frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

Know What Your Goals Are and Why

https://dontbreakprod.com/posts/know-what-your-goals-are-and-why
1•dorkrawk•5m ago•0 comments

Google's harmful and careless use of AI

1•bbarnett•6m ago•1 comments

Bank of Canada Identifies Technical Path for Retail CBDC in New Research Paper

https://www.coindesk.com/business/2025/07/04/bank-of-canada-identifies-technical-path-for-retail-cbdc-in-new-research-paper
2•PaulHoule•9m ago•0 comments

Ant Mill

https://en.wikipedia.org/wiki/Ant_mill
1•piniiforever•9m ago•0 comments

What My Mother Didn't Talk About (2020)

https://www.buzzfeednews.com/article/karolinawaclawiak/what-my-mother-didnt-talk-about-karolina-waclawiak
1•NaOH•9m ago•0 comments

Show HN: Empromptu.ai – Agentic AI Building AI Apps

1•anaempromptu•12m ago•0 comments

PSA: Part of your Kagi subscription fee goes to a Russian company (Yandex)

https://ounapuu.ee/posts/2025/07/17/kagi/
2•LorenDB•12m ago•0 comments

Don't Fall for AI: Reasons for Writers to Reject Slop

https://mythcreants.com/blog/dont-fall-for-ai-nine-reasons-for-writers-to-reject-slop/
2•BerislavLopac•14m ago•0 comments

A £3.93/mo Nomad‑backed learning lab: Next.js · .NET · Postgres on a budget

https://craig.banach.dev/musings/4-pound-learning-lab
1•craigbanach•15m ago•1 comments

Vibe Game Dev: Nimble Fox Unity AI Assistant [video]

https://www.youtube.com/watch?v=F4l2ULvMsko
2•lastdong•16m ago•0 comments

Telnyx AI Agents now support native integration with MCP servers

1•maevesentner•17m ago•0 comments

Grep a million GitHub repositories via MCP

https://vercel.com/blog/grep-a-million-github-repositories-via-mcp
2•cramforce•17m ago•0 comments

MirageLSD: The First Live-Stream Diffusion AI Video Model

https://about.decart.ai/publications/mirage
2•jasondavies•18m ago•0 comments

Microsoft Names Threat Actors

https://learn.microsoft.com/en-us/unified-secops-platform/microsoft-threat-actor-naming
2•grepcat•22m ago•0 comments

MirageLSD: The First Live-Stream Video Diffusion Model (∞-Generation, 0-Latency)

https://mirage.decart.ai/
3•dnnssl2•23m ago•5 comments

Show HN: I made EmojiAssets: click an emoji, get favicons and Apple touch icons

https://www.emojiassets.com
1•volukren•23m ago•0 comments

The Illusion of Biosafety During SARS-CoV-2 Research

https://www.researchgate.net/publication/390171996_The_Illusion_of_Biosafety_During_SARS-CoV-2_Research_Multiple_Apparent_Occult_Lab-Acquired_Infections_Are_Identified_including_from_Synthetic_Infectious_Clones_Under_BSL-3_Conditions_at_a_Premier_US-b
3•hggh•24m ago•0 comments

Show HN: I built a context-engineering CLI/MCP tool

https://github.com/jerpint/context-llemur
2•jerpint•24m ago•0 comments

Show HN: FavBox is a local-firs browser extension for bookmark management

https://github.com/dd3v/favbox
2•dm_dd3v•25m ago•0 comments

Trump's Impact on International Student Enrollment

https://spectrum.ieee.org/trump-international-students
3•rbanffy•25m ago•0 comments

Your Boss Still Thinks AI Is a Fancy Calculator

https://substack.com/home/post/p-168583724
1•anonym29•25m ago•1 comments

Ask HN: Has anyone ever manage to Chromecast from Chrome Mobile browser?

