If your goal is to say, write science fiction, their reversion to classic LLM-isms, is really distracting and is what makes people say from a glance that it was written by an LLM. You basically can't use them at the moment in any real "natural" long-form writing. Everyone will call "slop" pretty quickly on the current frontier models.

Rosmin's DFT paper is worth a read.

While I'm skeptical that there is much of a moat, at least for the large players, it should at least hopefully set rosmine up with for the next job :)

It does seem to fix the current biggest issues with using LLMs for writing at various publishers. If you're The Economist, you have a very specific house style and you have a decent corpus of articles written in that style. At least on my reading of it, rosmine can use DFT to get a model to closely match its outputs, in terms of the language quirks that are generated, to that of the corpus it is fine tuned on. ie it will very much match the house style, particularly as it is used in writing, vs giving a system prompt to an LLM that has some Economist articles in its vast training set, and telling it to write in that style- it will do an ok job, but still exhibit LLM language quirks despite itself. Even if you feed it the specific "style guide" that they give their authors, I dare say the reality of their writing is the best place to learn, and it sounds like DFT can ground the writing of a model in a specific corpus like that.

[1]: https://rosmine.ai/2026/05/18/fixing-llm-writing-with-distri...

I have been to Luxembourg and to Hungary, Bulgaria & Greece - the otherwise obvious contenders for "poorest" in the EU and Luxembourg should not be in the picture.

It is a tax haven, with one of the highest GDP / person in the world, why is it, by magnitudes, the biggest recipient of EU largesse / person??!

They're super close to chimps (and definitely much closer than us), rather than "a very different species".

There are lots of reasons this stuff happens, but one of them is definitely that some kids aren't acting out for school reasons but for attention from their parents.

Now, like all good SciFi, it seems fairly prescient ....

However, we were strongly told that for early stage startups, some (CA) VCs would only bother looking at CA or DE companies.

I kind of get what you're saying, but let us not pretend that SW engineers are perfect rule followers either.

Having a framework to work within, whether you are an LLM or a human, can be helpful.

Super simple. (although I use rewrites at my dns layer for the whole local lan, but whatever)

It also solves issues my password manager has with multiple services on the one host but with different ports, but putting each on their own 2nd level domain.

Or at least went dark ..

I don't know if it is just that it is an Australian thing, but certainly my friend group would all just say "petrichor" for this scent.

The Australian's who coined the term specifically credited Indian perfumers for their matti ka attar; they had collected and distilled it for centuries before (western?) science investigated.

Like lots of scents, the fresh versus the preserved are different. Petrichor has a sharp ozone smell, that does not persist when preserved, where it ends up with an earthier smell afterwards.

Just give them the right writing prompt. "You are a writer for the Economist, you need to write in the house style, following the house style rules, writing for print, with no emoji .." etc etc.

The large models have already ingested plenty of New Yorker, NYT, The Times, FT, The Economist etc articles, you just need to get them away from their system prompt quirks.

certainly nothing friendly.

For others wondering, while most of the Franco-era laws were nuked in 1978, this does include lots of old laws (ie pre-20th C).

However, the source material starts with a sqashed commit in 1960 :) So no changelog before that. The BOE source though is pretty phenomonal, they've scanned files going back to the 1600s so far.

David has done some great work and some funny work. Sometimes both.

There goes another bit of my writing style that will get mistaken for an LLM.

The key difference, is that the US is many jurisdictions (Federal + 50 states + a lot of others, from counties to cities to territories to MANY others), and the variance amongst those is high.

The key thing well regulated places like Sweden get right, is that in consumer contracts you have minimum bars that you must meet regardless of what you can get the consumer to agree to. So, for instance, return policies, for goods bought online have minimum standards they must meet.

In the US, these things have huge variability. There are well regulated states, and well, the others.

Still, thanks for the process you use.

My wsl is pretty long lived now, through quite a few ubuntu upgrades and installations of stuff that I probably no longer need.

It seems like there are two options:

a) The "founder" of the code disappears in to the ether, and it is the equivalent of "version X only";

b) The "founder" stays involved, and if GPL 3 is updated, they can choose.

only b is worth speaking of. In b, isn't having someone in a position to make a choice much better than no one? What is the boogie monster that is the worry? The FSF puts out the 4.0 version, with a special "except for boramalper" clause, that lets you specifically monetise the hell out of it while keeping it closed source? I would not lose much sleep over that.

Stallman is a nutcase, in an endearing way (ok, maybe you have to have moved in the right circles). But he has put in place a system that needed just such a nutcase, who established clear black lines that could not be crossed, and who was also writing enough amazingly meaningful code that we needed to take his license seriously, that could then establish the institutions and governance to make it all live beyond him.