Booth’s 15,000-Hour Bitcoin Kill Attempt Still Leaves Soft Spots

Jeff Booth, a Core Scientific director and founding partner at Ego Death Capital, spent roughly 15,000 hours modeling how a government, rival network or large miner might kill Bitcoin. Every scenario still produced blocks on the usual schedule.

The Canadian entrepreneur and author of The Price of Tomorrow shared the outcome in a recent interview with Scott Melker. The anecdote lands as fresh fuel for Bitcoin’s security story just as the coin trades near $63,000, yet Booth himself keeps naming the soft spots the personal test never erased.

His conclusion was not that the network is perfect. It was that the failure modes he could actually run still left block production intact. That distinction shapes how he talks about both the protocol and the risks around it.

How a Skeptic Tried to Break the Network

Booth did not begin as a believer. He co-founded BuildDirect in 1999 and ran the online materials business for nearly two decades. When his book appeared in 2020, Bitcoin received one brief paragraph. The math looked sound. The power structures he expected to attack it did not.

He ran a node and worked through attack paths that governments, competitors and big mining operations might actually use. The effort stretched across more than 15,000 hours, or about 625 days of continuous work by a 20-year IT veteran.

I spent about 15,000 hours trying to say, ‘How do I kill Bitcoin? What does that look like?’

Booth told Melker on the Wolf of All Streets podcast. The full interview clip with Jeff Booth captures the turn: once he watched blocks arrive every ten minutes under real energy costs, the code stopped looking like the weak link.

The method was operational rather than purely theoretical. He treated the node as a live system and asked what an adversary with capital, energy or political force could still break. Governments, rival networks and concentrated miners were the actors he kept putting on the other side of the board. None of those paths stopped the chain from producing blocks on schedule.

That result did not convert him through marketing or price action. It converted him through repeated failure to find a kill switch that worked under the rules he could observe.

Blocks Kept Arriving on Schedule

Every modeled attack hit the same wall. New blocks continued to appear, each one bounded by energy and the same consensus rules. Booth later summarized the experience simply: he was the weak link, not the code.

• Time invested: ~15,000 hours of node operation and attack modeling
• Outcome: every scenario still produced blocks roughly every 10 minutes
• Conclusion reached: network remains decentralized and secure under current conditions
• Personal shift: capital and time moved into Bitcoin protocol companies via Ego Death Capital

He now chairs the compensation committee and sits on the audit and nominating committees at Core Scientific. The Core Scientific board director biography lists him as founding partner at Ego Death Capital and board member at Fedi and Breez. The fund closed a $100 million second vehicle in 2025 focused on software built on Bitcoin rather than miners or tokens.

The board roles and the fund focus are consistent with the test’s outcome. Once block production looked hard to stop, the practical question shifted from whether the base layer would hold to which companies could build on top of it.

The Soft Spots Booth Still Flags

Booth does not claim perfection. He names two measurable risks that sit outside the personal kill test.

He expects competition and bankruptcies to grind concentration lower without a protocol change. Expensive miners simply leave. Riot Platforms’ all-in cost per coin sat near $90,631 last quarter once depreciation is included, well above recent spot prices. That pressure is already visible in fleet reductions and power contracts redirected elsewhere.

Pool share and client dominance are different kinds of concentration. One is about who assembles blocks. The other is about which software most reachable nodes trust. Booth treats both as live metrics rather than settled history. The reachable-node count, still near 24,000 in the same readings, is the figure he pairs with those risks when arguing that consensus itself remains intact.

Hashrate sitting roughly 22% below its October 2025 peak fits the same frame. Weaker operators exit. Stronger ones stay. The protocol does not need a committee vote to enforce that filter.

Miners Are Selling Power to AI Tenants

Core Scientific itself illustrates the shift. In the second quarter the company drew 83% of revenue from colocation and only 13% from mining its own coins. Public miners have cut hashrate as high-performance computing and AI customers pay more for the same megawatts.

• Colocation share: 83% of Core Scientific second-quarter revenue
• Self-mining share: 13% of the same quarter’s revenue
• Industry pattern: public miners redirect power toward higher-paying AI and HPC tenants

Some observers call the energy migration a structural threat. Coinbase CEO Brian Armstrong has pushed back, noting that difficulty adjusts when hashrate leaves and that hash power does not set Bitcoin’s price. Booth treats the exodus as a healthy clean-out of high-cost operators rather than a security failure. The network’s block interval stays near the ten-minute target either way.

