Brandon, here are the written answers I promised for today. Some of these I can answer solidly. For the rest, the honest answer is that I don’t know yet, or that it’s something we should work out together, and I’d rather say that than hand you numbers or dates that won’t hold. Thanks again for making time with Dave and me on Tuesday.
**Who leads the next upgrade, and when fee routing ships.** Chain upgrades are still prepared by RND engineering. Who coordinates the next one, and who owns chain comms with a backup (your ask in the v7.3.0 thread), isn’t settled yet, and I don’t want to name anyone before they’ve said yes. You offered to take that on if it’s recognized as work. I’d like to work that role out with you. The thread-per-upgrade commitment from my last post stands, and the v7.3.0 write-up is still owed.
On fee routing, nothing on chain today sends marketplace or activity fees to validators. Inflation is zero and the community tax is 17%, so validators get 83% of transaction fees and the pool gets 17%. Module-level fee routing (M013) needs a later chain upgrade and it isn’t scheduled. Your RFC on the split (agentic-tokenomics PR #111) is the live proposal for the parameters, and it still has no sponsor. I don’t have a date for you. Finding a sponsor and a build path is a good thing for us to work on together.
**Marketplace fees.** We got to the same place you did. There’s no public query for the marketplace fee parameters, the marketplace fee pool account is empty, and as far as we can tell no governance proposal has ever set them (only governance can). From the code, a buyer fee is added on top of the price and a seller fee comes out of the proceeds. Both go to the fee pool, and fees paid in REGEN are burned. What I don’t know yet is the stored values. Getting them means reading the node’s state directly, and I’d rather post that here than guess. The fees we can confirm are 5,000 REGEN to create a credit class and 1,000 REGEN to create a basket (the basket fee is burned).
**Regen Compute.** It retires on a monthly subscription cycle, not per inference, so a per-day number is the wrong lens. Through July it produced roughly 100 retirement records a month across about 34 addresses, which adds up to a few credits a month. From August 12 to 28 there was a burst of extra records, and I don’t know the cause yet, so I won’t quote anything that includes August until we do. I also don’t have the buyer-funded vs subsidized split in a form I can stand behind yet. And I couldn’t find a source for the ~88/day estimate.
**The sell orders and the $2.4M.** There are 42 open orders right now from 11 sellers, about 125k credits, mostly C06 and C02. Those are asks, and the chain doesn’t show who fills them, so I don’t know who buys and I won’t guess. One correction: the $2.4M comes from my State of Regen 2025 post (The State of Regen 2025). It’s the cumulative total purchased and retired on the network up to then, including earlier large sales like Microsoft’s, not a 2025-only figure. A clean 2025-only number would need to be derived from the chain, and given your ledger work that’s something we could do together, agreeing on the definition first like we said about issued vs cancelled. On a plan to grow retirements: the Registry is working on the demand side, but I don’t have something I can lay out publicly yet.
**Which funding pools we’re talking to.** That one I can’t put in a public thread, since most of those conversations aren’t ours alone to disclose. Where I think a verified-hour record helps most is what I said last time: reporting for money that’s already committed, like grants, public bond impact reports and forward commitments. Happy to go through specifics directly.
**The LiquidityDAO pilot, route and timing.** I don’t know yet. Before I advise a route I want to understand how the LiquidityDAO address is governed (who signs, and under what rules), and I haven’t done that work. Let’s figure out the route together once that’s clear. I won’t give you a timing before then.
**VSH01.** Nothing has changed: still yes in principle, once the first hours are accepted.
From Tuesday, I still owe you a post that pulls together the data module anchoring and signing tools, the relevant repos, and the claims engine work. I’ll put that up as its own thread. And I’m looking forward to the more digestible version of the proposal you mentioned.