wernerb

Joined 13 years ago

Karma
298
Submissions
172
Member since
Oct 26
[ my public key: https://keybase.io/buck; my proof: https://keybase.io/buck/sigs/oNMUWC_61MuH3Las0BztFz1dDDJqKbK4E9Lq8fPujeA ]
View on Hacker News

Recent submissions

Comment · 12 hours ago
This can be seen as a "look europe, we are trying to address our huge debt peoblem (which is 120% of gdp in 2026), please continue to help us". France as well as the French themselves do not want to address the extreme cost of their early pensions to their budget (25% of the 2026 budget!). The retirees and employees often get amazing perks from the government. Employees and retirees of EDF get a 90% discount on their gas and electric bill. 21% of the total workforce works for government... Meanwhile kids can't get a decent education, I would be pissed off too.
Comment · 1 month ago
Would be great to have the models be easily splittable to decouple the "brain-part" that is aware of external systems/internet of the "brain-part" that is actually being tested. Afterwards the brains are joined again.
Comment · 1 month ago
Meta in their lawsuit negotiated 5B to be paid out only if other market players agree on implementing safeguards. In the same vein OpenAI can slow down competition, lock out opensource AI.

> OpenAI said on Wednesday that it was pushing for mandatory national AI safety requirements in the United States, citing concerns that advanced AI systems could accelerate its own development.

As well, the weaponisation of AI is already there, open ai can switch off safeguards at will. A massive swarm was allocated to solving Millenium problems in a few days. Imagine the US having such a weapon, they want to ensure other nations cannot have this kind of access. I imagine the US might be very interested in stifling globaal Innovation and be sympathetic to OpenAI for these reasons.

Comment · 1 month ago
A modern heat pump (r290) uses propane has a negligible GWP of 3
Comment · 2 months ago
> The classic answer is a monitoring platform: ....... All great tools. But for a few dozen checks, I did not want to operate a second distributed system just to know whether the first one is up.

There is advantage to keeping to "classic" tooling especially if you are not in an "I" context but in a "team" context. How large is the team that is going to use and maintain this custom tooling?

Comment · 3 months ago
A Kobo ereader is as open as I need it with koreader installed. The Kobo Libra 2 black and white is the best ereader I have ever had. (not the color version,,unfortunately a step back in reading quality for normal reading)
Comment · 3 months ago
Death stranding 1 and 2 also look absolutely amazing. They are not using unreal engine but their own. Not sure what magic they are pulling off there.
Comment · 4 months ago
Multipurpose too for the washer/dryer settings
Comment · 4 months ago
Won't the power that is saved not just be used for more LLMs at this stage?
Comment · 4 months ago
They have millions of conversations with code prompts and diffs and the user telling them the model is taking a wrong turn. This is quite valuable data that is probably already used to tweak Kimi code (composer 2). Moreover they have quite the enterprise client base that I suspect at least a portion of which will jump ship with the acquisition.
Comment · 6 months ago
A reason not to run k8s is if you want your server to reach C10 idle states. The k8s control plain with its polling and checking are quite heavy on the mostly idle server. I have reverted to just use Nixos and oci podman containers. Everything is declarative and reproducible
Comment · 6 months ago
DevOps lost the plot with the Operator model. When it was being widely introduced as THE pattern I was dismayed. These operators abstract entirely complex services like databases behind yaml and custom go services. When going to kubecon i had one guy tell me he collects operators like candy. Answers on Lifecycle management, and inevitable large architectural changes in an ever changing operator landscape was handwaved away with series of staging and development clusters. This adds so much cost.. Fundamentally the issue is the abstractions being too much and entirely on the DevOps side of the "shared responsibility model". Taking an RDBMS from AWS of Azure is so vastly superior to taking all that responsibility yourself in the cluster.. Meanwhile (being a bit of an infrastructure snob) I run Nixos with systemd oci containers at home. With AI this is the easiest to maintain ever.
Comment · 6 months ago
Regarding GitHub actions and it's secret manager. Any decently organized company would do well to stay away from well known secret interfaces. Instead use oidc auth to fetch secrets just in time, all short-lived for the duration of the pipeline.
Comment · 8 months ago
If there's ever any use case to leave an expensive SaaS for self hosted, you can find it at datadog
Comment · 9 months ago
We keep standardising without adding versioning :(

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton