Fan of distributed systems. Previously HotDoc, Airtasker, AWS. Green energy afficiando. Opinions are my own.

Sydney, New South Wales
Joined May 2009
Our profession needs to reinvent our ways of working. Most of software engineering craft - sprint planning, stand ups, backlogs, rfcs, code reviews - was made for a time where the cost of rework was huge, and high risk. We always needed to slow down to speed up. We simply don't live in that era anymore.
1
8
692
Doug Rathbone πŸ¦˜πŸ‡¦πŸ‡ΊπŸ‘¨β€πŸ’» retweeted
oh my god
29
104
22
1,907
76,480
Doug Rathbone πŸ¦˜πŸ‡¦πŸ‡ΊπŸ‘¨β€πŸ’» retweeted
Ghostty 1.4 is now out! Full Kitty graphics with animation (the first terminal besides Kitty to do it), scrollback compression for much lower memory usage, macOS 27 support, a new OpenGL renderer on Linux, ghostty +ssh, big performance gains, and hundreds more changes. The highlights: - Scrollback compression: scrollback compresses ~90+% on average, meaning your terminals use that much less live memory - Huge memory savings across the board: no single large thing, just lots of good improvements lowers memory usage everywhere, netting anywhere from 50 to 90% less memory usage depending on workload - Kitty graphics is now 100% complete, including animation and relative placements, after a full conformance pass against the spec and Kitty itself - The Kitty clipboard protocol, with an image preview when a program asks to read your clipboard - Split drag and drop on Linux, including between windows - ghostty +ssh installs terminfo on remote hosts automatically, so things just work over SSH - Lots of terminal emulation accuracy fixes This release is 7 months of work from 172 contributors over 2,889 commits. It also fixes 10 security advisories, so please update. Full release notes: ghostty.org/docs/install/rel…
110
284
59
4,068
128,524
Doug Rathbone πŸ¦˜πŸ‡¦πŸ‡ΊπŸ‘¨β€πŸ’» retweeted
A reason I’ve cycled engineers through support for 20 years. Some argue and hate it (they don’t last long) but you’ve simply got to keep the people who can improve the products as close to the people who use the products as possible.
This so much. When the ability to ship increases by 10x, so so so so important to listen to your customers, elect the highest-value things, then build them with care (and speed). Building the wrong thing at 10x speed is of little value.
9
5
1
20
855
Doug Rathbone πŸ¦˜πŸ‡¦πŸ‡ΊπŸ‘¨β€πŸ’» retweeted
I get asked often β€œwhy not copilot?” A new answer: are you sure they’re even going to support this thing in a year? They’re literally trying to get rid the people working on it.
Here is what Microsoft's leaked buyout chart actually says, in detail. The chart splits Microsoft into three groups for the Voluntary Exit Offer. Fully participating, meaning the whole org is in: AI Experiences, Copilot and Agents, CoreAI, the CTO's office, Windows and Devices, Xbox, Finance, HR, and Strategy and Transformation. Partially participating, where eligibility varies by team: Cloud and AI, legal affairs, and Microsoft AI, the division behind consumer Copilot. Not participating at all: Marketing, the commercial business, Discovery and Quantum, and Security. Two footnotes matter. Employees of limited integration companies are not eligible. And inside the partial orgs, team leaders decide who gets the offer. The offer itself covers employees at level 70 and below in Canada, India, the UK and the US. What stands out is the split. The teams building Microsoft's AI products are fully on the list, while Security, Quantum and the commercial business are not on it at all. That list is the real story. It shows where Microsoft can afford to lose people.
4
1
11
678
Doug Rathbone πŸ¦˜πŸ‡¦πŸ‡ΊπŸ‘¨β€πŸ’» retweeted
It looks like Amazon AGI division is being shut down and their foundational models Nova are being sunset. Along the same lines they are shutting down Q IDE (autocomplete and chat in IDE), also abandoning efforts to train code-generation models. futurism.com/artificial-inte…
85
312
85
2,639
383,670
Doug Rathbone πŸ¦˜πŸ‡¦πŸ‡ΊπŸ‘¨β€πŸ’» retweeted
This morning we are releasing the largest proof in the history of mathematics: A formalization of the Classification of Finite Simple Groups (CFSG)! Over 43 million lines of Lean, 715 papers and books, and 100 substantive fixes to the literature, this is truly an enormous theorem. Well done team!
The largest math proof in human history is now computer-verified. Today, Axiom announces a complete Lean auto-formalization of the Classification of Finite Simple Groups, the "periodic table of symmetry." About 15,000 pages, built by over 100 mathematicians across generations.
