Martin Kleppmann
banner
martin.kleppmann.com
Martin Kleppmann
@martin.kleppmann.com
Associate Professor at @cst.cam.ac.uk, researching decentralised systems and security protocols. Advisor to the Bluesky team. Wrote “Designing Data-Intensive Applications” (O’Reilly). he/him
Reposted by Martin Kleppmann
Hello #Local-First enthusiasts, We have great news to share with you! @fosdem.org'26 wil be hosting a full devroom dedicated entirely to local first software. We have gathered the best lineup ever, all major open source projects will be presenting! Come and join us, it is free! openlocalfirst.org
Open Local First
For the first time this year, the FOSDEM conference will host a devroom for Local First, CRDTs and sync protocols and sync engines. FOSDEM is the biggest Free and Open Source conference in Europe. Thi...
openlocalfirst.org
December 24, 2025 at 1:43 PM
What if using American cloud services becomes untenable for Europe due to deteriorating geopolitics? It's hypothetical for now, but no longer unthinkable. By @quentinsf.com: statusq.org/archives/202...

Local-first software to the rescue?
Living without America | Status-Q
Quentin Stafford-Fraser's blog
statusq.org
January 8, 2026 at 9:45 AM
Reposted by Martin Kleppmann
New on my blog: "Why Study CS? Thoughts on LLM-assisted software engineering" kmicinski.com/claude-code-...
Why Study CS? Thoughts on LLM-assisted software engineering
Dear students of Computer Science,
kmicinski.com
January 7, 2026 at 3:15 AM
Reposted by Martin Kleppmann
Sorry, but triumphant claims about autonomous vehicle safety are wildly exaggerated.

It's an open question whether today’s self-driving cars are any safer than those driven by humans.

And if reducing crashes is the goal, that isn’t even the right question.

My deep dive, in Bloomberg 🧵
We Still Don’t Know if Robotaxis Are Safer Than Human Drivers
And even if self-driving technology proves to be less dangerous, there are many better ways to improve traffic safety and prevent fatal crashes.
www.bloomberg.com
January 6, 2026 at 2:54 PM
The rise and fall of Stack Overflow. Graph shows number of questions posted per month.

Discussion: news.ycombinator.com/item?id=4648...
January 6, 2026 at 9:48 PM
Reposted by Martin Kleppmann
New blog post! A close look at Tahoe menu icons https://tonsky.me/blog/tahoe-icons/
It’s hard to justify Tahoe icons
Looking at the first principles of icon design—and how Apple failed to apply all of them in macOS Tahoe
tonsky.me
January 5, 2026 at 11:10 AM
The O'Reilly Learning version of Designing Data-Intensive Applications, 2nd edition, will have some multiple-choice quiz questions at the end of each chapter and at the end of the book, for anyone who wants to check their understanding (they're not included in the print edition).
January 1, 2026 at 3:29 PM
New policy: declare that the new year starts at 9pm so that we can end the party early and all go to bed at a sensible time. Happy new year everybody and good night!
December 31, 2025 at 9:17 PM
Great talk on AI-human collaboration on text documents, in particular on how to enable an LLM to make precise, fine-grained edits to the text youtu.be/kQSfXSliBzU?...
Lee Byron (OpenAI) - Teaching Models to Collaborate
YouTube video by Local-First Conf
youtu.be
December 31, 2025 at 4:39 PM
Reposted by Martin Kleppmann
I recently wrote some thoughts on wrapping Rust libraries for Wasm plus a couple patterns that I use to make wasm-bindgen feel a bit more manageable. No silver bullets here but maybe some sensible defaults (fight me in the comments 😛) notes.brooklynzelenka.com/Blog/Notes-o...
December 30, 2025 at 9:58 PM
Reposted by Martin Kleppmann
Systems don't pause between retries

If action a was applied at step i, then replaying a at any future step j is a no-op

Idempotence is a property of traces
December 30, 2025 at 5:00 PM
Reposted by Martin Kleppmann
Here's a tragedy of DOGE that most people will never know about, even though it has big consequences: no one is left to coordinate the transition to memory safe systems code. 🧵
"Memory Safety for Skeptics," published in the ACM Queue.

Arguing for why memory safety is worth pursuing, even amid competing priorities and limited budgets, and with or without Rust.
Memory Safety for Skeptics - ACM Queue
queue.acm.org
December 30, 2025 at 4:58 AM
Reposted by Martin Kleppmann
This has everything: malleabile encryption, complex packets, memory safety, unsafe terminal UI, wontfix.

Incredible.

