deepsummer

I never had any issue like that, but I don’t use it in German. Not surprised it’s more stubborn on German. After all it’s only imitating people speaking that language. I wonder what the topic was though. Was it a legitimate request, or just some discussion?

An RSS (Atom...) feed for your blog would be nice.

I get your point, but the RSS format (and Atom, and JSON feed..) and the feeds are also partially to blame: there is just no common HTML subset specified for RSS. Some feeds are formatting their posts specifically for RSS. But they are in the minority. Most are just dumping whatever HTML they put on the website. With CMS-specific CSS classes and Tailwind CSS, without declaring them anywhere. Or even requiring JavaScript. RSS Readers have no chance but remove most styles from the HTML and apply their own, hoping it renders ok.

I think it's an interesting idea, and I would actually buy it, but there's one thing missing for me: a simple API. I use neither Stripe nor GA. All I would need is a simple REST URL that I can ping with the event type, data (amount or number) and my key. Create a small markdown description that my coding agent can read, and I could get it running far faster than by using any integration with third-party services.

Three more comments:

- My main concern with applications like that is memory and CPU consumption. I haven't registered it, but after starting, it consumes 85MB. That's tolerable these days, but more than other menu bar apps I am using, and they have far more functionality. I appreciate small applications being a small footprint.

- Maybe I am not the target group, but I really prefer the more 'family-friendly' sounds. I don't want to explain to my kids why my computer is moaning.

- There may be a market for notification tools that offer a bit more flexibility (like letting users configure the text that's displayed). Or even offering a widget. I am aware of free tools like Übersicht for that, but I don't want to waste 1 GB of RAM for a small 64x64 widget, just because it uses a full HTML+JS rendering engine.

I think the UBI experiments were one-sided. They may have proven that the receivers have profited. But did they show that the givers profited as well, or at least did not suffer enough to significantly reduce their economic output and therefore risk the financing of UBI?

You'd need to check something like take a group of high-earners, increase their tax rate by say 100%, increase their cost of living, and validate that they neither leave the experiment nor work significantly less.

The thing is, there is no hive mind called "society". Everybody works in their own interest. Always, and in every "society".

Individual workers work on failing initiatives as long as they get paid. The outcome of the initiative may be bad for "society". But was it good for the individuals working on it? Maybe they got paid well. Maybe they enjoyed the work? Maybe the work was easy because they knew that it would fail anyway, so they didn't have to put much effort into it?

Maybe it was also good for the management or politicians? Maybe it was a step up in their career. And maybe, if they could jump ship before the failure became obvious, they could climb up the ladder to get to an even better position? You can always blame your successors for ruining the project.

And maybe it was good for whoever ordered it? If it's a local project, maybe they got subsidies from federal government bodies, and they don't even care whether it succeeds, as long as it created employment and the illusion of progress? Or if it's a private project, maybe they just tried a moon shot that, if it fails, was useful as a tax write-off?

In real life, there are so many layers to a 'failed' project. It can be a failure for some and a success for others. And those for whom it is a success will defend it, maybe even deceive to keep it running.

The prosecutor's description of the "ICE protest" is "setting off fireworks, vandalizing property, and shooting at police officers who responded. One officer was struck in the neck with a bullet and survived.".

https://www.cbsnews.com/texas/news/ice-detention-attack-defe...

"Prosecutors said that the group launched a premeditated terror attack on the detention facility inspired by antifa ideology, by setting off fireworks, vandalizing property, and shooting at police officers who responded. One officer was struck in the neck with a bullet and survived."

https://www.cbsnews.com/texas/news/ice-detention-attack-defe...

1992-me hates the author. Coming from 68k assembly, x86 was a nightmare. And together with the ridiculous number of registers, segments made up a huge chunk of that horrible experience.

If he wanted to get out and sell his stocks, why would he care if it collapses after he leaves?

"Selling the stock" is not a single transaction. The moment he leaves Tesla, or starts selling shares (which would take weeks or months), the stock collapses.

A Tesla without Musk isn't worth much. It's not that profitable. Tesla's P/E ratio is 350 or so - absolutely crazy. Ford's and Toyota's are about 10. The largest Chinese electric car maker's (BYD) PE/ratio is 20. People are buying Tesla shares because they believe in Musk. The moment he announces that he is out, his wealth is going down by 90% or more.

"Elon is currently sitting on 817,000 lifetimes worth of money."

I think that's a wrong perception, especially in this case. Elon Musk does not have nearly a trillion dollar in cash. He holds shares in companies that have a market worth of a trillion dollar. But as soon as he sells, and thus wouldn't be involved with them anymore, they would be only worth a fraction of that. Tesla, in particular, would probably be bankrupt.

