Gigi

Not doing DMs. Aspiring Saunameister.

It required years of practice, deep knowledge, and hard-won experience. Now anyone can be a software developer.

Source: mike.tech

It’s an interesting question. Software development used to be a craft. Something only a few people could do well.

Source: mike.tech

And for that metric, the winner is New York! Nope — it’s actually Vatican City, with 1.9% of people having a terminal. Almost 4x Luxembourg.

Source: mike.tech

The model is just one piece of the puzzle. Think of it like traditional software development: not every developer is exceptional, but a team of good enough developers with the right process can build great software. The same principle applies here.

Source: mike.tech

In parallel, theories of cultural evolution formalized the idea that socially transmitted information can exhibit variation, differential persistence, and inheritance, whether framed through gene-culture coevolution [10, 11, 12] or through the concept of memes [13, 14, 15]. Together, these traditions point towards a broader class of entities, the “viruses of the mind” [16, 17] whose evolution depends not on a particular material substrate, but on their capacity to reproduce and persist across biological, cognitive, or computational hosts.

Source: arxiv.org

It allows information to accumulate, be passed on, and recombine over generations, thus supporting cumulative culture [3, 4]. In this sense, languages can be viewed as population-level cultural systems with partially autonomous dynamics: they diversify, compete, hybridize, spread, and sometimes become extinct, showing patterns reminiscent of species embedded in ecological communities [5]. Their persistence depends largely on social learning, through which individuals acquire words, grammatical constructions, meanings, and communicative conventions by observing and interacting with others, especially during childhood [6].

languages can be viewed as population-level cultural systems with partially autonomous dynamics: they diversify, compete, hybridize, spread, and sometimes become extinct, showing patterns reminiscent of species embedded in ecological communities
Source: arxiv.org

We model transitions among uncoupled, coupled, and persistently dependent users, and show that the interplay between social transmission, recovery, and collective reinforcement can generate tipping points and technological lock-in. A central consequence is the possibility of runaway dynamics: once a critical threshold is crossed, small increases in adoption can trigger rapid population-level shifts toward persistent dependence, with abrupt losses in cognitive competence. The same framework, however, identifies conditions for cognitive immunization, based on reducing transmission and facilitating reversibility.

Source: arxiv.org

: the page edits and updates itself, so the CMS sits inside the static HTML rather than behind it.

You change the DOM. Click around the

the CMS sits inside the static HTML rather than behind it
Source: github.com

As with so many things, using LLMs to aid in programming requires judgement and balance: they can be extraordinarily useful across an entire spectrum of software activity, and shouldn’t be dismissed out of hand. But any implicit dependency on an LLM to comprehend or evolve a system must also be strenuously resisted: we do not want to become as gavage geese, being force-fed unnecessary complexity in hopes that an LLM will somehow spare us from asphyxiation.

requires judgement and balance
Source: rfd.shared.oxide.computer

The closer code is to the system that we ship, the greater care needs to be shown when using LLMs. Even with something that seems natural for LLM contribution (e.g., writing tests), one should still be careful: it’s easy for LLMs to spiral into nonsense on even simple tasks.

Source: rfd.shared.oxide.computer

LLMs are amazingly good at writing code — so much so that there is borderline mass hysteria about LLMs entirely eliminating software engineering as a craft. As with using an LLM to write prose, there is obvious peril here!

there is borderline mass hysteria about LLMs entirely eliminating software engineering as a craft
Source: rfd.shared.oxide.computer

For any writing where the expectation is clearly direct personal communication — e.g., chat, review comments, e-mail — LLMs should never be used to write. This is not to say that LLMs will not be used at all: they can be used to inform (in the case of review comments) or edit (e-mails) — but they should never be used to write text that the reader is very clearly expecting to be human-authored.

Source: rfd.shared.oxide.computer

This can be navigated, of course, but it is truly perilous: our writing is an important vessel for building trust — and that trust can be quickly eroded if we are not speaking with our own voice. For us at Oxide, there is a more mechanical reason to be jaundiced about using LLMs to write: because our hiring process very much selects for writers, we know that everyone at Oxide can write — and we have the luxury of demanding of ourselves the kind of writing that we know that we are all capable of.

our writing is an important vessel for building trust — and that trust can be quickly eroded if we are not speaking with our own voice
Source: rfd.shared.oxide.computer

Finally, LLM-generated prose undermines a social contract of sorts: absent LLMs, it is presumed that of the reader and the writer, it is the writer that has undertaken the greater intellectual exertion. (That is, it is more work to write than to read!) For the reader, this is important: should they struggle with an idea, they can reasonably assume that the writer themselves understands it — and it is the least a reader can do to labor to make sense of it.

