vercantez

Oh that's embarrassing thanks for the feedback. Will fix docs. Sorry about that. Glad it's working!

Great to see. Could have use this last month when we migrated from OpenNext on CF to React Router 7

We'll reverse engineer our way out of planned obsolescence

> I personally believe that harnesses matter almost as much as the models in 2026.

Agreed. Great work

I love the plotly graphs, especially the sankey diagram. Seems HN may be an older crowd than I expected.

Updated with weighted analysis.

UPDATE: Added a weighted average analysis based on story points and comments. SQLite ranks highest in points per story and Redis ranks highest in comments per post. Also added SQLite to the growth table. I had accidentally deleted this row in the original post.

Fixed.

SQLite is now reflected in the growth table

Purely based on headline occurrence but weighing based on storypoints and comments is a great idea. I'll update the blog, thanks.

We use cloudflare turnstile. Sometimes it blocks some VPNs. Very rarely it blocks some browsers.

Mistake in the analysis. Fixing now.

Yeah I found a mistake in the analysis. I'm updating the post to reflect SQLite's popularity.

Wow. One of the absolute greatest. The world truly is a different place because of Bill. Bill’s importance in the history of computing cannot be overstated. Hypercard is probably my favorite invention of his. So ahead of its time. Rest in peace Bill

Very cool! Actually practical use of scaling parallel test time compute. I've been using worktrees + agents to work on separate features but never considered allocating N agents per task.

I messed up the url. Go to https://app.camelai.com/hn/

Thank you for the feedback. This product was made to connect to your own database. I thought it was fun to connect it to the HN bigquery public dataset. We are break even on a good month.

Sorry! I noticed this after I submitted. The link should be https://app.camelai.com/hn/

Thanks for the feedback! We've noticed o3 doesn't tend to make graphs when it should but sonnet makes too many graphs... We'll have to keep tweaking this. Mobile definitely needs some work but I'm glad it worked for you.

Thanks for the clarification dang. I was misled by the listing which lists the author/publisher as "Y Combinator". Thanks for offering the official API.

Hacker News publishes this dataset freely to the bigquery data marketplace

https://console.cloud.google.com/marketplace/product/y-combi...

This product was built to connect to your own database but I thought it was fun to connect to the HN dataset

Go to app.camelai.com/hn/ or click the title link!

Good idea. Will do that

Not scraped. HN itself publishes a live dataset to bigquery. The product is meant to connect to your own database but I thought this was fun to connect to hn

Thanks for catching that! Should be fixed once the cache invalidates

> Limit additional context in retrieval-augmented generation (RAG): When providing additional context or documents, include only the most relevant information to prevent the model from overcomplicating its response.

This databricks blog/paper seemingly contradicts this:

> OpenAI o1 models show a consistent improvement over Anthropic and Google models on our long context RAG Benchmark up to 128k tokens.

https://www.databricks.com/blog/long-context-rag-capabilitie...

Wish they explained how they perform targeted edits. Is it doing speculative decoding like Anthropic’s code edit or something like Aider’s diff format?

This is impressively fast. Well done!

We originally spelled it qaml which stood for quality assurance, machine learning. That wasn't obvious to anyone and everyone would pronounce it as "qwamel" which we hated so we decided on camelQA instead. Thanks for the question. I was waiting for someone to ask it!

Thanks! Yes inference costs are non-negligible right now but we think this will come down over time