• Member Since 21st May, 2012
  • offline last seen January 28th


Call me Gryph for short, if you like. In-case the avatar, the name, and the themes of my stories didn't make it apparent; big Gryphon fan here.

Blog Posts

  • 28 weeks
    What Comes Next?

    Whew! A half-million word novel is no joke folks... Especially when it isn't even your first extremely-long-form-got-away-from-me-Gryphons-rarely-do-anything-small escapade.

    They say no plan ever fully survives first contact with the real world, but to fail to plan is to plan to fail. Plans are clay to work with as the situation evolves.

    So here is mine, currently:

    First, I again want to shout-out Keystone Gray, and The Campaigner.

    Read More

    1 comments · 237 views
  • 28 weeks
    Advocate: Cutting Room Floor Achievements

    Here, in no particularly organized fashion, are all the achievements (the vast majority by Generic Friendship) which sadly didn't make it into The Advocate for story/brevity/etc reasons.

    • Out of the Cold, Around the Fire - Go camping with a party of individuals. - “All right, all set. I’ll go set up the bear traps…”
    • Your Plus One - Have Roger recognize Malacandra as a member of your party. - “I could write the book on being the third wheel.”

    Read More

    0 comments · 159 views
  • 47 weeks
    The Advocate: Progress Update

    Just to relieve anyone who still is keeping an eye out; I am working hard on The Advocate!

    I have been trying to get away from the practice of tossing out a few chapters, vanishing for a month, then repeat. So I am presently building up a backlog.

    The good news is, that backlog is almost long enough that I'm ready to start releasing again, on a set schedule, right through to the end of the book!

    Read More

    5 comments · 238 views
  • 67 weeks
    Just a Quick Update

    For the interested: I'm still focused on getting The Advocate completed.

    I had planned to make some progress this week, and still might depending how the weekend falls out, but it is becoming clear it won't be as much progress as I'd hoped, if any.

    Read More

    5 comments · 286 views
  • 106 weeks
    State of Stories

    Yes - I am still writing daily on RO, and hoping to finish this year.

    With that out of the way; It's been a busy few months. Lots of good projects at work, and lots of family issues, medical and more positive ( a couple killer DnD sessions, among other things).

    As a result, It's been radio silence for a bit.

    I consider that a nice recharge break, now I'm back to finishing RO... With a small twist.

    Read More

    1 comments · 246 views

Latest Stories

Comments ( 57 )
  • Viewing 53 - 57 of 57

It seems another viable pathway to AGI just dropped. ¡Bad news! that it poses an S-risk specifically to folks like us, in precisely the way you spent 469K words battling. I quote the relevant parts below, emphases mine.

[…] one thing that might help is various forms of "human detector" hardware. By which I do not mean a webcam, but domain-specific hardware that just exists for interacting with humans and is nontrivial to fool. A cuddle machine perhaps.

It's not that this can't be fooled, the purpose is to encode a ground signal for human presence that is not *accidentally fooled*. Deep effort should have to go into fooling it, by which point your model has learned instrumental values over the latents it gained from the sensor.

Basically think of it as a way to narrow the hypothesis space so that the conditioning on past experience/history you train it on earlier is more likely to work. These grounding modalities are ways to eliminate parts of the wrong solution space.

For example if you had encoders for modalities like a cuddle machine, EEG headset the user wears while talking to the model, these encode tons of bits about the human form for the model to form instrumental values over the latent geometries of the encoders.

Then the point is that once the model is consequentialist and powerful enough to simply fool these hardware based terminal values, it is not a generic consequentialist choosing the most efficient solution to them, but choosing a solution conditional on its previous choices.

The major failure mode is that any aspect of the human form you encode like this becomes integral to the models values, which limits our future morphological freedom/potential evolution. But you're going to have this problem with any form of attachment the model has to humans.

After all if humans can change *totally arbitrarily* then humans are almost literally undefined. So to keep your AI attached to you, you will presumably be bound to some definition of human/sapient that it recognizes and optimizes over worldstates for.

I am sorry that I'm only just now seeing this! I have long stints offline from scheduling related reasons, so sometimes I just plain miss things.

I really deeply appreciate the kind words; Have no fear! My plan is to finish Advocate 'soon' (in relative terms) then finish the last Hegira book, then likely write some more FiO stories following on from Advocate.

Hi! I just wanted to say what a joy it is to read the current story you are working on (The Advocate.)

I feel I am in a really strange position as far as the fandom for MLP goes. I've only watched one episode. I don't think I have a desire to watch any more, but I have come to know and love a lot of the characters through the lenses that other authors on this site portray them through. I don't get anything in pretending to be anything other than human, and am quite content in keeping it that way.

About 7-8 years ago, I was recommended the story "Friendship is Optimal" by Iceman. I remember scoffing at the fact it was something from MLP. I was told to go in with an open mind about it. I guess my friend knew that I am very much into existential dread, because FiO hit *all* the right notes for me. So much so, that it has never quite gone away since reading it. The idea is simultaneously compelling and utterly revolting. What a balance.

And the authors. Oh, the authors. In the last year, I decided to reread FiO, and began a lazy circle branching out into other stories. It started easily enough. Jpabrony, Starscribe (seriously, how do they manage to write as much as they do? Starscribe, you're incredible.), Chatoyance, and so many others. From there, it isn't too hard to realize that while the Optimalverse has great stories, there are -so- many more here. And it's such a simple thing to start tugging at some other threads, using the trust built by the authors of these stories to read their other, non-Optimalverse works.

All of this from the viewpoint of someone who hasn't even dabbled in the source material from which these stories spring.

What an absolute treat. While I cannot possibly quantify the sheer joy I have received from this site, know that it is a great deal.

Guardian, you are one of those authors. Thank you, a hundred times over. While you owe nothing to any of us, I do hope you continue writing. At some point, I will take a look at some other stories you've written.

That's awesome to hear- so far how I'm feeling about the place I'm at too. I've seen our direct boss get in and work alongside us, and everybody's happy and great to work with! I hope things keep that way for the both of us!

It's all about having good coworkers, and low-stress IMO. I'd trade into any job, doing anything, as long as I could either stay where I am in the mountains, or work remote, and as long as work doesn't follow me home, and my colleagues are good.

I'm very, very blessed/fortunate to have the coworkers and boss I do (some of the best in the business). The code is stress enough.

  • Viewing 53 - 57 of 57
Login or register to comment