1•01jonny01•25m ago•0 comments

AWS Lambda silent crash – a platform failure, not an application bug

https://aaronstuyvenberg.com/posts/does-lambda-have-a-silent-crash
3•felineflock•30m ago•0 comments

Wave 11: Just Keep Shipping

https://windsurf.com/blog/windsurf-wave-11
1•pr337h4m•32m ago•0 comments

Comparing Claude Code vs. OpenCode (and testing different models)

https://www.andreagrandi.it/posts/comparing-claude-code-vs-opencode-testing-different-models/
2•andreagrandi•33m ago•0 comments

Show HN: TrueSift – AI-Powered Real-Time Fact-Checking Chrome Extension

https://www.truesift.dev/
1•terrib1e•36m ago•0 comments

Substack raises $100M at $1.1B valuation

https://techcrunch.com/2025/07/17/substack-raises-100m-from-chernin-group-andreessen-horowitz-skims-ceo-and-more/
4•1zael•39m ago•1 comments

Anthropic tightens usage limits for Claude Code – without telling users

https://techcrunch.com/2025/07/17/anthropic-tightens-usage-limits-for-claude-code-without-telling-users/
36•mfiguiere•42m ago•15 comments

To survive the AI age, the web needs a new business model

https://www.economist.com/leaders/2025/07/17/to-survive-the-ai-age-the-web-needs-a-new-business-model
2•sien•43m ago•1 comments

Are We Alone?

https://www.writervivek.com/2025/07/Are-we-alone.html
4•totaldude87•49m ago•0 comments
Open in hackernews

Ask HN: Is the Ask HN algorithm broken?

12•alexathrowawa9•3h ago
I don't understand the ranking algorithm looking in the ask section on HN

For example I saw a post with time 6 hours ago with 7 upvotes ranks higher than a post with 22 upvotes 1 hour ago?

What is the logic of the ranking?

I've noticed this a lot recently, as in the last year or so

Also why is a post from 2 weeks ago still on the front page of ask?

https://news.ycombinator.com/item?id=44448461

Anyone understand this algo?

Comments

isaacvando•3h ago
How did this end up on the front page with 0 comments??
charliebwrites•3h ago
By making this comment you changed the outcome
Jtsummers•3h ago
Enough upvotes in a short enough time. Now it'll be up there a bit longer because there's active discussion.
mellosouls•2h ago
mod mischief
heywintermute•3h ago
Mods have a way of boosting posts to refresh them back into the algorithm hence why you see some posts weeks ago showing back up
tough•3h ago
second chance pool
toomuchtodo•3h ago
https://news.ycombinator.com/item?id=26998308
Jtsummers•3h ago
https://news.ycombinator.com/item?id=44595003

I'm guessing you're asking because of this submission (yours) from a little while ago that's somehow even off the Ask page, not just the front page, despite having a good amount of active discussion.

Ask HN seems to be more sensitive to two things than other discussions:

1. Comment:Score ratio. The ratio of comments to points is 3:1, 69 comments and 23 points as I write this. None of the discussion threads are very deep, but when a discussion has a lot of comments but not a lot of points it tends to get moderated (automatically, not by dang & co.) down faster. This probably relates to the flame war detection, though I've seen that kick in more often when comment threads get very deep and more linear than branching.

2. Flags. When Ask HN submissions get flagged, they move down faster than article submissions.

The Parekh discussion has a high score and a high comment count, so (1) is not applicable to it (more comments than points, but more like 4:3 so not over whatever the threshold may be). I have no idea if it was flagged, but if it was since it had active discussion the mods can turn off flags (as it impacts rankings) on a submission by submission basis.

alexathrowawa9•1h ago
Makes sense, I can see #1 being an issue
elpocko•2h ago
The points shown on a post are misleading. There are other, mostly invisible factors influencing the ranking of all posts, not just Ask HN. You can't even trust the timestamps on posts and comments, because mods occasionally decide to falsify them and make you think you're going crazy by inducing feelings of deja-vu.