legulere

Google search ads are just highly optimised over now 26 years. There's no reason that OpenAI won't do the same. How successful they are going to be is not clear though.

My favourite example is amazon.com and pets.com. In hindsight amazon looks like an obviously good investment and pets.com doesn't, however amazon was heavily invested in pets.com itself.

Zero allocation is also a thing in rust. What you can learn is to reduce allocations in your critical paths.

Early-onset case numbers also rise in Europe though.

They are the biggest models supported in Github Copilot Pro

https://docs.github.com/en/copilot/reference/ai-models/suppo...

It's more like they realised they can also commute by bike or foot never realizing before that that was even an option

Microsoft tried with (Open)XPS and failed

Our visual perception has a higher bandwidth than listening. With text alone we can read faster than typical speaking speed and then a picture is worth a thousand words.

Darwin probably didn’t know this exact species but only other species of the same family and predicted that there are/were some members of the family that are carnivorous

With a heat pump asked central heating system run in cooling mode condensate can form at your pipes in the wall and your radiators or floor when using under floor heating. So you have to limit its cooling power to prevent that.

With an AC the advantage is that the condensate is forming inside the AC when cooling the air.

If your customers accept it, why not? It's much cheaper and testing in production is not really new.

Aren’t they also cheaper to built? Cheap backup capacity seems more important than efficiency and cost to run for Dunkelflaute scenarios to me

They could have explained that the fancy word Lochstab just means hole rod in German

Not everything has to be ARC though and arenas are strictly better than GCs lacking synchronization issues for allocation and not needing any scanning for deallocation.

That seems like a latency rather than a throughput issue. In compilers I would expect latency to be a lesser problem.

Let’s say you could get 10x speed up with 10k and 11x speed up with 100k. The 100k option still wouldn’t make sense.

It’s a contrived example, but to me it’s not improbable that a lot of the spend comes from tokens that are used very inefficiently. After all developers were pushed to try using LLMs without looking at costs.

I’m not saying the others are wrong, just that you can only tell something about marginal cost effectiveness, not total cost effectiveness of LLMs by measures to reduce AI spending

The baby boom happened in some countries at different times. In the US from 1946 till 1965, in Germany 1955-1969. In Japan fertility dropped after 1973

You often pay for regulatory compliance and liability with SaaS. Would you risk running your own vibe-coded payroll or accounting system for your company?

You could only conclude that if they would have dropped LLMs completely. It just means they see the benefit/cost optimum at a lower point than 100k/programmer.

I also doubt a longterm 2x multiplier for most developers.

There's https://camelcamelcamel.com to check how prices developed.

There's still the possibility of shenanigans with different models that can be overpriced for long so they look like a good deal when they get a normal price

PV making sense and PV+batteries being enough are two different things though. I have 10 kW on my roof as well here in Germany, in winter there’s not enough sun for heating with a heat pump though.

The weather is easy to measure though, which leads to a continuing improvement of the models. In a lot of fields it’s very difficult to make experiments and measure the results. In those fields models can stay very bad for a long time.

Because they don’t want to wait for a grid connection that can take years and they need backup generation capacity anyways.

You’ll see them all connecting to the grid in the following years simply because it’s cheaper

Only in some regions that’s enough. A lot of regions have winters where solar output is greatly diminished and wind can stop for days. Batteries are too expensive to store electricity for days to weeks. On top of that you have higher demand if you have electrified space heating.

Median automation engineer in Germany made 78 840 € in 2025, according to government statistics

https://web.arbeitsagentur.de/entgeltatlas/beruf/59351

How sure we are that something causes cancer and how often it causes cancer are not the same. Those WHO classifications are about the former. Usually the two are pretty much entwined, because you can more easily find larger effects. The effect of meats is a counterexample though: because it causes specific types of cancers that are otherwise rare you can be pretty sure it causes cancer even though the effect is not as big as e.g. smoking.

Uranium mining, milling, conversion, enrichment and fuel fabrication are all not free.

Because of protection from radioactive radiation, you have higher costs handling anything in a nuclear reactor compared to a coal plant. Then you have the issue of runaway nuclear reactions, hydrogen buildup etc.

To quote the article: "Don't confuse motion with progress."

We've been using probabilistic algorithms a long time now, also known as randomised algorithms. You probably won't find a program that is not using a hash table.

Compilers actually actively try to stay deterministic in their output.

Here in Germany there was a bidding frenzy for 3G frequencies. The end result was simply that 3G was more expensive than in other countries and reception was worse.