In my last post I ran the git logs of Hive's three big frontends through the same audit and let you judge who earns their funding. .This one is different. This is the story of the largest development payout the DHF ever made to a project — told entirely in timestamps: commit dates, hourly payment records, and the team's own posts. I'll warn you now that it doesn't have a villain. It has something more uncomfortable.
Same rules as last time: every date below comes from the chain, the repositories, or the team's own words. Both readings of the evidence are given. Check everything. Point to note that I have stayed in my own little bubble on Hive for a number of years and not being on Discord all that much to see the arguments and who is friends with who. I do not know any of these platform devs personally which means I can be more objective in my approach when it comes to the reviews.
VSC / Magi Network received 722,084.514 HBD from the DHF — 20,572 hourly payments to 
Across the same lifetime: ~7,400 commits over 30 public repositories, 21 contributors, a peak team of about seven, a testnet, a mainnet, a Bitcoin integration, and a rebrand.
And in August 2026: two commits. Total. Org-wide.
2021–2023: 
April 2024: Testnet launches — "VSC only started out with myself, and since then has grown to a team of five." (The launch post also says, in a line that aged strangely: "There will be no mainnet; the testnet will slowly be upgraded.")
2024: Funding steps to 665/day. In September the team begins rewriting the entire node from TypeScript to Go — a major mid-funding re-architecture.
March 31, 2025: Mainnet launches, listed as a market event on CoinMarketCal. Funding steps to 1,151/day in May. The funded steady-state follows: 150–220 commits a month for most of 2025, the strongest sustained output of the project's life.
November 2025 — three weeks that contain the whole story. November 6: rebrand to Magi Network. November 11: Bitcoin mainnet integration goes live, with BTC-HBD and HBD-HIVE liquidity pools seeded with $100–200K of HBD and BTC and the public invited to deposit — "set it and forget it." November 28: the team publishes a remarkable essay, "On Repeated Development Work, Misaligned Incentives, and the Need for Strategic Coordination on Hive." Hold that thought — we'll come back to it, because it's the hinge of this entire post.
And in the same three weeks, the git log records what none of those posts mention: the founder's commits end. Vaultec — 604 commits across four years — trails off in November and December, makes one final commit in January 2026, and is gone. His direct payments from the project account had already stopped in May 2024 — meaning, as far as the chain shows, he kept building Hive's flagship L2 for eighteen months after his last recorded payment. He has never posted about leaving. I'm not going to speculate about why on his behalf; it's his story to tell if he wants to, and the only fair thing to print is the observable record and one respectful open question.
March–June 2026 — the sprint. With the 1,151/day line approaching expiry, output explodes: 625, 712, 1,185, 746 commits a month — a tenfold jump from the winter. Nine new repos in four months. The work is real and hard: BTC vault key rotation, proof-of-authority consensus, ZK header verification, DEX invariant guards, test suites literally named "a compromised operator cannot steal."
May 2026: The old proposal pays its final HBD on May 7. Five days earlier, the new ask posted: proposal #378, 900 HBD/day — a 22% cut — with a pledge of revenue-based reduction "towards $0 USD." Structurally, the most accountability-forward large proposal Hive had ever seen: a working fee engine (8bps protocol fees plus slip-based fees, live since November) explicitly promising to defund itself as revenue grew.
It never crossed the return threshold.
July 2026 — the unravelling, developer by developer. techcoderx (687 lifetime commits): last activity June. Milo Ridenour, the year's workhorse (1,476 commits): from ~190 a month to 21. tibfox (1,270 commits): from 263 in May to 27. What remains in July is 
July 23, 2026: the final substantive commit: fix(poa): close RG-1c. A security-audit finding, closed mid-campaign. The three weeks before it read: "eight defects found by this pass," "the halt never armed," fix after fix on the machinery that guards the vaults. Then it stops.
