French financial newspaper Les Echos clearly shows their opinion (and possibly fears) of their owner LVMH this week: 4 stories against taxing the richest and for more austerity. Usually their editorial articles are more nuanced and balanced (even though it’s a liberal newspaper)

Just published: Outbox Pattern Survival Guide (3 min read) medium.com/@tpierrain/o... #architecture #outboxPattern #reliability #scalability

Just visited (at last) the Soulages Museum in Rodez! Great to witness the evolution of the artist, in a beautiful building. Extra bonus: the fantastic Café Bras on the premise 😋

most people want a quick and simple answer to why AI systems encode/exacerbate societal and historical bias/injustice and due to the reductive but common thinking of "bias in, bias out," the obvious culprit often is training data but this is not entirely true 1/

Five Quick Reads: Durable v Disposable code, Domain Models as memory, getting AI to generate its own coding rules, what AI should stand for, and what loses more time than you save with AI martinfowler.com/articles/202...

have you tried simply not doing war crimes

“The intention of modelling is to surface insights” — @ruthmalan.bsky.social

Running a random subset of tests for a small amount of time as a hook for an AI agent is super smart.

It's cool to see all the holes and imperfections being gradually addressed in the overall AI coding developer experience. This week I've been playing around with Context7 mcp server which ensures your coding assistant has upto documentation. github.com/upstash/cont...

New paper hot off the press www.nature.com/articles/s41... We analysed over 40,000 computer vision papers from CVPR (the longest standing CV conf) & associated patents tracing pathways from research to application. We found that 90% of papers & 86% of downstream patents power surveillance 1/

Hélder Câmara describes 3 types of violence: 1. Institutional aka structural injustice (as described below) 2. Revolutionary (against the oppressor) 3. Repressive (against the revolutionaries). Those in power will claim that only n°2 is violence. en.wikipedia.org/wiki/H%C3%A9...

🧵 I’ve reviewed conference talk submissions for +10 years. Here are my tips: (Side note, I've sinned against all these tips myself 🙃)

Using OpenFoodFacts to check for ultra-processed food (UPF); I’m very happy to report that Picard Surgelés, a French company manufacturing and selling frozen products, updated many products to remove most or all its additives! ❤️ Half of tested even went from NOVA 4 to 3. Now updating the DB.

From an engineering perspective, software wants to be decoupled. From a business value perspective, software wants to be deeply interconnected. That's the fundamental friction that software design, and Domain-Driven Design specifically, attempts to address.

“The way you work is as much a part of your product as the product itself” #dddeu

A somehow funny 3x3 mosaic collage of pictures of @cyetain.com Jabe Bloom on stage in multiple very dynamic body postures during his closing keynote at #dddeu

"Create contexts in which options can be discovered, evaluated and selected. These contexts themselves constrain the kinds, amounts and complexity of the things discovered." - @cyetain.com at #dddeu

Some serious Herbert Simon vibes during @cyetain.com closing keynote at #dddeu 2025!

Domain-driven Design’s magic is to convert complex problems into predictable, repeatable, orderly software - @settling-mud.bsky.social with @johncutle.fish at #dddeu

That’s how we do Domain-Driven Design for real: @ericevans0.bsky.social can’t resist but going in-depth into learning the domain of how LLMs work, down to monosemantic and polysemantic neurons! #dddeu

Ignore Hype and Backslash about AI! - @ericevans0.bsky.social on #AI at #dddeu

Feature Flag Alternatives to consider: • Better Split (of user stories) • GoF Design Patterns • Domain-Driven Design • Strangler Application Pattern • Branch By Abstraction • Circuit breaker Dorra BARTAGUIZ at #DDDEU

Thanks Tobias Goeshel for sharing your 4-month experiment of using AI agents to solve a domain problem directly! Lots of cues of what’s ahead of us, and all this thanks to last year @ericevans0.bsky.social talk! #dddeu

Now Tobias Goeshel "How to train your domain model", starting by explaining why he's a genAI sceptic. (Research shows it basically makes dumber, energy cost, hallucinations etc) but is still curious on ways it could serve us. #dddeu @dddeu.bsky.social

"Red flag. We stopped building pyramids 5000 years ago, but software architects still do" #dddeu

Gregor Hohpe starting off his keynote about platform engineering with two fundamentally different economies at #DDDEU.

