frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

Open in hackernews

Alephic Writing Style Guide

https://www.alephic.com/company/writing
29•otoolep•3d ago

Comments

noahbrier•3d ago
Thanks for the shout-out, glad you enjoyed it.
brudgers•14h ago
No space for poetry.

No space for fun.

All arguments from authority

That only missionary position

Is permissable.

throwanem•1h ago
Well, I'm sure it makes a decent system prompt.
Barrin92•1h ago

  Do                                  Don't
  Think from first principles         Repeat what everyone else is saying Draw from 
  diverse references and domains      Stay confined to marketing jargon
"We paint visions of transformative change while keeping one foot firmly in practical reality."

Famously non-jargon, non-marketing language never encountered before in a tech blog

ozornin•1h ago
Another writing style guide, barely distinguishable from many others, but written in tone like it's radically different. Maybe this is precisely the point, though.
suddenlybananas•1h ago
> Use "1800s/1900s/2000s" instead of "19th/20th/21st century". Using century numbers is confusing for many people because the 19th century actually refers to the 1800s. For clarity, always use the specific years.

Way to assume your audience are morons.

skrebbel•1h ago
Personally I love it when people do this. It's just.. it saves me the few seconds it takes, every time, to subtract "1" from the number.

It's not that hard with "20th century" which I'm conditioned to map to "last century" immediately, but when someone did something in the 17th century I first think "Oh like 1750-ish" and then I realize I'm a century off, and it disrupts my reading flow.

nashashmi•6m ago
[delayed]
skrebbel•1h ago
I recognize that the following is a classic HN "middlebrow dismissal" but I'd be wary of taking writing advice from a company that has this text (and nothing else) on their homepage:

> Alephic is an AI-first technology foundry built to tackle marketing's most complex challenges. We don't just advise—we engineer, prototype, and deploy custom AI systems that help marketing teams do the impossible.

I read this twice and I still have no idea what they do!

Veen•41m ago
That’s not bad writing; it's deliberate ambiguity. Alephic wants potential clients to understand that they do "AI stuff," so when a company is looking for help with "AI stuff", they can say, "Yes! Of course we do that!" — regardless of the specifics. B2B services companies frame their offering ambiguously to cast the widest possible net. They'll write a custom pitch once they know what a lead actually needs.
BrenBarn•16m ago
That's especially ludicrous considering the linked article, which is listing a bunch of style guidelines which would probably make an AI stand out like a sore thumb.
kristel100•27m ago
Refreshing take. Especially liked the emphasis on frictionless rhythm. Most style guides focus too much on rules and not enough on what it feels like to read the thing.
notepad0x90•18m ago
This is nice but they forgot to mention the most important rule all guides like this should mention first: There are no hard-and-fast rules, all rules have an exception.

For example, including unnecessary sentences and paragraphs is somethings necessary. You can do without them but with them you get character, voice, a smoother transition. How do you know what is necessary and what isn't? That's the whole point of the rule I mentioned earlier.

Firefox moves to GitHub

https://github.com/mozilla-firefox/firefox
378•thefilmore•4h ago•226 comments

FastVLM: Efficient vision encoding for vision language models

https://github.com/apple/ml-fastvlm
256•nhod•8h ago•45 comments

A programming language made for me

https://zylinski.se/posts/a-programming-language-for-me/
14•gingerBill•57m ago•4 comments

Ask HN: How are you acquiring your first hundred users?

10•amanchanda•51m ago•11 comments

TransMLA: Multi-head latent attention is all you need

https://arxiv.org/abs/2502.07864
66•ocean_moist•6h ago•12 comments

Open Hardware Ethernet Switch project, part 1

https://serd.es/2025/05/08/Switch-project-pt1.html
138•luu•4d ago•17 comments

Anti-Personnel Computing (2023)

https://erratique.ch/writings/anti-personnel-computing
10•transpute•1h ago•3 comments

15 Years of Shader Minification

https://www.ctrl-alt-test.fr/2025/15-years-of-shader-minification/
84•laurentlb•3d ago•16 comments

Air Traffic Control

https://computer.rip/2025-05-11-air-traffic-control.html
173•1317•1d ago•57 comments

The Barbican

https://arslan.io/2025/05/12/barbican-estate/
541•farslan•18h ago•200 comments

Alephic Writing Style Guide

https://www.alephic.com/company/writing
29•otoolep•3d ago•12 comments

Revisiting Image Maps

https://css-tricks.com/revisiting-image-maps/
25•thm•3d ago•12 comments

We Fixed 2k+ Security Issues (2023)

https://www.sudhishkr.com/posts/20230609_how-we-fixed-2kplus-security-issues/
3•mooreds•1d ago•0 comments

Can you trust that permission pop-up on macOS?

https://wts.dev/posts/tcc-who/
288•nmgycombinator•15h ago•198 comments

A conversation about AI for science with Jason Pruet

https://www.lanl.gov/media/publications/1663/0125-qa-jason-pruet
148•LAsteNERD•13h ago•125 comments

RIP Usenix ATC

https://bcantrill.dtrace.org/2025/05/11/rip-usenix-atc/
167•joecobb•17h ago•37 comments

Understanding LucasArts' iMUSE System

https://github.com/meshula/LabMidi/blob/main/LabMuse/imuse-technical.md
121•todsacerdoti•10h ago•26 comments

How to avoid P hacking

https://www.nature.com/articles/d41586-025-01246-1
59•benocodes•4d ago•50 comments

A community-led fork of Organic Maps

https://www.comaps.app/news/2025-05-12/3/
301•maelito•21h ago•195 comments

HealthBench – An evaluation for AI systems and human health

https://openai.com/index/healthbench/
155•mfiguiere•15h ago•129 comments

University of Texas-led team solves a big problem for fusion energy

https://news.utexas.edu/2025/05/05/university-of-texas-led-team-solves-a-big-problem-for-fusion-energy/
245•signa11•21h ago•167 comments

Launch HN: ParaQuery (YC X25) – GPU Accelerated Spark/SQL

118•winwang•17h ago•71 comments

Reviving a modular cargo bike design from the 1930s

https://www.core77.com/posts/136773/Reviving-a-Modular-Cargo-Bike-Design-from-the-1930s
183•surprisetalk•19h ago•146 comments

Trial by Fire: The crash of Aeroflot flight 1492

https://admiralcloudberg.medium.com/trial-by-fire-the-crash-of-aeroflot-flight-1492-ee61cebcf6ec
5•shmeeed•2h ago•2 comments

Writing N-body gravity simulations code in Python

https://alvinng4.github.io/grav_sim/5_steps_to_n_body_simulation/
126•dargscisyhp•3d ago•23 comments

Offline vs. online ML pipelines

https://decodingml.substack.com/p/offline-vs-online-ml-pipelines
4•rbanffy•3d ago•0 comments

Ruby 3.5 Feature: Namespace on read

https://bugs.ruby-lang.org/issues/21311
206•ksec•19h ago•97 comments

NASA study reveals Venus crust surprise

https://science.nasa.gov/science-research/astromaterials/nasa-study-reveals-venus-crust-surprise/
78•mnem•3d ago•77 comments

FedRAMP 20x – One Month in and Moving Fast

https://www.fedramp.gov/2025-04-24-fedramp-20x-one-month-in-and-moving-fast/
80•transpute•9h ago•53 comments

Policy of Transience

https://www.chiark.greenend.org.uk/~sgtatham/quasiblog/transience/
35•pekim•2d ago•7 comments