frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

What Happens When Technical Debt Vanishes?

https://ieeexplore.ieee.org/document/11316905
1•blenderob•40s ago•0 comments

AI Is Finally Eating Software's Total Market: Here's What's Next

https://vinvashishta.substack.com/p/ai-is-finally-eating-softwares-total
1•gmays•1m ago•0 comments

Computer Science from the Bottom Up

https://www.bottomupcs.com/
1•gurjeet•1m ago•0 comments

Show HN: I built a toy compiler as a young dev

https://vire-lang.web.app
1•xeouz•3m ago•0 comments

You don't need Mac mini to run OpenClaw

https://runclaw.sh
1•rutagandasalim•3m ago•0 comments

Learning to Reason in 13 Parameters

https://arxiv.org/abs/2602.04118
1•nicholascarolan•5m ago•0 comments

Convergent Discovery of Critical Phenomena Mathematics Across Disciplines

https://arxiv.org/abs/2601.22389
1•energyscholar•6m ago•1 comments

Ask HN: Will GPU and RAM prices ever go down?

1•alentred•6m ago•0 comments

From hunger to luxury: The story behind the most expensive rice (2025)

https://www.cnn.com/travel/japan-expensive-rice-kinmemai-premium-intl-hnk-dst
1•mooreds•7m ago•0 comments

Substack makes money from hosting Nazi newsletters

https://www.theguardian.com/media/2026/feb/07/revealed-how-substack-makes-money-from-hosting-nazi...
5•mindracer•8m ago•1 comments

A New Crypto Winter Is Here and Even the Biggest Bulls Aren't Certain Why

https://www.wsj.com/finance/currencies/a-new-crypto-winter-is-here-and-even-the-biggest-bulls-are...
1•thm•8m ago•0 comments

Moltbook was peak AI theater

https://www.technologyreview.com/2026/02/06/1132448/moltbook-was-peak-ai-theater/
1•Brajeshwar•9m ago•0 comments

Why Claude Cowork is a math problem Indian IT can't solve

https://restofworld.org/2026/indian-it-ai-stock-crash-claude-cowork/
1•Brajeshwar•9m ago•0 comments

Show HN: Built an space travel calculator with vanilla JavaScript v2

https://www.cosmicodometer.space/
2•captainnemo729•9m ago•0 comments

Why a 175-Year-Old Glassmaker Is Suddenly an AI Superstar

https://www.wsj.com/tech/corning-fiber-optics-ai-e045ba3b
1•Brajeshwar•9m ago•0 comments

Micro-Front Ends in 2026: Architecture Win or Enterprise Tax?

https://iocombats.com/blogs/micro-frontends-in-2026
1•ghazikhan205•11m ago•0 comments

These White-Collar Workers Actually Made the Switch to a Trade

https://www.wsj.com/lifestyle/careers/white-collar-mid-career-trades-caca4b5f
1•impish9208•12m ago•1 comments

The Wonder Drug That's Plaguing Sports

https://www.nytimes.com/2026/02/02/us/ostarine-olympics-doping.html
1•mooreds•12m ago•0 comments

Show HN: Which chef knife steels are good? Data from 540 Reddit tread

https://new.knife.day/blog/reddit-steel-sentiment-analysis
1•p-s-v•12m ago•0 comments

Federated Credential Management (FedCM)

https://ciamweekly.substack.com/p/federated-credential-management-fedcm
1•mooreds•13m ago•0 comments

Token-to-Credit Conversion: Avoiding Floating-Point Errors in AI Billing Systems

https://app.writtte.com/read/kZ8Kj6R
1•lasgawe•13m ago•1 comments

The Story of Heroku (2022)

https://leerob.com/heroku
1•tosh•13m ago•0 comments

Obey the Testing Goat

https://www.obeythetestinggoat.com/
1•mkl95•14m ago•0 comments

Claude Opus 4.6 extends LLM pareto frontier

https://michaelshi.me/pareto/
1•mikeshi42•15m ago•0 comments

Brute Force Colors (2022)

https://arnaud-carre.github.io/2022-12-30-amiga-ham/
1•erickhill•17m ago•0 comments

Google Translate apparently vulnerable to prompt injection

https://www.lesswrong.com/posts/tAh2keDNEEHMXvLvz/prompt-injection-in-google-translate-reveals-ba...
1•julkali•18m ago•0 comments