LLM-generated prose undermines a social contract of sorts: absent LLMs, it is presumed that of the reader and the writer, it is the writer that has undertaken the greater intellectual exertion
Source: rfd.shared.oxide.computer

But there are deeper problems: LLM-generated writing undermines the authenticity of not just one’s writing but of the thinking behind it as well. If the prose is automatically generated, might the ideas be too? The reader can’t be sure — and increasingly, the hallmarks of LLM generation cause readers to turn off (or worse).

Source: rfd.shared.oxide.computer

First, to those who can recognize an LLM’s reveals (an expanding demographic!), it’s just embarrassing — it’s as if the writer is walking around with their intellectual fly open. But there are deeper problems: LLM-generated writing undermines the authenticity of not just one’s writing but of the thinking behind it as well. If the prose is automatically generated, might the ideas be too?

LLM-generated writing undermines the authenticity of not just one’s writing but of the thinking behind it as well
Source: rfd.shared.oxide.computer

LLMs can be useful researchers. LLMs are well-suited for the kind of light research tasks for which one would historically use an internet search engine, especially as LLM hallucination has been attenuated and sourcing has improved (and, frankly, as the quality of Google search results has declined).

the quality of Google search results has declined
Source: rfd.shared.oxide.computer

Note that this may be opt-out (that is, by default, a model may reserve the right to train on uploaded documents), but can generally be controlled via preferences — albeit occasionally via euphemism. (OpenAI shamelessly calls this checked-by-default setting "Improve the model for everyone", making anyone who doesn’t wish the model to train on their data feel as if they suffer from a kind of reactionary avarice.)

OpenAI shamelessly calls this checked-by-default setting "Improve the model for everyone", making anyone who doesn’t wish the model to train on their data feel as if they suffer from a kind of reactionary avarice.
Source: rfd.shared.oxide.computer
  • Urgency: Urgency seems natural with a tool that can seemingly do so much knowledge work so quickly, but with respect to LLM use, too many organizations have seemingly enshrined urgency over all else. These organizations treat LLMs as an opportunity to increase pace over all else, seemingly without regard for setting direction. Urgency is certainly important, and LLMs absolutely afford an opportunity to do work more quickly — but that pace must not come at the expense of our responsibility, rigor, empathy and teamwork.
These organizations treat LLMs as an opportunity to increase pace over all else, seemingly without regard for setting direction
Source: rfd.shared.oxide.computer
  • Urgency: Urgency seems natural with a tool that can seemingly do so much knowledge work so quickly, but with respect to LLM use, too many organizations have seemingly enshrined urgency over all else. These organizations treat LLMs as an opportunity to increase pace over all else, seemingly without regard for setting direction.
too many organizations have seemingly enshrined urgency over all else
Source: rfd.shared.oxide.computer
  • Teamwork: We are working together on a shared endeavor, and we must be sure that our LLM use does not undermine our sense of teamwork. Specifically, we must be careful to not use LLMs in such a way as to undermine the trust that we have in one another. In some cases, LLM usage is going to be warranted (or otherwise expected), and in others it won’t be.
we must be careful to not use LLMs in such a way as to undermine the trust that we have in one another
Source: rfd.shared.oxide.computer
  • Empathy: Be we readers or writers, there are humans on the other end of our language use. As we use LLMs, we must keep in mind our empathy for that human, be they the one who is consuming our writing, or the one who has written what we are reading.
Be we readers or writers, there are humans on the other end of our language use
Source: rfd.shared.oxide.computer

Oxide employees bear responsibility for the artifacts we create, whatever automation we might employ to create them. That is, human judgement remains firmly in the loop: even if or as an LLM is generating an artifact that we will use (writing, test cases, documentation, code, etc.), their output is the responsibility of the human using them.

Source: rfd.shared.oxide.computer

As in previous versions, all human-written training data is commercially licensed or owned by the company. As ever, Pangram 4 was not trained on any user-submitted data, customer data from API users, or data obtained from unauthorized internet crawls.

Source: www.pangram.com

Some research indicates that using language models to edit text alters its style in measurable ways; for example, academic papers rewritten by an LLM tend to accumulate hedging words ("may," "typically," "suggests") and emphasis words ("strong," "robust," "consistent"). Related work finds that LLM edits shift a text's meaning even when the model is instructed to correct only grammar, and that LLM-written peer reviews weight clarity and significance differently than human reviewers do. Elsewhere, there are indications that it is difficult to remove the signatures of an AI-generated draft even with extensive editing: writers who edit LLM output toward their personal style still produce text that reads closer to LLM writing than to their own.

