frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

OpenCiv3: Open-source, cross-platform reimagining of Civilization III

https://openciv3.org/
450•klaussilveira•6h ago•109 comments

The Waymo World Model

https://waymo.com/blog/2026/02/the-waymo-world-model-a-new-frontier-for-autonomous-driving-simula...
791•xnx•12h ago•481 comments

Show HN: Look Ma, No Linux: Shell, App Installer, Vi, Cc on ESP32-S3 / BreezyBox

https://github.com/valdanylchuk/breezydemo
152•isitcontent•6h ago•15 comments

Monty: A minimal, secure Python interpreter written in Rust for use by AI

https://github.com/pydantic/monty
143•dmpetrov•7h ago•63 comments

How we made geo joins 400× faster with H3 indexes

https://floedb.ai/blog/how-we-made-geo-joins-400-faster-with-h3-indexes
19•matheusalmeida•1d ago•0 comments

Dark Alley Mathematics

https://blog.szczepan.org/blog/three-points/
46•quibono•4d ago•4 comments

A century of hair samples proves leaded gas ban worked

https://arstechnica.com/science/2026/02/a-century-of-hair-samples-proves-leaded-gas-ban-worked/
84•jnord•3d ago•8 comments

Show HN: I spent 4 years building a UI design tool with only the features I use

https://vecti.com
257•vecti•8h ago•120 comments

Show HN: If you lose your memory, how to regain access to your computer?

https://eljojo.github.io/rememory/
191•eljojo•9h ago•127 comments

Microsoft open-sources LiteBox, a security-focused library OS

https://github.com/microsoft/litebox
320•aktau•13h ago•155 comments

Sheldon Brown's Bicycle Technical Info

https://www.sheldonbrown.com/
317•ostacke•12h ago•85 comments

Hackers (1995) Animated Experience

https://hackers-1995.vercel.app/
403•todsacerdoti•14h ago•218 comments

An Update on Heroku

https://www.heroku.com/blog/an-update-on-heroku/
328•lstoll•13h ago•236 comments

PC Floppy Copy Protection: Vault Prolok

https://martypc.blogspot.com/2024/09/pc-floppy-copy-protection-vault-prolok.html
19•kmm•4d ago•1 comments

Show HN: R3forth, a ColorForth-inspired language with a tiny VM

https://github.com/phreda4/r3
50•phreda4•6h ago•8 comments

I spent 5 years in DevOps – Solutions engineering gave me what I was missing

https://infisical.com/blog/devops-to-solutions-engineering
110•vmatsiiako•11h ago•34 comments

How to effectively write quality code with AI

https://heidenstedt.org/posts/2026/how-to-effectively-write-quality-code-with-ai/
189•i5heu•9h ago•132 comments

Learning from context is harder than we thought

https://hy.tencent.com/research/100025?langVersion=en
149•limoce•3d ago•79 comments

Make Trust Irrelevant: A Gamer's Take on Agentic AI Safety

https://github.com/Deso-PK/make-trust-irrelevant
7•DesoPK•1h ago•3 comments

Understanding Neural Network, Visually

https://visualrambling.space/neural-network/
240•surprisetalk•3d ago•31 comments

I now assume that all ads on Apple news are scams

https://kirkville.com/i-now-assume-that-all-ads-on-apple-news-are-scams/
985•cdrnsf•16h ago•417 comments

Introducing the Developer Knowledge API and MCP Server

https://developers.googleblog.com/introducing-the-developer-knowledge-api-and-mcp-server/
21•gfortaine•4h ago•2 comments

FORTH? Really!?

https://rescrv.net/w/2026/02/06/associative
43•rescrv•14h ago•17 comments

I'm going to cure my girlfriend's brain tumor

https://andrewjrod.substack.com/p/im-going-to-cure-my-girlfriends-brain
58•ray__•3h ago•14 comments