I think it is too early to answer that question. If I really wanted to, I could 'vibe code' Photoshop. But my estimate is that it would take me 2 years, as a solo dev. And by 'vibe code' I mean not write a single line by hand - I would still define and specify an outline for important data structures and algorithms. And then work through the system feature by feature, with many prompts needed for each feature, until it is perfect. It is much faster than coding by hand, but still a lot of work.

Currently, I am working on a smaller-than-Photoshop solo side-project that would have taken years, but now I am close to a beta after 6 months. If the main work is coding, I am easily 10x as productive. But those productivity gains don't hold up when teams are larger, because communication and processes are not really accelerated. So be a bit more patient. It's too early to expect vibecoded Photoshops.

APIs - MCP is the obvious better alternative, but as you said, most software and most online services don't provide a full API. Most none at all.

I watched it last week, and I loved every moment of it. It's really the Muppet Show, not one of those boring remakes and spin-offs that never worked. Even my 9 year old daughter who would usually roll her eyes when I would even suggest watching Muppets, watched it till the end. Probably mostly because of Sabrina Carpenter, but whatever.

Would love to get more episodes.

I can understand why you want an AI to use a desktop. But it's still absurd to use the least efficient interface possible for interactions with the outer world.

Having said that, fun project, good luck :) I am sure quite a few people would want to try it.

As much as I like the Claude models, they are expensive. I wouldn't use them to process large volumes of data. Gemini 2.5 Flash-Lite is $0.10 per million tokens. Grok 4.1 Fast is really good and only $0.20. They will work just as well for most simple tasks.

I understand your argument. But I have worked at two companies that worked pretty much like you described. They call it 'project-oriented'. They threw lots of engineers at projects, hired freelancers, and got it working as fast as possible. Once it was done, they only left a maintainer or two.

That model works fine for a few years. Then you need a bigger change. Often, the system is built on top of some enterprise project, heavily customized, and you stay at your outdated version until it becomes unsupported. The maintainers don't care, and often don't have the capability to upgrade, so they just leave it as long as it keeps working. Or maybe some law has been introduced and requires a bigger change. Or the market just changes, and you need to support new APIs, new payment methods, new integrations...

The maintainers tend to quit every 1-2 years and are replaced with someone only trained by the previous generation. With every generation, the maintainers get worse. After 3 generations, all product knowledge is gone. To make things worse, the maintainers do stupid things in the code because they don't fully understand it, and it begins to rot. In the worst case I know, no one even knew what branch was deployed on production and what the last changes were.

Then, after 5-10 years of decay, some requirement comes along that would require a major refactoring. Everybody is overwhelmed, no one understands the internals, and eventually they decide that it the project is now so outdated that the only solution is to replace it. Management doesn't care because they can blame their predecessors.

In my experience, that's how it always works. I know at least 5 major projects that took over a year to develop, and costing millions, in at least one case tens of millions, that died like that.

Great idea. Whether brainstorm mode is actually useful is hard to say without trying it out, but it sounds like an interesting approach. Maybe it would be a good idea to try running a SWE benchmark with it.

Personally, I wouldn't use the personas. Some people like to try out different modes and slash commands and whatnot - but I am quite happy using the defaults and would rather (let it) write more code than tinker with settings or personas.

I understand it. For example, with AI you don't need to remember stuff. Like there is a command in MacOS (two actually) to flush the DNS cache. I used to memorize it because I needed it like twice a week. These days, I can't remember it. I just tell Copilot to flush the cache for me. It knows what to do.

And it's like that for many things. Complicated Git commands that I rarely need. I used to remember them at least 50% of the time, and if not, I looked them up. Now I just describe what I need to Copilot. But also APIs that I don't need daily. All that stuff that I used to know is gone, because I don't need to look it up anymore, I just tell Copilot or Claude what to do.

I recently had the same realization and moved all my functions to a simple stand-alone server. Besides the normal AWS costs, what scares me most about AWS is the possibility that someone could try to DOS me, leaving me with a huge AWS bill, because there is no real way to limit AWS spending.

The main reason why I keep coming back to cloud providers is databases. I don't feel comfortable setting up a high-availability db setup, and I don't want the responsibility of managing backups. But if you go to, say, Hetzner, you won't be able to use a cloud database in the same network.

Let me ask you a related question: if there was a study that handwriting is better for your brain than typing, should secretaries have quitted when typewriters and computers were introduced?