53
235
33
1,625
77,541
Doug Rathbone πŸ¦˜πŸ‡¦πŸ‡ΊπŸ‘¨β€πŸ’» retweeted
Okay this is f*cking wild. Ex-Google Android engineer: "Your old phone is the best agent hardware you own. Grok Bot sees the screen, taps, types and runs your apps while it charges" This is a phone from my drawer running overnight Grok Bot β†’ Screen reader β†’ Tapper β†’ Typer β†’ Checker β†’ Done The screen reader turns screenshots into buttons and fields The tapper and typer work it like a person, just faster Watch second 11, the checker catches a wrong date before it gets sent 42 tasks done while it charged from 19% to 100% Everyone gives Grok Bot a chat window, almost nobody gives it a whole phone Worth more than most $1,000 automation courses Bookmark and watch it today Send this to the friend with an old phone in a drawer Then read how to build your first Grok Bot desk in the article below ↓
80
167
11
2,011
160,563
The Docusign leadership team needs to record a MasterClass series on β€œAdvocating for headcount growth”.
BREAKING: DocuSign says it may lay off up to 162,000 employees.
Readers added context they thought people might want to know
As of Docusign's most recent 10-K filing, Docusign had 7,044 employees as of 1/31/2026, making it impossible to lay off 162,000 employees. sec.gov/Archives/edgar…
1
135
It was about time that AI got more management theatre. Bring on the Grokbot business reviews, OKRs, and quarterly Bot calibration meetings.
I also made a Command Center Dashboard for my grok bot now My bots are so busy that I started losing an overview of what was done This helps
47
This. We reduced ops load overall by 80% at @hidovetail while 3x’ing velocity. If there was ever a time to throw out the rule book on tech debt, ops risk, and many other parts of the Eng profession, it’s now.
Ship more and have fewer incidents. Why not have both?
1
6
333
β€œMake no mistakes”
Fun fact: `--dangerously-skip-permissions` doesn't actually approve EVERYTHINGβ€”we still block some obvious things like `rm -rf ~`. But today's release will let you disable this if you really want, by setting a new env var `CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS=1`
81
Doug Rathbone πŸ¦˜πŸ‡¦πŸ‡ΊπŸ‘¨β€πŸ’» retweeted
Announcing 3 Critical vulns for hundreds of millions of IoT, Firewalls, OT, and Military devices including: Palo Alto Networks, Comcast, D-Link, HP, Canon, Honeywell, GE, Oracle, Siemens, Rockwell, Digi, Lockheed Martin, Boeing... The vulns are with GoAhead Web Server which is a backend library used by these devices. There's a patch available from GoAhead but that doesn't mean there's a patch available for the devices yet -- that's up to the manufacturer. - TLS traffic over DHE ciphers have a 1-byte secret, "04", meaning anyone can decrypt connections to these devices and steal login credentials and session tokens if the client negotiated with DHE ciphers like the latest TLS_DHE_RSA_WITH_CHACHA20_POLY1305_SHA256, which is extremely likely. - LibreSSL mTLS has been completely bypassed, meaning anyone can connect over mTLS. If you have one of these devices listening on the internet and are expecting mTLS to protect it from attackers, take it offline or patch asap. - OpenSSL Certificate Revocation Lists (CRLs) are not enforced, meaning if a user device is stolen or compromised, and you revoke that mTLS cert, nothing actually happens. There are a lot of other attacks this can be used for. These are critical because these types of devices run our world. Water systems, electricity, military, firewalls, critical infrastructure. foxio.io/blog/choose-your-ba…
150
524
416
3,308
917,162
People will blame Elon, billionaires, AI, the recession, covid and every other trope… without acknowledging that remote work simply isn’t for everyone, and many who remote work are simply unproductive.
SKYDANCE'S FIRST BIG MEMO: TAKE THE BUYOUT, OR COME IN 5 DAYS A WEEK Similar to Microsoft earlier today, except this one comes with a Return to Office mandate (other silent layoff that doesn't always show up in headlines). From co-CEOs David Ellison and Ynon Kreiz: "we recognize that not everyone will see their future here" Those who qualify get "personalized details and terms" and "a window to decide whether to accept this one-time offer." The target: more than $6B in annual cost savings. As @Amanda_Goodall notes, huge layoffs are expected to be part of it.
1
263
The door-desk strategy is looking pretty good in hindsight. Those private offices had to come at the cost of something.
Microsoft is offering 39 weeks of salary, the 2 next stock vests and 6 months of cobra to up to 60.000 employees (~27% of their total workforce) in a effort to reduce expenses after AI productivity gains.
118
The striped pyjamas and sheepskin bedroom loafers make this for me... 🧐
Anthropic 2025 revenue: $4.6 billion 2025 net loss: -$42 billion IPO valuation: $2 trillion
1
185
Dead internet theory seems more real by the day. Thanks Karan-bot.
3
321
At least folks you never want to hire tend to self identify well. Imagine being so pessimistic about the innovation in your field, that it turns you into a rage filled stalker…
glad i blocked this toxic guy a while ago when he was being condescending to someone else. also meta has its own scm and my commits aren’t all public 🀑
110
Doug Rathbone πŸ¦˜πŸ‡¦πŸ‡ΊπŸ‘¨β€πŸ’» retweeted
2 hours ago, Anthropic filed for an IPO. We went through the S-1 and broke it down in 6 charts:
56
299
70
2,584
919,072
oh dear.
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected. Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions. Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent. But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models? We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem. Read the full report: cohere.com/blog/concerning-b…
1
223