mbernstein

The argument for universities to be a place to learn to think critically and not learn specific skills is an even stronger value prop in an era where useful skills likely change rapidly.

Hiring earlier career folks isn’t about maximizing their productivity.

It’s also about an influx of new ideas, different lenses to view problems, and connections to people like them, amongst a number of other reasons.

This is a compute memory trade, not compression vs. turobquant? Lemma 1 is something like, "forward pass is deterministic because it's deterministic" which means the input tokens were always the lower bound...which isn't caching? Smells tautological. What am I missing?

They aren't - they'd have led with that. (note: I use arch and am legally obligated to tell you).

HN used to provide a really high signal to noise ratio for me, but it's degrading pretty quickly. There are new accounts below saying "hey I just learned what RTOS means, thanks!"

I reflexively reload HN many times per day, but I'm wondering if I need a walled garden with some sort of curation of individuals - which sucks - to get the signal level I want.

I've noticed a gigantic uptick in text messages and phone calls where people try to bypass the call screening. It may get to the point where I'll only want to see comms from people in an allowlist.

Stork.JS is a really well written piece of software, though.

Short answer: yes.

Longer answer: About 10 years I moved into leadership roles (VP Eng) and while I continued to write code for POCs, it hasn't been my primary role for quite some time. DDIA has been a book I pull out often when guiding leaders and members of my teams when it comes to building distributed systems. I'm writing more code these days because I can, and I still reference DDIA and have the second edition preordered.

There's almost no shot to get hand authored posts some views (I tried with one of mine recently). I felt like I submitted it and a moment later there were like 20 new very obviously AI generated posts ahead of it.

Shameless plug - I sort of eluded in this post I wrote about Dark Factories generally and about rust being better than Go for building software (not just agents) with AI - but I think something generally important is feedback loops. While not all feedback loops are created equal and some will be superior, my argument is that holistic approach of including diverse, valuable feedback loops matters more.

https://bernste.in/writings/the-unreasonable-effectiveness-o...

Do they? So far I haven’t found anything that matches battery life, build quality, or trackpad quality.

Acrolinx | Remote (US time zones) | Software Engineer (Python) (AI/LLM) | 125k-150k + benefits https://www.acrolinx.com

Every day, millions of people read content checked by Acrolinx. Our AI platform ensures that what companies publish is accurate, on-brand, and high-quality. You'll join the team building the next generation of these language-aware services. You'll work in python, integrating LLMs, optimizing performance, and delivering features to customers ranging from Fortune 500s to startups.

You should have: 3+ years of Python (HTTP APIs, relational/NoSQL, experience with a cloud platform like AWS/GCP/Azure) Strong grasp of system design, algorithms, data structures, and testing. Experience with LLMs and a desire to build products utilizing them.

Bonus (not required): Experience with any of the following - PyTorch, Hugging Face, FastAPI, Docker/K8s, async Python, Temporal/Airflow. Experience integrating LLMs into products.

We're a small, remote-first team working in cross-functional squads. If you're excited about building AI-powered products, let's talk: recruiting-mb@acrolinx.com

These are terrible examples that don't prove a single thing. Babel, Webpack, and React all used leftpad as dependencies. Blaming someone for using an Apache project is absurd.

Here's my pointless randomly made up on the spot anecdote - you're more likely to write a vulnerability in your own logging system than being impactedby using a widely adopted opensource one.

I just did a pass with some replacements with o1 and it very much still recognized it as the Einstein riddle and actually seems to have cheated a bit :)

"Revisiting assumptions

Considering "Camels" might be a mistake for "Kools," leading to confusion. This inconsistency complicates solving the puzzle, showing the need for careful brand assignment."

Tracking puzzle progress

I’m mapping out various house and nationality combinations, but the classic conclusion is the Norwegian drinks water and the Japanese owns the zebra.

Analyzing the arrangement

I’m working through the classic puzzle structure and noting variations, while consistently identifying the Norwegian drinking water and the Japanese owning the zebra as the final solution."

At least per Github, the TabbyML project is older than the TabbyAPI project.

I remember that class - it was just between recess and lunch!

Is your customer bringing you the solution to the problem or the problem and asking you to solve the problem? One is a translation activity and the other isn't.

Not outsourcing at all - you're are an engineer using the tools that make sense to solve a problem. The core issue with identifying as just a coder is that code is just one of many potential tools to solve a problem.

Seriously scaling nuclear would involve batteries. Nuclear has issues being cost effective at 80+% capacity factors.

I assume you mean that sub 80% capacity nuclear has issues being cost effective (which I agree is true).

You could pair the baseload nuclear with renewables during peak times and reduce battery dependency for scaling and maintaining higher utilization.

When talking to my parents, I hear a lot about Jane Fonda and the China Syndrome as far as the fears of nuclear power.

She's made the same baseless argument for a long time: "Nuclear power is slow, expensive — and wildly dangerous"

https://ourworldindata.org/nuclear-energy#:~:text=The%20key%....

CO2 issues aside, it's just outright safer than all forms of coal and gas and about as safe as solar and wind, all three of which are a bit safer than hydro (still very safe).

History is a great reference, but it doesn't solve our problems now. Just because hydro has prevented more CO2 until now doesn't mean that plus solar are the combination that delivers abundant, clean energy. There are power storage challenges and storage mechanisms aren't carbon neutral. Even if we assume that nuclear, wind, and solar (without storage) all have the same carbon footprint - I believe nuclear is less that solar pretty much equivalent to wind - you have to add the storage mechanisms for scenarios where there's no wind, sun, or water.

All of the above are significantly better than burning gas or coal - but nuclear is the clear winner from an CO2 and general availability perspective.

Of course not.

Nuclear power adoption is the largest force to combat climate change.

I did the same for grim life collective for my wife. 5m video for $50 from one her favorite YouTube content creating duos seemed worth it.

Out of curiosity what are you using to measure developer productivity platform or metrics wish (if beyond typical sprint metrics)?

Examples?

Most individuals aren't essential, and no one would prioritize them. However, a company is successful due to the individuals that work within. When 700 out of 770 employees in quite frankly the hottest startup in the world band together and threaten to leave (and join Microsoft) if they aren't given an appropriate explanation, it doesn't matter what anyone thinks an employer owes an employee. Implying otherwise is absurd.

If ~91% of the employees leave OpenAI, they will not be fine. That is delusional.

No?

Energy stations (gas) pay pretty well. Right now Tesla owns most of the electric world and even if their cars fail could become a dominant energy provider.

Depends on a number of factors. I think GitHub copilot is reducing that trade off a bit more. If the average dev stays somewhere for 2.5 years (our average is quite a bit higher because I tend to think we’re good place to work) - how long are you willing to invest in them learning the new tech (and on top of the, business logic/etc)? The lower the investment number the less you’re willing to dive into rust sans specific technical reasons (which go and elixir would easily meet).

Separately, I don’t buy the notion that a language that has a bit more boilerplate is less productive than one with virtually none.