The thing is, there is no going back. There will be no significant demand for output that's created by humans even though a machine can do it as well. You can try to find a niche where AI is worse than humans. But that will be increasingly difficult to find.

So if you want to continue doing things without AI, that's fine. But most likely it will be a hobby, not a job.

It can certainly also be used the other way round by people who defend Trump no matter what. But I have seen enough people who clearly weren't even able to discuss Trump's policy because the thought that Trump could be right about anything was unacceptable to them. And often that thought caused a very emotional reaction.

As Scott Adams would tell you, it doesn't matter whether he made it up. If you believe him, that's your reality. If you don't believe him, your reality is that he's faking it. You can chose your reality and that act upon it.

A lot of comments here mention his comics or his controversial pro-Trump opinions in the last 10 years, but I would like to emphasize and point out his influence he had over the lifes of so many people with his life strategies and explanations, microlessons, memes and ways to look at the world. Like

* systems over goals: the theory that you shouldn't set yourself specific goals, but instead just find a system how to work towards your goals regularly

* talent stacks: the theory that, in order to succeed in life, you don't need to be the best in one skill, but good enough in a useful combination of several skills that can be used together

* the idea that managing your energy is more important than managing your time

* the Adams rule of slow moving disasters: any kind of disaster that takes many years to manifest can be overcome by humanity. Scary are those disasters that don't give you enough time to react.

* rewiring your brain: that by finding the right way to look at something, you can modify your own behavior. He wrote a whole book full of recipes to change your behavior and feelings.

* despite not listening to Rap, a long time ago when Kanye West had one of his first successful songs, someone sent Adams the lyrics to some song and by looking at the lyrics Adams recognized West as a unique genius

* you should never trust a video as proof of anything, if you can't see what happened before or after. It's most likely taken out of context. Just like most quotes are worthless without context.

* "perception is reality": that how someone perceives a fact is more important than what actually happened

* "simultaneous realities": realities are shaped by how people perceive them. And two people can disagree on something, while both are right at the same time, because they view the same thing through two different lenses and thus live in different realities.

* TDS (Trump Derangement Syndrome): the observation that many people hate Trump so much that they lose the capability of rational thought and either just shut their brain down when talking about anything related to Trump, or want to do the opposite of what Trump wants

* "word-thinking": when someone find labels for things or people, and then forms opinions based on the label

* detecting cognitive dissonance: when someone just shuts down their brain because the experienced reality doesn't match their expectation

* "tells for lies", like analyzing people on TV and looking for clues that they lie

* coining the term "fine people hoax" for a video snippet that was constantly repeated on media to show Trump having one opinion, even though when watching the whole video it was clear that he meant the opposite.

* "logic doesn't win arguments", the rules of persuasion, and the theory of 'master persuaders'

* he predicted Trump winning the 2016 election when Trump had just announced his campaign, long before the primaries, because he recognized a 'master persuader' in him.

And there are probably many more things I don't remember right now, but his books and blog shaped my way of thinking, and I am using his way of looking at the world every day.

I must admit I didn't really follow 'Coffee with Scott Adams' - I think he kind of jumped the shark when having to fill at least 30 minutes every day, and I am not that interested in politics. But that doesn't diminish his accomplishments.

I think "number of questions asked" is the wrong metric. Because it feels like all the questions have already been asked. Whenever I need to know something, I can google it and find answers on Stack Overflow. I can't remember the last time I actually had to ask something. Or the last time I found a question that didn't already have a good answer. Stack Overflow's library of question is pretty complete, and the only reason for new questions are new tools.

Certainly LLMs are a huge factor, but I feel that LLMs rarely give good (and trustworthy!) answers to the things I would check on Stackoverflow. Just like LLMs are no good replacement for API references because they get the details wrong all the time.

I think a couple of years ago there was a real window of opportunity for crypto. People could have invented crypto versions of real-life things like insurances and mortgages.

But that never really happened, because everybody was busy getting rich with shitcoins and ape NFTs. Which seemed easier than building up complex organization on a crypto foundation. Along the way they destroyed all trust and goodwill that Bitcoin had created. Maybe, in a few years, there will be another opportunity, but right now it will be hard to convince anyone to invest money in a crypto business.

An AI that's breaking copyright laws shouldn't be legal. So yes, it's kind of like putting it in jail.

I think an AI should be treated like a human. A human can consume copyright material (possibly after paying for it), but not reproduce it. I don't see any reason why the same can't be true for an AI.

Some work will be replaced with AI. If you regulate the AI, the work will also be relocated.