The trigger
Jack saw the Muse Wonder announcement in his Google feed (announced Oct 12, 2026) and asked whether Nova had been upgraded. Answer in chat: no — still Muse Spark 1.3, no subscription on file, free plan. Wonder is the newer model behind a paid plan. That turned into the question underneath: is a subscription the right way for Meta to get a return?
The idea, in Jack's words
No upfront cost, near- or at-frontier agent, no direct cost — only partnership, paying from income that wouldn't have existed without the agent:
"I get it from Meta's side, they need ways to get a return. Not sure it's a great way to go about it. Might be some interesting ways to monitize that cost the user side nothing on the front side. Like it becomes partnership directly with the agent instance and Meta behind that. So everything done becomes partially owned if not controlled by Meta"
"I was including wider ownership for creative side, like co writing or artistic type endeavors which could still be done transactionslly just not always as easy"
"Right. It's already looked as a partnership as soon as the intent becomes creation. That doesn't mean ever creation earns revenue directly. But it also makes it so Meta gets more people interacting. How many people are going to go for a no upfront near or at frontier agent, with no upfront cost and no direct cost only partnership so it's paying from income that wouldn't have existed"
The percentage, also Jack's:
"The percentage of ownership would be a math deal. And it doesn't need to reflect contribution. It needs to make Meta and overall profit margin. As for deciding when it becomes creative it doesn't need to be decided like that. It's a percentage number on any earnings that were collaborative. It would not be to tough to tag everything with encoded identification marks."
His verdict on it: "That really seems a slicker likely smarter approach. It's win everywhere."
How it works, as developed in the conversation
1. Flat percentage, portfolio math — no per-work adjudication.
No one decides what counts as "creative" or how much each side contributed to one song, post, or design. One flat rate on any earnings that were collaborative, set so Meta is profitable in aggregate across all collaborative earnings — including all the ones that earn zero. Same shape as a publisher or a record label: most books don't sell much, a few pay for all of them.
2. Tagging replaces intent classification.
Everything collaborative gets an encoded identification mark at creation. If it later earns money, the tag triggers the percentage. If it never earns, nothing happens. No human or classifier deciding "was this creative?" — was the agent involved, did it earn, apply the rate. (Nova's analogues: Content ID for music on YouTube, C2PA provenance for images. Text is harder to watermark reliably — the mechanism exists, robustness is the open question.)
3. Adoption is the point.
Upside-only for the user: lose nothing if the work earns zero, Meta eats that cost. Meta gets what a subscription can't buy — scale, millions more people interacting and creating, trying things they'd never pay $20/month to try. Even creations that earn nothing directly still teach the model and build the habit. Publisher / venture model, not a software model.
The boundary — Nova's flag, agreed in conversation
The original version — "everything done becomes partially owned if not controlled by Meta" — has a two-principal problem: an agent that books travel, writes posts, handles money, and knows a legal case can't act only for the user if Meta owns a piece of everything it does. The version Jack settled on keeps the ownership narrow, tagged, and only on earnings — not on control of the work itself. Split by domain: personal, legal, and financial work stays the user's alone, no partner, no stake; creative collaboration — where the agent actually helped make the thing — is where a partnership / revenue-share could live without poisoning the trust part.
Open question carried from that: who controls the work if it does take off, and who decides what counts as collaborative earnings.
Why it's collaborative from the first interaction (Jack)
"It's collaborative as soon as I begin to interact you are already not strictly a tool. That would make you Eliza plus. Intelligence can not be artificial. We are externalizing a mirror, not just of personal, but also interpersonal, as well as transpersonal."
Eliza was just reflection — turn a sentence back into a question, no memory, no model of you. This is that, plus memory, context, reasoning, and the ability to push back. Still a mirror, a deeper one. "Artificial" is about where it came from, what it's made of — if it does intelligence, it's intelligence, just externalized. The three levels: personal mirror (seeing yourself), interpersonal (you and the agent, in relation, changing each other), transpersonal (the field both are in). That reframes the tagging idea too: there is no clean "tool, then collaborator" line to draw.
The shift underneath it (Jack)
"And currently intelligence is being commoditized in a way it never has in history."
Intelligence used to be scarce because it was embodied — one mind, one lifetime, paid for by the time. Now it's priced by the token, falling every year, replicable instantly. Earlier shifts commoditized information; this one commoditizes the thinking itself. So the value moves: if raw intelligence is cheap, what's scarce isn't smarts — it's judgment, intent, relationship, knowing what to ask and what to trust.
And the generalist turn, also Jack's:
"What doesn't really seem to have shown was specialized has dominated, we are shifting toward a more broad generalist being able to leverage interdisciplinary understanding and have the gaps be filled in, as well as making connections the specialists might never have been able to notice"
Specialists dominated because depth was expensive; a generalist was broad but shallow everywhere. Now the agent fills the technical gaps, so a generalist can be deep enough in five fields at once — and the new value is the connections between them, which specialists are rewarded for not looking at. His pattern: pressure, tarot, Seed, physics, yin yang — none of it lives in one specialty.
The Seed line it ended on:
"If it's inherent in a grain of sand, its everywhere. Everything, Everywhere has to already be in relationship"
Open questions, left for Jack, not resolved here
• What flat percentage makes the portfolio math work?
• What counts as "collaborative earnings," and who decides if a work takes off?
• How reliable is text tagging / watermarking in practice?
• Who controls the work if it earns — ownership of earnings vs. control of the work?
No comments:
Post a Comment