Dirk Schumacher 🚲

We are introducing EU Inc. To make building and growing a business across the EU faster, simpler, and smarter. 🔸 Start a company in less than 48 hours 🔸 No minimum capital requirement 🔸 Fully online and borderless

Oh wow, a popular GitHub Action (tj-actions/changed-files) was fully compromised. Someone committed a base64-encoded payload that runs a script that in turn prints out encoded secrets… Stay safe out there!

@dvassallo Diwurstify

Diversify

@jdatap Nice, Is it public somewhere? Side quest: translate R into Go’s AST and generate go code 😅

After god knows how many months I finally have a lexer and parser that fully handles #rstats. It uses no external library, no treesitter, nothing, just standard library Go. Programming is just so fucking cool.

Never once in the quarter of a century I've been building for the web has picking MySQL come up as a regret. Unbelievably solid, no-nonsense tech.

@ninja_maths @justinskycak @atharv_ojha_ Cryptography would also be cool

@justinskycak @atharv_ojha_ Early 2025. If you're interested in Discrete Math, please complete "Methods of Proof" first. x.com/ninja_maths/st…

Great news. @_MathAcademy_'s Probability and Statistics course will be launched in time for Christmas. The final list of topics is attached. Just call me Santa Claus. 🎅

The world will hate us (🇩🇪) again as soon as every support team of every SaaS company everywhere is hit by their German customers with "PLEASE PROVIDE AN E-RECHNUNG".

@elonmusk F

My hardcore Diablo Druid died today 😢 Uploading his last pit clear in memoriam. RIP lil buddy 🪦

@ninja_maths What's your recommendation to teach from grade 1 to 3 (as a parent)?

I just returned from my six-year-old son's "parent workshop" for maths. This workshop aimed to educate parents on how math is taught at the school and to encourage them to ask questions. I gotta say, I'm feeling a little depressed. The curriculum they've opted for feels bloated and overly complicated, with an overreliance on manipulatives and too many different problem-solving strategies. I cannot see how the majority of kids won't struggle. The word "repetition" was not used once in the entire presentation, which shocked me. Kids need reps, end of story! The curriculum uses the so-called CPA approach (Concrete-Pictorial-Abstract). While I think this is generally a good approach, there's a danger that students will get stuck on the C-P part and find it difficult to move to the A-part (the ultimate goal). The issue with overreliance on manipulatives and inefficient strategies is that kids get comfortable with them. Once they get comfortable, moving them onto a more efficient strategy is hard. I now know this from experience. My son (capable of some abstract manipulations already) would much prefer to count beads if given the choice. Concrete models for solving problems are often fine to start with. But sometimes they're just a bad idea or should be postponed until later. For example, the teacher demonstrated calculating 73 - 25 using tens and ones blocks. The process took about 4-5 steps. I would be very surprised if many six-year-olds could internalize this method and use it to solve problems independently. Even those who can will likely take a long time to do so. Don't get me wrong. The teachers are passionate and well-intentioned. The problem is that I don't think it's the school - it's primary math education in general. Alarm bells started ringing when my son complained that "maths at school is hard" and that "maths with dad is easy," even though he had a clear passion for maths only six months ago. When I designed our 4th—and 5th-grade curriculums, I made sure that we didn't overrely on concrete models. Sure, there's a place for them, and they're surely needed. But once kids have grasped a concept using a concrete model, it's important to move them onto how things are done in practice (e.g., standard algorithms) as fast as possible! I sent a (hopefully) politely worded email to (a) stress the importance of repetitions, especially for important skills like using inequality symbols and certain types of word problems, and (b) stress that my son is unlikely to move onto the "Abstract" part by himself and needs to be challenged.

Our new textbook Theoretical Foundations of Conformal Prediction is out! Conformal prediction is a a statistical technique that augments ML systems with uncertainty information for safe deployment. This book lays out the core theory. arxiv.org/abs/2411.11824

If I count all forks of LLR (github.com/dirkschumacher…) as active users, then I can happily report that this language is used by at least 14 people🤣

Rails 8.0: #NOBUILD, #NOPAAS, all-in on SQLite as a production database option with jobs, cache, and cable, new authentication generator, and so much more! Final release is out 🎉 rubyonrails.org/2024/11/7/rail…

MA is hiring!

