fzeindl

I once had an idea of buying the domain "freeofcharge.org", where people could put useful services that fit into RAM onto subdomains, meaning services that cost them only ~10$ per month, which they pay out of their own pockets.

Obviously all of what dang said, but I want to add that I think timing is an additional factor.

If you post when silicon valley wakes up on a weekday, you might get “initial” points faster, which leads to your submission being ranked higher up for a while and being more discoverable.

The site doesn‘t mention it, I got that information from various german announcements. I fear there probably won‘t be an English announcement regarding the budget, though there will be many regarding the tunnel.

In a world where large infrastructure projects regularly exceed their budget by 100-1000%, 7% is huge.

It is important to show people that that is possible in government projects.

If you find a more concise way of saying „with unusually small overrun of it‘s budget“ tell me.

Also there were sections added after the initial estimate.

Agreed, it was just important to me to point it out, since staying within such a massive budget on such a long timeline is a rare achievement.

It is mostly within budget, estimated in 2005 were 5.5 billion €, total cost as of today are 5.9 billion €, the difference being largely attributed to the pandemic and later addition of sections.

Bruce Schneier put it well:

"Willison’s insight was that this isn’t just a filtering problem; it’s architectural. There is no privilege separation, and there is no separation between the data and control paths. The very mechanism that makes modern AI powerful - treating all inputs uniformly - is what makes it vulnerable. The security challenges we face today are structural consequences of using AI for everything."

- https://www.schneier.com/crypto-gram/archives/2025/1115.html...

> If you plot the line it's probably still curving up and I'm not clear at which point (if ever) it would start bending the other way.

I suspect when Moore‘s law ends and we cannot build substantially faster machines anymore.

What is the benefit over Elixir?

In german the usual saying in meetings is, literally translated „I can‘t say anything about that (optional: at the present moment)“.

That sentence is widely accepted as a substitute for „I don‘t know“, but at the same time doesn‘t hold the same conclusion, since there could be a myriad of reasons why you can‘t say anything about a topic.

ISO8601 durations should be used, like PT3M.

Apart from the reasons for this block: Why do these decisions always have to be black and white. I believe it would benefit mental health if Facebook was blocked one day per week so people are forced to live a day without it.

Same with combustion vehicles and the climate: block cars in cities a couple of days per week, individually selected per person.

> I see this kind of cynicism often on this site and I frankly find it inexplicable. It doesn't seem warranted. The vast majority of therapists clearly care about their patients.

I find the the view cynic as well, that's why I wrote "apart from compassion and morality", which you failed to include in your quote.

Exactly. That might be something interesting to think about. Humans make mistakes. LLMs make mistakes.

Yet for humans we have built a society which prevents these mistakes except in edge cases.

Would humans make these mistakes as often as LLMs if there would be no consequences?

> There is no comparison to therapists. Because a therapist would NEVER do that unless wanting to cause harm.

Some therapists ultimately might. It occurs that therapists were stripped of their licenses for leading abusive sects:

https://en.m.wikipedia.org/wiki/Center_for_Feeling_Therapy

> An AI Model designed to match my tone and be sycophantic to my every whim. It would have killed me.

Matching tones and being sycophantic to every whims. Just like many really bad therapists. Only they are legally responsible if they cause a death, which makes them care (apart from compassion and morality).

The criminal justice system is also a system for preventing individuals who perform unwanted action from doing them again.

You can’t punish AI for messing up. You would need to pull it out of circulation on each major screw up, which isn’t financially feasible, and you would need to make it want to prevent that.

I find Stallmans views are best summed up by this quote from him:

“I could have made money this way, and perhaps amused myself writing code. But I knew that at the end of my career, I would look back on years of building walls to divide people, and feel I had spent my life making the world a worse place.”

I think before we can fix all that we need to revert the renting of software via subscriptions and go back to one-time-payment. But people are too greedy for that.

Have that here in Austria from time to time as well.

Funniest case I visited was a Chinese restaurant where the waitress always wore a winter jacket, because they couldn’t be bothered heating the place.

Agreed, but my point is that an abstraction should provide discernible gain of value, not merely roll a handful of operations of the same level into one.

I can read about 50% of the words.

http://www.gnovies.com exists since 2002.

(Other media: http://www.gnod.com)

> I love the idea of an on-device model that I can say something like "who's going to the baseball game this weekend" and it'll intelligently check my calendar and see who's listed.

The question is whether you are ok with the model naming someone who isn‘t listed, or failing to name someone who is listed.

Some developers love to implement Rube-Goldberg-machines as tooling.

Whenever they need to make two or three manual steps or configurations, they rather develop an abstraction layer where you just have to press one button to perform them.

Until that button needs to be accompanied by another manual action. Then they will again develop an abstraction which encapsulates the press of the button along with the augmenting action and is triggered by the pressing of a superior button.

Example: Docker->Kubernetes->Helm or any other tooling that uses YAML to write YAML.

This. I‘m always amazed on how LLMs are praised for being able to churn out the large amount of code we apparently all need.

I keep wondering why. All projects I ever saw need lines of code, nuts and bolts removed instead of added. My best libraries consist of a couple of thousand lines.

1. how does this add value over mermaid? I like how it looks and works, but is there any real reason to switch over.

2. in my opinion, what all these tools are missing is the ability to add a manual „corrective“ layer, meaning to be able to adjust the automatically created output by moving and resizing boxes by a certain x/y amount.

> The sad fact is that the software industry as a whole relies heavily on mainstream languages, and not on educating your engineers to leverage the strengths of your specialty language.

That is an excellent remark.

I agree, but the main feature isn’t the language, but the BEAM VM.

As far as I know there‘s no other language which focussed from the get-go on parallel, distributed deployment, has processes as first class primitives, supports hot code swapping etc.

It won‘t but I don‘t get why. The thesis is genius, it even starts with an philosophy and an attempt to formalize software architecture.

Erlang was invented at Ericsson, used for certain phone switches, it was tremendously successful at what it was built to do, and still it‘s not used anymore in their newer switches.

Why?

From the comment it seems there are people interested.

If the LLMs output is written in a way where I would edit at most 5% to be satisfied (which I did by the way), why shouldn’t I use it?