io/primer…
1
77
Bryan Hirsch retweeted
Thariq
@trq212Jul 24
We removed ~80% of the Claude Code system prompt for our newest models, this is what we've learned about writing system prompts, skills and Claude.MDs for them.

The new rules of context engineering for Claude 5 models

Article · @trq212
463
1,899
705
16,288
4,942,698
Bryan Hirsch retweeted
Simon Willison
@simonwJul 23
I think loops were a short-lived patch for models that couldn't reliably keep working on long problems until they hit a defined goal Fable and GPT-5.6 (and probably Kimi K3 as well) can just do that out of the box, unassisted
Wes McKinney
@wesmckinnJul 22
I think loops are bullshit
141
60
28
1,276
157,939
Load more