Labeeb

Hi 👋, I'm Labeeb. I write backend deep-dives that help engineers build scalable systems.

[Aryan Mittal @aryan_mittal007]

Its good time that junior developers realise that yes they need to master the basics enough to know when claude is giving them 50 line solution for a 7 liner , you’re a nobody without knowledge

[Ben Dicken @BenjDicken]

Database performance and human efficiency are limited by the same thing: the universal scalability law.

In a perfect world, throwing more cores at Postgres would scale performance linearly. Two problems: contention and coherency.

Contention = some resources are shared across processes. Shared buffers. WAL. Synchronization and locking are required here, diminishing returns with scale.

Coherency = data in shared resources must be consistent! CPU cache flushing. IPC. Disk sync. With enough going on, this reverses the effectiveness of increasing parallel work.

This is one of N reasons why sharding a database becomes a necessary scaling axis with increasing QPS / data size.

It's the same reason why companies reverse efficiency when they get too big! A small, talented, focused team can outperform a much larger bloated one.

[Shaan Puri @ShaanVP]

Daily writing is the new weightlifting.

Back in the day, 90%+ of the population worked manual labor on farms. We didn’t have “gyms” because everyone was breaking their back working all day.

Once we got machines, we needed a new way to keep our body in shape - lifting heavy objects (3 sets of 10), voluntarily, at the gym.

The same thing is happening for writing. Everyone is outsourcing their thinking to AI.

The brain is a muscle, and it will atrophy if you don’t use it.

[Puneet Patwari @system_monarch]

A database is halfway through saving your data and the power dies. By all logic, that half-written record should be corrupt garbage. Somehow it isn't.

So before touching the real data, the database first scribbles "about to do X" into a plain append-only log. After a crash, it replays that log and finishes or undoes what was mid-flight.

This is called a write-ahead log. Write down the intent before the act, so a crash is always recoverable.

The trick to surviving a crash is deciding what you'll do before you do it.

[Greg Powell @ThoughtfulTechy]

The idea is simple.

Any specific tool, platform, or skill you learn in tech has a shelf life.

What felt cutting edge two years ago might be irrelevant now.

The real advantage isn't your current knowledge.

It's your learning speed.

For you to be successful longterm your focus shouldn't be on clinging to what you already know.

Your differentiator is being able to dive into something unfamiliar and get up to speed fast.

Treat learning itself as your main skill because everything else is temporary.

♻️ Share this with your network

[Roy Carrilho @RuiCarrilho5]

ended up finding this book on the matter. it's apparently a more academic, practical version of DDIA.

and it's free!

[Gunnar Morling 🌍 @gunnarmorling]

Call me old-school, but I don't feel working on multiple things at once to make up for agent thinking time is the way to go. Unless it's trivial work, the context switching tax is brutal, even just for two sessions in parallel.

[Ben Cera @Bencera]

Conferences are a waste of time.

Chit-chat won't get you product-market fit.

Just keep grinding in silence.

[Maximilian @maxedapps]

This is so, so bad ...

[grubles @notgrubles]

Anthropic is the new late 90s evil Microsoft.

[Ritesh Mishra @riteshmis]

Slippery slope, remember when Microsoft thought they owned the browser

[Lee B. Cyrano @leebriskcyrano]

guy with no moat:

[Gergely Orosz @GergelyOrosz]

It’s a bit rich for Anthropic’s CEO to badmouth open models, after Anthropic silently nerfed Claude in April (and later admitted in a postmortem)

Open models cannot and be “nerfed” in a way that Anthropic has already done, and can do in the future.

They also threaten Anthropic

[Hassan @nutlope]

This model is insane at design.

I asked GLM 5.2 (left) and Opus 4.8 (right) to build me a landing page and you can't even tell the difference.

GLM cost $0.06 while opus cost $0.49. More than 6x cheaper while being faster + more token efficient.

Another win for open source AI.

[Yuchen Jin @Yuchenj_UW]

DeepSeek is the GOAT. 🐳

They just published DSpark, a new speculative decoding method that boosts throughput by 51% to 400%.

They also open-sourced DeepSpec, the training framework behind it.

This is the real open AI.

[sphinx @protosphinx]

China won in manufacturing in part because the US made building things increasingly difficult through creeping regulation.

They’re repeating the same mistake in AI.

When China eventually gets its chips right, we’ll look back at this era as the moment the winds changed.

[Matthew Berman @MatthewBerman]

> mythos is so good at cyber it can't be released

also

> mythos can't detect 20k fraudulent chinese accounts attacking it

[François Chollet @fchollet]

Agentic coding forces you to design clean interfaces and document them well. An agent cannot read the implicit mental model shared by your engineering team, it can only read your API contracts and docstrings.

[Ronit Pereira @CAronitpereira]

“At the end of the day, what have I got? A successful Singapore. What have I given up? My life.”

- Lee Kuan Yew

[Gary Marcus @GaryMarcus]

COMPANY BUILT ON IP THEFT RAGES AGAINST THEFT OF ITS OWN IP