Source: www.pangram.com

We believe it is important to distinguish this category of document from both fully AI-generated and fully human text. This matters to us both because we believe in transparent authorship, and because, as a popular and emerging technology, language models shape the world in ways we do not yet fully understand – but we do have some indication. Some research indicates that using language models to edit text alters its style in measurable ways; for example, academic papers rewritten by an LLM tend to accumulate hedging words ("may," "typically," "suggests") and emphasis words ("strong," "robust," "consistent").

Source: www.pangram.com

nerated or fully human-written; rather, we live in a time where a text's prose and its argument can plausibly have different provenance.

We believe it is important to disting

Source: www.pangram.com

AI only 0.0041% of the time, or roughly one false positive for every 24,000 documents.

  • Pangram 4 also has a marked reducti
Source: www.pangram.com

To be clear, I think LLMs are incredibly useful: they are helpful for brainstorming, invaluable for comprehending text, and they make for astonishingly good editors. (And, unlike most good editors, you can freely ignore their well-meaning suggestions without fear of igniting a civil war over the Oxford comma or whatever.) But LLMs are also lousy writers and (most importantly!) they are not you. At best, they will wrap your otherwise real content in constructs that cause people to skim or otherwise stop reading; at worst, they will cause people who see it for what it is to question your authenticity entirely.

LLMs are also lousy writers and (most importantly!) they are not you
Source: bcantrill.dtrace.org

When you use an LLM to author a post, you may think you are generating plausible writing, but you aren’t: to anyone who has seen even a modicum of LLM-generated content (a rapidly expanding demographic!), the LLM tells are impossible to ignore. Bluntly, your intellectual fly is open: lots of people notice — but no one is pointing it out.

the LLM tells are impossible to ignore
Source: bcantrill.dtrace.org

Because holy hell, the writing sucks. It’s not that it’s mediocre (though certainly that!), it’s that it is so stylistically grating, riddled with emojis and single-sentence paragraphs and "it’s not just…​ but also" constructions and (yes!) em-dashes that some of us use naturally — but most don’t (or shouldn’t).

Source: bcantrill.dtrace.org

So, writers beware: readers are in revolt. You should fully expect your writing to be run through Pangram. If your position is that we should be fine with an LLM crafting prose from your prompt, spare us all the wasted cycles and just give us your prompt.

If your position is that we should be fine with an LLM crafting prose from your prompt, spare us all the wasted cycles and just give us your prompt.
Source: bcantrill.dtrace.org

I had tried to use LLMs themselves for LLM identification, but I found that their false negative rate was far too high: they were chipper in accepting stuff that I was certain was LLM-authored. Other services seemed to look for basic LLM tells, but as an avid (and unapologetic!) user of the em-dash, these superficial techniques make me shift nervously in my seat (and I found them to be so broadly unreliable that they didn’t earn regular use).

as an avid (and unapologetic!) user of the em-dash
Source: bcantrill.dtrace.org

The war on spam started to turn when we could identify it at scale; could something similar happen to LLM-authored writing? Like spam, an LLM’s influence is readily identifiable to humans reading it; surely this is a solvable problem?

Like spam, an LLM’s influence is readily identifiable to humans reading it; surely this is a solvable problem?
Source: bcantrill.dtrace.org

Beyond having to endure aggravating stylistic tics, when reading a piece that has had substantial LLM assistance, we — the readers — don’t know what is real and what isn’t. As I wrote in RFD 576, to use an LLM to write is to void the social contract between writer and reader: we readers shouldn’t be expected to labor to understand a sentence that the writer themselves didn’t work to create.

to use an LLM to write is to void the social contract between writer and reader
Source: bcantrill.dtrace.org

And there are consequences that outlast the piece: 71% of the respondents in Cynthia’s survey also "avoid the author in the future" (!!). Revealingly, the respondents are not after linguistic perfection, but rather authenticity: 98% reported preferring an author’s own (imperfectly) written piece over an LLM-polished one. Finally, be wary of dismissing Cynthia’s respondents as a self-selecting group: active readers on social media are exactly the folks most likely to repost or otherwise promote writing they like — the early adopters and the tastemakers of online prose.

the respondents are not after linguistic perfection, but rather authenticity
Source: bcantrill.dtrace.org

71% of the respondents in Cynthia’s survey also "avoid the author in the future"

Source: bcantrill.dtrace.org

In fact, the tells of LLM writing are so grating ("and here’s why that framing matters!") that our brains pull an LLM-triggered ejection handle, bailing us out mid-sentence in an act of self-preservation. And the first person plural there is deliberate: as Cynthia Dunlop writes, of the 668 developers that replied to her survey, 78% "stop reading immediately" when they detect an LLM. And there are consequences that outlast the piece: 71% of the respondents in Cynthia’s survey also "avoid the author in the future" (!!).

