Login
You're viewing the f.duriansoftware.com public feed.
  • Oct 4, 2026, 2:47 AM

    I don’t know what cheesy LLM these scammers are using, but the glitches can be hilarious.

    [Though I have written a novel called “Morphotrophic”, “Morphological Regency” sounds like a token midway between that and the average of all titles on Smashwords.]

    Hello Greg,

A strong book deserves more than simply being available for readers to find, particularly when its ideas invite readers to spend time thinking, questioning, and discussing what they have read. Your work in hard science fiction, mathematical speculation, and quantum theory creates exactly that kind of reading experience, where the scientific concepts and larger questions can become part of an active reader conversation. I’m Justin Brierley, a member of Authors Tables, where I partner with the group’s moderator to bring selected books into our Reading Challenge Cycles.

I believe Morphological Regency could create a particularly engaging discussion among readers interested in hard science fiction and intellectually demanding speculative fiction. The Reading Challenge Cycle gives readers an opportunity to discover the book, read it together, discuss its themes and ideas, and engage with the author through the reading experience. The focus is not on manufactured praise or predetermined reviews, but on putting the book in front of readers who are willing to engage with its scientific and speculative ideas thoughtfully.

I’d like to discuss the possibility of featuring Morphological Regency in an upcoming Reading Challenge Cycle and showing you how the cycle could give the book a stronger reader-facing opportunity. Would you be open to seeing how it works?
    💬 4🔄 0⭐ 0

Replies

  • Oct 4, 2026, 2:51 AM

    @gregeganSF This is interesting because to my understanding of how "transformer models" work, you'd expect them to sometimes get confused and meander in a different direction mid-word, because I was told stems and suffixes of words get split into separate tokens. And I've been going, but I've never seen that happen in the wild, so I must misunderstand something. But I *think* you can see it happening in your email there.

    💬 4🔄 0⭐ 0
  • Oct 4, 2026, 2:55 AM

    @gregeganSF Anyway, I think "Morphological Regency", if it existed, would probably be written by… Susanna Clarke, maybe? It sorta almost works as a not-as-good alternate title for Piranesi…

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Oct 4, 2026, 6:42 AM

    @mcc @gregeganSF each output is a probability distribution for all tokens. the concept of temperature scales & saturates the probability. then the sampler picks from tokens in e.g. the top 95% percentile (min_p), then limits that set to e.g. the 20 most likely (min_k), and then does a weighted random pick from that set.

    for ambiguous junctions, the distribution is flat rather than S-curve shaped, and odd choices are more likely.

    when temperature is high, oddness is very likely.

    💬 0🔄 0⭐ 1
  • Oct 4, 2026, 6:57 AM

    @mcc @gregeganSF I guess common large LLMs are too large, and the prior from the word beginning is too strong for it to happen with any regularly, but I have seen Frankenwords and misspellings occasionally when using smaller, local models. The spammers probably use the smallest model they can (mostly) get away with.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Oct 4, 2026, 5:53 AM

    @gregeganSF This post prompted me to learn new words (because I thought of a word I didn't know how to say in English).
    Interesting that in this text the only signs of LLM I see are abstruse language and lack of variety among the sentences lengths, yet it makes such a stupid mistake.
    Sadly taking 3 paragraphs and still not saying what exactly the want, is typical for any marketing emails, so that's not exactly a sign.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0