Plaintext exfiltration if you just click a link.

I can’t be convinced gpg can be used safely.
December 27, 2025 at 4:53 PM
Reposted by Martin Kleppmann
🧵 Could Bitchat have 5x the range and 100x the throughput for the same power expenditure? I explored how Wi-Fi Aware could improve the reliability and throughput of Bitchat and mobile ad-hoc networks in the absence of internet connectivity. #bitchat

👇
December 23, 2025 at 4:09 PM
We got a white Christmas (with a light dusting of snow) in southern Germany! Merry Christmas to all who celebrate it! 🎄❄️
December 24, 2025 at 10:05 PM
Reposted by Martin Kleppmann
Oh, that's a useful feature. You can enforce causal consistency in general (which includes read-your-writes and other guarantees) with that mechanism. We've had that in MongoDB for some time, though, and customers have found it hard to use: emptysqua.re/blog/how-to-...
How To Use MongoDB Causal Consistency
The missing manual for a useful MongoDB feature.
emptysqua.re
December 22, 2025 at 11:46 PM
Reposted by Martin Kleppmann
🐘 Oh, that's nice: Postgres 19 is going to ship read-your-writes for standbys when using asynchronous replication, via a new command WAIT FOR LSN (similar to WAIT_FOR_EXECUTED_GTID_SET in MySQL). Very cool!

www.postgresql.org/docs/devel/w...
December 19, 2025 at 9:34 PM
Interesting to see what research articles cite our paper on local-first software. Today: automatically monitoring sea turtles 🐢 www.sciencedirect.com/science/arti...
Toward decentralized ecological monitoring: A local-first web-based AI platform for sea turtle detection
Accurately monitoring sea turtle populations is crucial for informing effective conservation strategies; however, traditional methods for assessing sp…
www.sciencedirect.com
December 22, 2025 at 9:37 AM
We stayed at an Airbnb that had some issues but was overall pretty good. Left a review that was 4 stars in two categories, and 5 stars in all other categories. Inexplicably Airbnb turned that into an overall 2-star review. The host is understandably upset. Anyone know wtf is going on here?
December 21, 2025 at 10:22 AM
Reposted by Martin Kleppmann
It's seriously amazing how fast the AT Protocol developer experience is, like from idea to early results is wowzers.

Like, I've spent all of maybe two days working on @questionable.fyi, and I've already account profiles working with arbitrary data on the network.

My next thing to do is OAuth... 😅
December 21, 2025 at 2:58 AM
Reposted by Martin Kleppmann
Using an age keyserver as a demo, this article demonstrates how to add a transparency log to a centralized service step-by-step.

We use Tessera for the tlog, VRFs for privacy, and the Witness Network. It all takes just 500 lines to integrate!

The result of years of work making tlogs accessible.
Building a Transparent Keyserver
We apply a transparency log to a centralized keyserver step-by-step, in less than 500 lines, with privacy protections, anti-poisoning, and witness cosigning.
words.filippo.io
December 19, 2025 at 3:32 PM
Reposted by Martin Kleppmann
Raft has a real problem guaranteeing Read Your Writes, because there might be two leaders serving reads at once. In theory, you could use a timed lease to ensure there's only one leader.
LeaseGuard: Raft Leases Done Right!
A new Raft enhancement for fast, consistent reads.
emptysqua.re
December 19, 2025 at 1:49 PM
Reposted by Martin Kleppmann
DRAM prices mooning will hurt everyone and might cause a consumer tech recession on its own in the next 12 months even without the AI bubble popping.

Hyperscalers are going to jack up prices for cloud resources, electronics manufacturers (Apple, Samsung, etc.) will raise prices, this is gonna hurt.
December 15, 2025 at 5:41 AM
Reposted by Martin Kleppmann
This is giving me life. I’ve always appreciated Hiromi’s completely ecstatic way of performing and this is no exception:

youtu.be/XGjy73DPScc?...
Blackbird & Jesu, Joy of Man's Desiring - Hiromi Uehara (Complete Transcription)
YouTube video by Cornelius Tsen
youtu.be
December 14, 2025 at 4:35 AM
Reposted by Martin Kleppmann
Our focus is specifically on getting *CS* people into formal verification, hence the name, "Logic for Systems" (as opposed to "Logic for philosophical syllogisms", say). There's a book accompanied by a verification tool specifically designed for education (not a lot are!).
forge-fm.github.io/book/
How to Read this book - Logic for Systems: Lightweight Formal Methods for Everybody
forge-fm.github.io
December 13, 2025 at 12:25 PM