My first product (Testimonial Creator) is live on @ProductHunt 🚀
Would like to see some banger testimonials by you guys 🔥
Check it out
Create testimonial cards in seconds, ready to share anywhere
https://t.co/xdi42hoqKW
Everyone building modern systems carries responsibility for the legal and ethical weight of their decisions.
GDPR’s “right to be forgotten” sounds simple. A user wants their data deleted. You delete it. Done.
Except it’s not.
Most data architectures are built on immutable constructs. Append-only logs. Event streams. Kafka topics. These aren’t bugs, they’re features. Immutability is what makes distributed systems reliable. But immutability and “the right to be forgotten” are fundamentally at odds.
How do you delete something from the middle of a file designed never to change?
And that’s before you ask: where else did that data go?
Analytics tables. Materialized views. ML training datasets. Snapshots from six months ago sitting in cold storage. The original record is just the root.
By the time a deletion request comes in, that data has already branched into a dozen downstream systems — each with its own retention logic, or none at all.
GDPR doesn’t give you a technical blueprint. It gives you a legal obligation and leaves the architecture to you. Intentionally, because tech moves faster than legislation can keep up with.
Compliance isn’t a checkbox. It’s a question you have to answer at design time, not after the fact.
Deleting the source record is the easy part. Proving it’s actually gone everywhere it propagated, that’s the real engineering problem nobody talks about enough.
Got myself a little something 👀
Been searching for this book a while now and I can’t wait to see what @martinkl added to this edition
Has anyone read the 2nd edition, and if so, how do you like it?
Why are people so gullible.
Lots of profiles (probably engagement farmers) posting about Anthropic removing Claude Code from the 20$ Pro plan just to set you up and go to a phishing website made in Webflow so that they can steal your account and data.
To anyone who fell for it, my condolences, and hope you learn how to check and test links and not be so gullible.
Here is the REAL Claude pricing page:
https://t.co/veB8OagkBH
Here is the FAKE pricing page: https://t.co/OrbXxTybEI
Job hunting humbled me HARD. Thought I was hot shit with my experience, until the ghosting, the 5-round interviews only to get told that you were a bad fit at the 2nd round, and the “quick” take-home project that ate my entire weekend (for free).
I applied to 80+ roles. Half were fake postings, jobs that were already filled or never real to begin with. Got excited for “final round” only to get hit with round 4, 5, and 6. Spent 12+ hours on unpaid projects that probably went straight into someone’s portfolio. Got the feedback that was purely some word slop or even confusing, since some of the roles that I apply to are entry roles that require some amount of knowledge, but also not senior/mid level knowledge. Zero transparency.
There genuinely needs to be new workers rights laws:
1) no fake job postings
2) max 3 interview steps
3) no free labor (“take-home projects”)
4) new hiring transparency laws
@J_D_Herz@SandraW5386@beyoncegarden Continue in the fucking DM's or just shut up. Stop arguing in the comment section with these long meaningless comments.
Belief is not something that can be explained as a hypothetical through questions and examples.
The belief is something that you know that is to be true, whether it's something as simple as a pen or something as big as God, that you will be willing to put your life at risk because you BELIEVE it, and not know it.
The kid only does hypothetical stuff and many of the debates he is in is mostly political. Can't argue with someone that is not in the field of study as you are in.
Not saying that JP is someone who needs to be praised by any means, but someone that doesn't grasp the concept of the things he is saying (not to agree, but to understand) is someone that is not competent to saying anything nor debate.
@DickusMaxximun@beyoncegarden Saying "The end" doesn't do anything if there is no explanation to How did he not answer the question. Explain it how he didn't answer it and then we will see if it's The end or no.
If you are a guy who thinks like this, I can't imagine you reading any or even one philosophy book.
Having a opinion like this gets you nowhere if you don't have anything to back up the "discourse" part. I even said that there is no discourse in the video, just a deeper understanding of the meaning behind the question.
Well that’s also a big problem, because of the amount of things people say who are actually inteligent and well spoken can be BELIEVED to have been generated by a LLM.
There is no point in arguing about this since I know that my words are mine and not LLMs, but you believe what you want.
Good question, and here's a "short form" answer.
This attack relied on a sneaky postinstall script inside the fake plain-crypto-js package. When you ran npm install, it automatically executed and dropped the RAT, that’s how it got in.
pnpm (v10+) and Bun are way stricter by default:
- They disable preinstall/postinstall scripts from dependencies (you have to explicitly allow them).
- pnpm also uses a content-addressable store that verifies every package with hashes, so tampering is much harder to pull off.
In this exact case, the malicious code probably wouldn’t have run at all on pnpm or Bun. That’s why I’m pushing the switch.
npm’s defaults just give too much trust these days.
This axios supply chain attack on npm is pure madness.
Maintainer creds hijacked, malicious versions 1.14.1 & 0.30.4 sneak in a fake “plain-crypto-js” dropping cross-platform remote access trojans. Over 100 million weekly downloads compromised, bypassed CI, no tags, self-destructing payload. Classic.
NPM dependency hell isn’t a bug anymore, it’s the entire business model at this point.
I’m convinced the only real fix is ditching npm entirely and switching to pnpm or Bun. Cleaner resolution, less bloat, and maybe we finally stop getting owned every other week.
Exactly. When a bug appears in production, you ask the AI to fix it… and it often “fixes” the symptom while introducing 2-3 new ones elsewhere because it doesn’t actually understand the system architecture, trade-offs, or why certain decisions were made.
Then you spend hours debugging the AI’s fix, only to realize you still need deep knowledge of your own codebase to verify it’s correct and safe.
AI agents don’t truly understand the codebase they generate. They pattern-match and predict tokens.
Relying on “just ask the AI” for deep ownership is the exact vibe-coding trap: you end up with code that “works” today but becomes a black box tomorrow when requirements change, bugs appear in production, or you need to scale/refactor.
Real product builders don’t outsource understanding to probabilistic models. They maintain mental models of their own system. Otherwise you’re not building, you’re babysitting an unpredictable co-pilot that never actually “quits” because it was never responsible in the first place.