legulere

I really like the concept of novelty budget: keep most parts boring and use new stuff only where you can gain an advantage from it.

in just under one hour.

I wonder what the survivorship bias is though. How many other problems did they try but fail? Did they try to solve this problem but with another prompt? Still very impressive though.

You need to write some unsafe code in a kernel, but most of the code does not have to be, which allows you to eliminate memory unsafety from almost all code and give more scrutiny to the parts the compiler cannot guarantee for you. I don’t know though if the affected code in openbsd would have needed to be unsafe. Moving towards rust is a possible way for kernels as Linux has shown, but I guess for OpenBSD the pros and cons are different, as it’s striving for a more minimal system and has been affected far less by memory unsafety issues.

I guess the problem is with the price of the sensors. Just look how expensive the Aranet 4 home shown in article is. There are worse devices for less like the IKEA alpstuga. I also don’t know how much electricity they pull.

You can build gas power plants on site at your data center and don’t have to wait for the grid to be upgraded. In the long run gas will be replaced by cheaper renewables

Housing suffers from being based on a naturally scarce resource land. Markets drive prices higher but new land cannot be created

Part of Elon Musks strategy seems to sell some kind of hype that does not materialise or at least won't for long (Mars, autonomous Cars) The vast amounts of money collected are then used to develop products that are still a significant progress in its market. Now AI is where all the hype is. It's difficult to sell some hype without AI currently.

It generates tokens by estimating what the next token is going to be.

Sure it cannot think like a human, but given it's input, it should give a good statistical answer (approximating not of how long it actually takes, but what a human would say how long it takes).

The Chinese Government isn't like Russia where power is mostly legitimised by having power, like the law of the jungle. A big part of its legitimacy is from economic growth and lifting people out of poverty.

The examples in Clean Code show that Uncle Bob himself misunderstood heuristics as strict rules that are to be taken to the extreme. There's nothing to not misunderstand.

With fixed you do pinning on GC memory, which can have a negative performance effect. You can also do unchecked pointer arithmetics on references with the Unsafe class, which avoids that. A lot of the methods of Span use that internally.

I don’t think it’s different politics directly in China. The people believe that change means change for the better. In the west people have lost all hope for progress.

I let Claude translate a horribly written vb program writing some xml data into a pdf form. Most of the code I didn't even read until much later, I just checked the end result. The code won't be touched again, and if it will simply be replaced. Some code is foundational and you should put a lot of effort into it, a lot of code isn't though.

Other than that agentic coding has not really been working that well for me at our main codebase though.

Even earlier in Germany: https://en.wikipedia.org/wiki/East_German_uprising_of_1953

Here in Germany they very often talk about the importance of what happened in Poland for the reunification

The societal advantage of raising prices with demand is that it will lead to more supply generally, but in some cases supply is just simply limited and cannot be expanded enough sensibly. This leads just to people earning money without any additional economic benefit and often the poor suffering because of it (like with oil currently).

Of course DRAM manufacturers are taking advantage of the current situation, they're companies and making money is what companies are for. The problem is that DRAM manufacturing works in boom-bust cycles, building a new factory is capital intensive and slow, so additional supply will come too few and too late to press prices down to a sensible rate above manufacturing costs.

France still gets around a quarter of its uranium from russia and partnerships between Framatome and Rosatom are still going on: https://www.politico.eu/article/france-russian-state-owned-n...

How about the VVER-440 still running e.g. in the Czech Republic and Slovakia that lack proper containment? Still only Russia being Russia?

Their water use is mostly for cooling needs from the heat produced from their electricity use.

You should also include the water needed to produce the electricity, which is the biggest water user in the US:

The three largest water-use categories were irrigation (118 Bgal/day), thermoelectric power (133 Bgal/day), and public supply (39 Bgal/day), cumulatively accounting for 90 percent of the national total.

https://www.usgs.gov/faqs/how-much-water-used-people-united-...

Chernobyl lacked a containment and there are still reactors of the same type running without containment in Russia.

Europe managed to get off Russian Gas, but didn’t manage to get off Russian uranium industry. You correctly identified the chokepoints and Russia can’t be replaced fast there.

Don't forget the organisational overhead. You'll need managers and communication overhead between developers grows superlinear (see Brook's law).

It's not the 20x efficiency of harvesting technology compared to what agrarian societies that make them make sense. It's the productivity of the other 95% of the population that makes their labor cost so high that such expensive machines make economic sense.

To me it reads more like they are struggling to scale with requests and are trying to find ways that hurt users the least.

Like this? https://www.thinkgeoenergy.com/vulcan-energy-achieves-drilli...

Economic viability depends on many things, lithium prices have been pretty volatile in the past, battery production in Europe as customers are just scaling up.

It’s not 100% of oil production capacity that is lost, but 20%. You need to cut demand by that, so electric cars can help extremely, because most oil is consumed during car use, not production.

This isn't any different than the "person who wrote it already doesn't work here any more".

It is very different. With empathy you can often deduct why people wrote code the way they did. With LLMs there often is no reason.

I think you cannot take the step from any turing machine being representable as a neural network to say anything about the prowess of learned neural networks instead of specifically crafted ones.

I think a good example are calculations or counting letters: it's trivial to write turing machines doing that correctly, so you could create neural networks, that do just that. From LLM we know that they are bad at those tasks.

Much simpler: just store session ids in Redis.

I skimmed over the previous articles in this blog and they don't seem to mention the one use case JWTs were made for: having a separate authentication server from the application server. Most developers will only need this for integrating into corporations with single sign in or social logins (sign in with Facebook/google/apple...). There you won't write the authentication server but integrate with them. Session Ids are dead simple to get right securely. Just use them.

That's the idea behind it. The reality is that patents are written in a way to reveal as few as possible while blocking other companies as much as possible.

There's another perspective you can see in the comparison with the dot com boom. The web is here to stay, but a lot of ideas from the beginning didn't work out and a lot of companies turned bankrupt.