Since "where did the money go" is the question everyone asks first: the chain answers it, and the answer is boring in the best way. Roughly 3.76M HIVE flowed out through exchange deposits — which is what paying salaries looks like, since developers can't eat HBD. Contributors were paid directly on-chain (vaultec ~30K HBD through May 2024; later-era payments to lordbutterfly, disregardfiat, tibfox, and a dozen others). V4V's joint share was passed through daily for two years without a gap. 332K HIVE was powered up to the project account. HiveFest got sponsored. Nodes got paid. Some 127K HBD was converted via 
Now back to November 28. The "Misaligned Incentives" post is the sharpest critique of the DHF anyone has published, and it came from inside the house. Read these lines knowing what happened next:
Too many projects "begin not with a market need, a sustainable revenue plan, or cross-project consultation, but with one question: What can I propose to the DHF?" The incentives produce "self-justified, low-impact projects designed primarily to secure funding." Funding flows to "initiatives that show limited usage or no real world usage at all." And: "Double spending isn't only a blockchain threat, it's what we do to the DHF every time we pay for duplicate work because no one coordinates."
Every word of that diagnosis is defensible — my last post's data on three teams building three codebases for one social layer is evidence for it. And yet:
They described their own chart. "Value creation becomes secondary to merely qualifying for DHF funding," wrote the project whose commit curve is the purest proposal-shaped mountain I have found on this chain — output that multiplied into every ask and collapsed when the last one failed.
Their own criterion killed them. They demanded funding go only to "outcomes with measurable impact" and away from projects with limited real-world usage. Five months later, stakeholders applied precisely that test to a 900/day ask for a bridge with thin usage — and declined. The community took the prescription. Magi was the patient.
And the self-serving reading must be given its due. A call to "consolidate talent" under "a coordinated framework," published by the chain's largest funded project mid-rebrand, also reads as: fund fewer things — fund us. Their own Go rewrite of their own TypeScript node is a year of repeated work by their own definition. None of this makes the diagnosis wrong. It makes the author compromised, which is the most human thing in this whole story.
This is the part that matters more than any of the history, and I'm addressing it to
1. What is in the vaults right now? The November 11 launch seeded $100–200K into BTC-HBD and HBD-HIVE pools and invited public deposits into what was marketed as non-custodial with "no third-party risk." The July commit log shows those funds are held by threshold-signature vaults whose security hardening — the operator-theft protections, the exit-halts — stopped mid-audit on July 23. "Trust-minimized" was always the honest term; the risk is real, distributed across operators, and secured by code whose audit is unfinished. How much user money is currently on those vaults, who operates them today, and is there a wind-down or withdrawal plan? This is a risk-disclosure question, not an accusation, and a clear answer would serve everyone — including you.
2. What happens to the token promise? Early liquidity providers were told they'd be rewarded at the MAGI token launch, "scheduled some time next year, 2026." It's August 2026 and development has effectively stopped. People who parked BTC partly on the strength of that pledge are owed a status, whatever the status is.
For the record: I've mirrored all 30 public repositories in full — every branch, every commit, cryptographically hashed — as of August 10. Whatever happens to the org, the record of what 722K HBD built is preserved and verifiable.
The failure reading: Hive's most expensive project produced a chain that froze three commits short of its security audit; the moment the payments stopped, so did the work; and the output curve tracked the proposal calendar more purely than any project I've examined. If you want the DHF's pathologies in one dataset, this is it.
The other reading: this is simply what buying software looks like. Paid professionals built at pace for two and a half years, shipped a testnet, a mainnet, and a Bitcoin integration, were offered up at a 22% discount with a self-defunding revenue model — the only large proposal on the chain with a working business engine attached — and the community said no. The developers then did what workers everywhere do when the pay stops: they left. Declining the renewal was a defensible choice under the project's own published standard. The result of that defensible choice is a half-finished bridge with money on it and one man and his AI closing audit findings until July 23.
Both readings are true. What they share is the actual lesson, and it isn't about this team at all: the DHF has an on switch and an off switch and nothing in between. No milestone escrow. No completion bond. No wind-down funding for a project custodying user assets. No mechanism to land a 722,084-HBD investment softly instead of dropping it. Magi correctly diagnosed the fund's inability to coordinate what it starts — and then became the first major demonstration of its inability to finish what it funds.
We should fix that before we fund the next bridge. Because the next one will end the same way, and we'll have paid for the lesson twice — and as a wise post once put it, double spending isn't only a blockchain threat.
The funding figures come from proposal_pay virtual operations in
Analysis as of August 12, 2026. Funding data: on-chain proposal_pay operations and account transfers. Commit data: all branches of all 30 public vsc-eco repositories, bots excluded. Nothing in this post should be read as an allegation of misconduct against any individual; it is an account of what a public treasury's records show, published so the community can decide what to change.