78% "stop reading immediately" when they detect an LLM
Source: bcantrill.dtrace.org

As to the second question, readers emphatically care. In fact, the tells of LLM writing are so grating ("and here’s why that framing matters!") that our brains pull an LLM-triggered ejection handle, bailing us out mid-sentence in an act of self-preservation. And the first person plural there is deliberate: as Cynthia Dunlop writes, of the 668 developers that replied to her survey, 78% "stop reading immediately" when they detect an LLM.

our brains pull an LLM-triggered ejection handle, bailing us out mid-sentence in an act of self-preservation
Source: bcantrill.dtrace.org

Reading is important to me. While I’m not a quick reader nor an especially voracious one, I have found that long-form reading has had a profound influence on me over my life.

long-form reading has had a profound influence on me over my life
Source: bcantrill.dtrace.org

ing is important to me, writing is too: writing not only allows us to convey our ideas, but the very act forces us to test and distill them — at once making our ideas more robust

writing not only allows us to convey our ideas, but the very act forces us to test and distill them
Source: bcantrill.dtrace.org

To those who read broadly, the hand of the LLM is so clear it’s as if the writer’s intellectual fly is open. In fact, it’s so jarring that I have to believe that those writing with LLMs are either not reading enough to see the LLM’s obvious structural tells — or (and?) they aren’t even reading their own content. (A confession: with particularly egregious pieces, I have fantasized about sentencing the author to read them aloud, certain that they themselves will be unable to endure the slop that they are foisting upon the rest of us.)

those writing with LLMs are either not reading enough to see the LLM’s obvious structural tells — or (and?) they aren’t even reading their own content
Source: bcantrill.dtrace.org

unsere Plätze besetzen, aber immerhin: wir gehen mit dem besten moralischen Gewissen der Welt unter!

Es leben der Sozialismus!

Wir haben hier drüben schließlich den behördlichen Wohlfühl-Sozialismus perfektioniert – komplett mit Formular A38, moralischen Stuhlkreisen und einer EU-Verordnung für die korrekte Krümmung von KI-Algorithmen. Und da der Sozialismus historisch gesehen nun mal die zuverlässigste aller Abkürzungen in die wirtschaftliche Sackgasse ist, gleicht das aktuelle Schauspiel eher einer Wiederholungstat als einer Überraschung.

Wenn die deutsche Autoindustrie heute unter Druck gerät, offenbart das lediglich die Schwächen des gesamten Wirtschaftsstandortes. Historisch gesehen war der technologische Wandel immer gnadenlos: Kodak verpasste die Digitalkamera, Nokia das Smartphone. Dass nun deutsche Hersteller im Bereich Software und künstliche Intelligenz hinterherhinken, zeigt, dass das Land in der Breite den Anschluss an die digitale Wertschöpfung verliert.

Wer von der Pferdekutsche zum Verbrennungsmotor wechseln konnte, sollte theoretisch auch den Sprung zum autonomen Elektrofahrzeug schaffen. Das eigentliche Problem Deutschlands liegt nicht in der Existenz dieses Wandels, sondern in einer strukturellen und kulturellen Innovationslosigkeit, die das Land zunehmend daran hindert, neue Märkte aktiv zu prägen.

Der traditionelle USP deutscher Marken löst sich im autonomen Zeitalter teilweise auf. Wenn der Fahrgast auf der Rückbank sitzt und E-Mails liest, spielen Fahrwerksabstimmung, Motorleistung und das Design des Lenkrads keine Rolle mehr. Stattdessen rücken Software, Infotainment und nahtlose digitale Integration in den Vordergrund – Bereiche, in denen deutsche Hersteller traditionell schwerer Fuß fassen als Tech-Giganten aus dem Silicon Valley.

I choose to build on the Nostr and Bitcoin ecosystem because nothing else I’ve found has ever given me this level of ownership, and personal fulfillment, and I couldn’t imagine being anywhere else. Even if almost nobody gets it yet.

This is what I love about open source software, and it’s what keeps drawing me in deeper. I don’t have a strong background in handwritten code, but I have a career of solving complex problems in creative ways, and as any of you who know me well will probably agree, I am fucking relentless when I set my mind on accomplishing something.

FIPS is great, and it just works. The difficult part is explaining it to people.