Evaluating and mitigating the growing risk of LLM-discovered 0-days

https://red.anthropic.com/2026/zero-days/
36•lebovic•1d ago•11 comments

Female Asian Elephant Calf Born at the Smithsonian National Zoo

https://www.si.edu/newsdesk/releases/female-asian-elephant-calf-born-smithsonians-national-zoo-an...
5•gmays•1h ago•0 comments

Show HN: Smooth CLI – Token-efficient browser for AI agents

https://docs.smooth.sh/cli/overview
77•antves•1d ago•57 comments

Show HN: Slack CLI for Agents

https://github.com/stablyai/agent-slack
40•nwparker•1d ago•10 comments

The Oklahoma Architect Who Turned Kitsch into Art

https://www.bloomberg.com/news/features/2026-01-31/oklahoma-architect-bruce-goff-s-wild-home-desi...
20•MarlonPro•3d ago•4 comments

How virtual textures work

https://www.shlom.dev/articles/how-virtual-textures-really-work/
28•betamark•13h ago•23 comments
Open in hackernews

Why “negative vectors” can't delete data in FAISS – but weighted kernels can

https://github.com/nikitph/bloomin/tree/master/negative-vector-experiment
21•loaderchips•1mo ago
The fix for machine unlearning in vector databases turns out to be conceptually simple, but it requires changing the semantics of retrieval.

Standard FAISS-style indices store vectors and compute:

argmax ⟨q, vᵢ⟩

If you insert -v, nothing happens. It’s just another point. The original vector is still maximally similar to itself and remains rank-1.

This isn’t a bug—it’s a consequence of selection-based retrieval.

If instead you store (vector, weight) pairs and evaluate: φ(q) = Σ wᵢ · K(q, vᵢ)

you get a different object entirely: a field, not a selection. Now inserting the same vector with w = −1 causes destructive interference. The contribution cancels. The attractor disappears.

Deletion becomes O(1) append-only (add the inverse), not a structural rebuild.

FAISS-style: Vec<Vec<f32>> → argmax (selection) Weighted form: Vec<(Vec<f32>, f32)> → Σ (field)

We validated this on 100k vectors: • FAISS: target stays rank-1 after “deletion” • Field-based model: exact cancellation (φ → 0), target unretrievable

The deeper point is that this isn’t a trick—it’s a semantic separation. • FAISS implements a selection operator over discrete points. • The weighted version implements a field operator where vectors act as kernels in a continuous potential. • Retrieval becomes gradient ascent to local maxima. • Deletion becomes destructive interference that removes attractors.

This shifts deletion from structural (modify index, rebuild, filter) to algebraic (append an inverse element). You get append-only logs, reversible unlearning, and auditable deletion records. The negative weight is the proof.

Implication: current vector DBs can’t guarantee GDPR/CCPA erasure without reconstruction. Field-based retrieval can—provably.

Paper with proofs: https://github.com/nikitph/bloomin/blob/master/negative-vect...

Comments

jey•1mo ago
That makes sense, but how do you efficiently evaluate the Gaussian kernel based approach (“operator-based data structures (OBDS)”)? Presumably you want to do it in a way that keeps a dynamically updating data structure instead of computing a low rank approximation to the kernel etc? In my understanding the upside of the kNN based approaches are fast querying and ability to dynamically insert additional vectors..?
loaderchips•1mo ago
Thank you for the thoughtful comment. Your questions are valid given the title, which I used to make the post more accessible to a general HN audience. To clarify: the core distinction here is not kernelization vs kNN, but field evaluation vs point selection (or selection vs superposition as retrieval semantics). The kernel is just a concrete example.

FAISS implements selection (argmax ⟨q,v⟩), so vectors are discrete atoms and deletion must be structural. The weighted formulation represents a field: vectors act as sources whose influence superposes into a potential. Retrieval evaluates that field (or follows its gradient), not a point identity. In this regime, deletion is algebraic (append -v for cancellation), evaluation is sparse/local, and no index rebuild is required.

The paper goes into this in more detail.

CamperBob2•1mo ago
Hey man, nice slop
ricochet11•1mo ago
This isn’t just X, it’s Y.