More diversity is only possible on top of a common platform, with 1) on top the economy of speed and 2) at the bottom the economy of scale 🏭 #DDDEU 👫 🚘

If software engineers had named the automobile... ...it'd be called PistonCrankshaftGearWheelAssembly 😂 - Gregor Hohpe #dddeu

Platform used to be a good word to have a project funded. But not its AI! 😂 Badge engineering is not platform engineering 😂 Gregor Hohpe mixing fun and deep content in a great talk on Platform engineering #dddeu

And that’s a start of #DDDEU by @gienverschatse.com herself on the huge stage of the Antwerp Elizabeth Hall! 🎉 #ddd

this was a career highlight for me — getting to work with @cyriux.bsky.social and @malk-zameth.bsky.social and @bitboss.bsky.social and others! At times fun, serious, joyous, reflective, playful, and so *generous*

so good! fly.io/blog/youre-a...

I will be attending @ruthmalan.bsky.social’s workshop tomorrow at #dddeu, still time to join me there!

Got the cover design for the second edition of DDIA from the publisher. Now just need to finish the effin’ manuscript…

🔥 There hope! Social networks polarize us right now, but could they help us enrich the debate our democracies need? A research paper with former Taiwan Minister of Digital suggests a promising approach: Prosocial Media arxiv.org/abs/2502.10834

This looks like a good event today, free to attend. Sensible people I respect (e.g @kentbeck.com, @skamille.themanagerswrath.com, @gergely.pragmaticengineer.com) with likely sensible takes www.oreilly.com/CodingwithAI/

This was yet another chance for me to talk about the #ArchitectureAdviceProcess in a deep, meaningful and nuanced way with an architect I respect a lot, @gienverschatse.com. I learned a lot. You will hopefully too.

The DORA research found that there is NO TRADE-OFF BETWEEN SPEED & QUALITY Teams with good scores on both, spend 44% more time on new features!!! Read “Accelerate” by Nicole Forsgren et al. 4/4

You’ll notice that “test coverage” is not one of the desirable properties of tests in testdesiderata.com. Coverage by itself means nothing. Lack of coverage means something but increasing coverage doesn’t necessarily mean anything.

Super session de @cyriux.bsky.social à propos d’architecture à #mixit ce matin 👏. Merci pour la dédicace (moi aussi j’ai appris pleins de trucs cools grace à toi à l’époque 🙂). La Feuille de triche des ANTI-patterns de Stabilité que tu évoquais : medium.com/@tpierrain/f...

L'architecture est une tension entre une vision descendante "problème/solution" et une vision montante "contraintes" #MiXiT @cyriux.bsky.social @mixitconf.org

Présentation brillante et synthétique des différentes architectures par rapport aux différents niveaux de modularité ! Il faut que tout le monde voit ce talk (pour ce sujet et pour les autres) #MiXiT @cyriux.bsky.social @mixitconf.org

I've never had any luck trying to explain what it's like to be from Michigan—but here on forward I think I'll send this article and call it good. www.nytimes.com/2025/04/18/u...

'Bluesky has overtaken its flailing rival X in hosting posts related to new academic research, indicating the platform is fast becoming the go-to place for scholars to share their work.'

1. LLM-generated code tries to run code from online software packages. Which is normal but 2. The packages don’t exist. Which would normally cause an error but 3. Nefarious people have made malware under the package names that LLMs make up most often. So 4. Now the LLM code points to malware.

The American public:

Paraphrasing Gitte&Jakob: "Making autonomous teams work requires more than magic, even though we often act as if it magic would work"

I don’t remember who I heard it from - but the optional sprint length is one day. You meet in the morning, talk about what needs to be done. You work through the day together merging in changes continuously. At the end of the day you talk about how it went and then go home. Repeat