mhh__

People turning up in hedge funds (i.e. much smaller) and trying to rewrite the bit of a bank they used to work in's equivalent of this article is so annoying.

They look for latent structure in the data.

you will wet yourself in an M&S

asda

Along these lines always watch the flow of entrepreneurial Asians.

Lamport phrases this as "Thinking doesn't guarantee we will be right; not thinking guarantees we will be wrong" - specifications are for reasoning about systems in the deepest sense

AISI is basically the crown jewel of the British government at this point in that its actually pretty good.

So I think one of the main failure modes of vibe coding is that unless you have a very aggressive approach the onus is pretty much solely on the developer for the code to be good.

The volume of code, addiction to said volume of code, and fact that the vibe coder may not have read it basically makes review impossible both logistically and in that IME it seems to upset the vibe coder to even suggest that it's fine to take a bit longer and do something good as opposed to some overfit mess.

It might be that we look back on this as like trying to review the assembly output of a compiler but I don't see it that way at the moment.

I increasingly agree with the spirit of this. I'm not even convinced that they'll never be able to "program" I just find that these tools are to midwits what flame is to a moth and destroy the systems-thinking that drives good engineering (as opposed to merely coding)

You have to be good enough to know what to ask for but not good enough to ask yourself what is needed next, or to say no to it - for example, you can fairly easily ask an agent to write small changes and so on, almost no one does.

Management are also seem the most prone to psychosis (IMO at least) e.g. if you are a once-technical manager it probably feels great to kind of being doing programming again with a minion who doesn't eat lunch?

I am yet to meet more than probably 2 vibe coders who uses their newfound "productivity" to try _new_ things as opposed to just pump out correct-looking slop - the problem here isn't just that there is now more code, but also that for some strange reason people seem to choose to automate the thinking as well as the coding.

I am yet to see any AI driven productivity around me. They can definitely turn a days work into a lot less time, this is magical, but I'm yet to see any "real" projects be delivered in a timeframe that's particularly different to what they would've taken previously.

"Because it's there"

One canary I have for Linux vs Windows is whether Bloomberg ever support Linux natively on the terminal. You can actually use the API from inside WSL quite easily already.

If it was the most expensive it wouldn't transact at that price, the price clears roughly to minimise excess demand i.e. gas is often the energy you can actually at a given time.

So what if it's a cold night with no wind? I'm not saying renewables aren't cheap, I'm saying that you can't fully construct a grid on them and that this has not been communicated.

Our energy is expensive and volatile because of this (and subsidies to implement it). You don't seem to be capable of following my argument. It's not a free lunch. Renewables are also slightly problematic for grid stability although I expect we will have enough fast storage / (possibly synthetic) inertia eventually.

"What if we don't fix the roof?"

"but it's sunny?"

> logic and experience

This is patronising drivel.

We burn gas when there is no wind & sun. We do this because there no alternative other than the lights going out. You are not seeing the whole.

We have already done that. Do you know why gas often sets the price?

Can't imagine that slop is going to save them. Gitlab is a totally directionless, beyond self-hosting which I think is commendable, shoddily implemented product. I don't hate it, in that it is at least predictable, but the lack of basically any interesting view on how software should be developed or even look is such a waste.

> experience

We have some of the most expensive energy in the world, what gives? I was told renewables were cheap

Would you rather be biased or blind?

Do you have any substantial counterargument beyond "you aren't allowed to think that"?

Recent piece in critic magazine by Chris bayliss for example.

I've basically only just seen this stuff started to be discussed critically in the media in fairly recent years.

not including $work discussions with energy traders.

If the grid balance is dominated by bursty renewables then you can potentially price the stable / on-demand generation out of the market (or lead to a massive contango to incentivise said producers)

Unfortunately in Britain at least politicians are absolutely dead set on taking the piss / abusing this by e.g. adding huge amounts of subsidy and stealth taxes into what should be price discovery mechanisms (or for example when was the last time you heard someone talking about how cheap renewables are and discuss the CfD schemes).

normally when you buy electricity it costs money!

You have to think about these things as a portfolio rather than just by minimum price.

If you have a steel mill for example you need to be able to basically guarantee a certain level of energy production to run it viably because the risk of there not being any power during adverse weather is enough to make it unviable (you can't just turn these things off). This is the reason why gas and nuclear probably aren't going away (or at least shouldn't).

This is a lesson in how electricity isn't really a commodity e.g. it's very very difficult to send some electrons from one side of the world to another.

"rareleds" on Instagram is fantastic. Vintage LEDs set to apex twin and so on

This is more of a question of Affordances than type systems as per se e.g. you can do this quite happily in C# or something it's just that the amount of visual clutter is more than the actual type definition.

So I heard someone recently in person say "Oh you can just have the AI do things that don't really matter like database transaction"

It's so sad that given these amazing tools the average programmers attitude is to automate the things that should be their edge as an engineer.

Torvalds said that great programmers think about data structures. Midwits let the AI handle it.

Another one for my theory that web search makes LLMs useless for anything other than searching the web.

I don't know if that's true. Linux users are curious and will try more stuff but people mistake that they file bug reports (and usually detailed ones...) with complaining more.

Apple's MO is that it's their baby. End of. They don't do open. Their compiler is closed source, and so on.