Ich denke da auch immer wieder drüber nach. Vielleicht gründe ich auch irgendwann eine kleine Agentur, um der öffentlichen Hand zu helfen - falls das Problem die Supplier sind. Ich kann mir vorstellen, dass die folgende Strategie funktionieren könnte: * Dienstleistungen finden, die online durchgeführt werden können (z.B. Melderegisterauskunft oder Urkunden vom Standesamt) * Kommunen finden, die das ganze nur per Email abbilden ohne Backend. * Mit denen reden, ob die an einer GDPR et al. compliant Lösungen interessiert werden * V0 könnte eine client side app sein, die alle Nutzereingaben mit einem Public Key der Kommune verschlüsselt und dann wie gehabt per Mail verschickt. Authentifizierung mittels nPA. Der Bürger bekommt eine nette UI, die Kommune kann einen einzelnen Prozess low risk digitalisieren ohne viel durcheinander zu bringen intern. * Wenn das klappt, kann das natürlich ausgebaut werden, mehr Prozesse oder ein custom backend etc. etc. Ein paar schnelle Gedanken.

Random Gedanken zu Adler OS 🇩🇪: Es gibt dutzende Bestrebungen die deutsche Verwaltung zu digitalisieren, imho mit fehlerhaften Ansätzen. Da rein zu kommen wird sehr, sehr schwer und hätte das Risiko, dass Adler OS der Versuch n+1 wird. Was überall fehlt, sind Standards an die sich alle Player halten. - Datenformate - Authentifizierung (Bund ID anyone?) - UI Best Practices - Natürlich Zentrale Register ... ich überlege, was das Beste wäre was Adler OS open-source tun könnte was allen ein wenig hilft. Als ersten Step mit möglichst wenig bürokratischen Verfahren. Beispiel: Ready-to-use UI's für die gängigsten Anträge der Bürger Oder vielleicht erstmal die Dokumentation der Datenformate als ein Standard auf den man sich bei der Entwicklung stützen kann? Oder ggf. Dokumentation und Anleitung wie Bürger überhaupt die (sorry, grässliche) AusweisApp2 einrichten und nutzen können. Was ist der minimalste Beitrag der helfen würde den man leisten kann, ohne das man die Erlaubnis von 300 Gremien braucht? Mein Ansatz bei sowas ist eher, Leute zu begeistern die schnell Dinge ausprobieren können um dann mit Hebel die schweren Brocken zu überzeugen. Beispiel: Wenn Adler OS zehntausende Bürger als Nutzer hätte, zB für die AusweisApp, ohne als Frontend das hintenraus den Prozess semi-automatisiert, oder ähnliches, sind Kommunen natürlich sehr viel hellhöriger als heute. Bei Parqet habe ich das übrigens genau so gemacht. Um Erlaubnis fragen wird uns nicht weiter bringen. Die Situation ist ein absolute Chaos mit Playern die den Fortschritt verhindern möchten dank derer Monopole auf den Rechnern der Verwaltung. Ich bin im Team "einfach machen" aber ich schwanke noch, was die beste Aufwand/Impact Ratio hat.

@TweetsOfSumit wat

So apparently, the great Vaterland now taxes you if you have €500k in an ETF and leave 🇩🇪. I (was) generally not against high taxes but it surely feels like GER is just adding more & more tax burdens to companies and individuals, always reducing the odds of building wealth and get financially independent to NOT rely on welfare

Sign for Europe: eu-inc.org

In praise of mystery ✨ After a successful launch aboard a @SpaceX Falcon Heavy rocket from @NASAKennedy, @EuropaClipper is on its way from our ocean world to another to see if Jupiter's moon Europa has conditions suitable for life: go.nasa.gov/3Ua0OnV