escarpment

/sh.mb

I can't quite recall when I first used this slash command... perhaps early summer this year, or maybe even before that. Regardless, it was a formalization of something I'd been doing in LLM•space for a while now, and I just gave it both a term — "shadow micro-blogging" — and a formal string, "/sh.mb" to denote it in a prompt.

And here's the thing: I almost always used it at the end of the prompt, not at the start; there's an intentionality in this positioning that is meaningful. Unlike slash commands in the various models, e.g.:

...the way I use mine doesn't tell the model what to do or perform, but instead tells it what I am doing with the prompt. This of course is not a canonical function or operator in model space, but, unsurprisingly, the model understands the way I use it (for the most part, with occasional lapses in perception), which feels quite organic for me, given the idiosyncratic way I've used LLMs all this time.

Speaking of which, what exactly is this slash command about anyway? ::chuckle::

Well, it's what its nether half says: micro-blogging — sparse bits of text, prose, or story that function like a blog post; with the shadow part of it locating the text not on a usual blog page but instead is, weirdly, within a prompt to a large language model.

How and why this came about, and what it is really for, is grist for another post here... as it will require a bit of investigative plumbing into the psychology (or sanity) of the LLM interlocutor, an orthogonal topic for another day. Be that as it may, the process of shadow microblogging evolved quite organically in the writer-interlocutor's ongoing, longitudinal engagement with these novel, synthetic minds over the past several years now.

Hint: it is all of a piece with QS (or "qualic sharing") which itself is yet another topic for some other bloggy day in this space. Stay tuned.

[ Yesterday: a night scene sandwiched by day's images... ]