(Bsky thread) "This turns the maintainer into an unwitting vibe coder"

https://bsky.app/profile/fullmoon.id/post/3meadfaulhk2s
1•todsacerdoti•19m ago•0 comments

Software development is undergoing a Renaissance in front of our eyes

https://twitter.com/gdb/status/2019566641491963946
1•tosh•19m ago•0 comments

Can you beat ensloppification? I made a quiz for Wikipedia's Signs of AI Writing

https://tryward.app/aiquiz
1•bennydog224•20m ago•1 comments

Spec-Driven Design with Kiro: Lessons from Seddle

https://medium.com/@dustin_44710/spec-driven-design-with-kiro-lessons-from-seddle-9320ef18a61f
1•nslog•20m ago•0 comments
Open in hackernews

Being a Force Multiplier

https://substack.com/home/post/p-165651243
20•jandrewrogers•7mo ago

Comments

roenxi•7mo ago
This article could do with a good edit to cut out the middle section, which appears to be a list of mostly meaningless platitudes. Although I can't argue with great managers "organise ... a free team lunch" - managers of the world take note; deficiencies on the free lunch front could be what is holding you back from greatness.

The basic idea of greatness being small optimisations in a large number of areas is worth repeating a few times though. The majority of greatness comes from avoiding making any well known basic mistakes and a strategy of working through all the details and checking for small problems can do a lot to enable that. Big dramatic gestures generally do not.

davedx•7mo ago
Yeah I recall reading this works in industrial contexts too (lots of small optimisations accumulating into significant gains).
hansmayer•7mo ago
Ugh. Please lets stop adopting military concepts into the field of business leadership. Given that most humans and engineers in particular perform best when in position of having high autonomy, which is exactly the opposite of military environment, why do we keep borrowing from there? Is it because all the expired military "experts" who are trained to fit in and not think for themselves, lost all the wars in the past two decades and need a new job? Why do we allow people who are severely under-educated even compared to a junior-LLM-assisted rookie to tell us what we need to do? No, please don't be a "force multiplier", just look around and do what makes sense in your specific environment. You are way smarter than that.
eptcyka•7mo ago
I think most people feel like they’re performing well when given a high level of autonomy, which is not the same as performing well.
hansmayer•7mo ago
Well, no. This was proven scientifically a long-time ago, so no, what you think does not disprove what was already proved by scientific experimentation of workplace psychologists back in the 70s...
eptcyka•7mo ago
Could you reference some of those works? I do not disagree that high performers need a high degree of autonomy, but I do not believe that anyone can be turned into a high performer by just adding more autonomy to their work life.
hansmayer•7mo ago
Not off the top of my head right now, but pick up any classic book on managing software teams and you will find heaps of such scientific references, for example "Peopleware" comes to mind.
roenxi•7mo ago
People who aren't high performers, as a general rule, can't be transformed into them by any system (although there are some fascinating explorations of what a low performer can achieve with the right capital investments). A good strategy is one that either achieves the best possible results with a large number of average performers or turns medium-high performers into high performers and really enables the high performers to shine.
hansmayer•7mo ago
Really? So you are saying things are set in stone?
ofwellkgi•7mo ago
It shouldn't come as a suprise to anyone that happy people make way better workers than angry and sad ones. Autonomy makes people happy, and on top of that experts usually know what they're doing, some even like their field and actually enjoy taking on challenges.

I think organizations of any type or size have a habit of discounting the power of spite aswell. You can do way worse than lose productivity, revolutions happen because people are unhappy.

MoreQARespect•7mo ago
>Given that most humans and engineers in particular perform best when in position of having high autonomy

High autonomy militaries outperform low autonomy militaries too.

drcongo•7mo ago
Author could have used Substack itself as an example - it's a force multiplier for right wing conspiracy theories and propaganda.
asplake•7mo ago
> You don’t obsess over one thing. You move lots of little things forward. No grand initiatives. No reorg. Just constant, low-key, under-the-radar nudging in the right direction.

It's not terrible advice, but it scales less well than the writer thinks. To really scale, you:

1. Engage with the right challenges (large or small)

2. Invite others into the process, celebrate their successes etc

3. Coach others to start from #1

Perhaps its organisational scope isn't much bigger than the team, but to my mind, the article doesn't go far enough beyond #2.

Do it, and you're the best kind of leader, one that makes other leaders. That's what scales.