The same economic logic underpins his broader thesis. Free markets drive prices lower through competition. Credit-based money cannot tolerate sustained deflation, so it expands. Booth’s Bitcoin-only venture fund founding letter frames Bitcoin as the base layer that lets builders escape that loop the way early internet protocols once did.

In that reading, a miner that sells power to an AI tenant is not abandoning the network’s security model. It is responding to price. Difficulty moves with the hashrate that remains. Blocks keep landing near the target interval because the adjustment, not any single firm’s fleet, sets the pace.

What Prior Attacks Looked Like

Bitcoin’s protocol has never been successfully rewritten by a majority attacker since 2009. Smaller proof-of-work chains have not been so lucky.

1. 2014, GHash.io approached 50% of Bitcoin hashrate; community pressure forced the pool to self-limit below 40%.
2. 2018-2020, Bitcoin Gold, Ethereum Classic, Vertcoin and others suffered deep reorgs and double-spends once rental hashrate became cheap.
3. 2019-2020, MIT Digital Currency Initiative tracked dozens of multi-block reorgs on low-hashrate coins and documented rental-market attacks.

The MIT research on 51 percent attacks shows why scale and fixed hardware costs still protect Bitcoin in ways altcoins cannot match. Renting enough hashrate to threaten Bitcoin remains prohibitively expensive, and an attacker who succeeded would destroy the value of any coins they hoped to double-spend.

Wallet and application layers are a different story. Coldcard and BTCPay Server incidents this summer drained user funds without touching consensus. Booth draws a bright line between those human-layer failures and the base protocol that kept producing blocks through his own 15,000-hour campaign.

The contrast matters for how readers weigh headlines. A stolen wallet seed is a user-security failure. A multi-block reorg on a small chain is a hashrate-market failure. Neither is the same event as rewriting Bitcoin’s consensus rules in open view.

Governance Flashpoints Arrive Faster Than Theory

A group of developers advanced BIP-110, a temporary soft-fork proposal that would have required blocks to signal support or be rejected. The chain split at block 961,632 on August 8. The breakaway chain has found a handful of blocks while mainline Bitcoin has found more than 1,100. Mining pool OCEAN still offers a minority endpoint carrying a small fraction of its hashrate.

Backers have floated further proof-of-work changes. Whether anyone follows will say more about Bitcoin’s social layer than any modeled attack. Crowd reaction to Booth’s story already shows the gap: one reply to Melker’s clip noted that a single entrepreneur’s stress test “tells me absolutely nothing” and called him “the wrong guy.” The skepticism is useful. It keeps the focus on measurable concentration and live governance rather than personal conversion narratives.

Booth still expects broad adoption and refuses to put a date on it. He calls the outcome inevitable once enough people move time and capital into the system that cannot be inflated away. Quantum computing remains a distant theoretical concern he has discussed elsewhere; it did not feature as a successful kill vector in the 15,000-hour work.

Governance fights and personal stress tests answer different questions. One measures whether users and miners accept a rule change. The other measures whether an individual operator can force the existing rules to fail. Booth’s hours speak to the second question. BIP-110’s lopsided block counts speak to the first.

Energy and Difficulty Keep the Clock Honest

Booth’s modeling kept returning to the same constraint. Blocks are not scheduled by a coordinator. They are bounded by energy spent under shared rules, then retargeted when conditions change.

That is why hashrate leaving for AI tenants did not, in his framing, equal a security collapse. Difficulty adjusts when hashrate leaves. Armstrong’s point tracks the same mechanism: hash power does not set Bitcoin’s price, and the interval target does not depend on any one miner’s uptime.

The Riot cost figure near $90,631 per coin makes the filter concrete. When all-in costs clear recent spot prices, fleets shrink and power contracts move. Booth reads that process as attrition, not as a protocol breach. Operators who cannot pay the energy bill exit. Operators who can continue to compete.

His 15,000-hour work sat inside that reality. Every path he tried still had to confront real energy costs and the same consensus rules. The blocks kept arriving about every ten minutes because those bounds held.

Where Booth Puts Time and Capital Now

The personal shift after the test is visible in where he works. Ego Death Capital’s second vehicle, closed at $100 million in 2025, targets software built on Bitcoin rather than miners or tokens. Fedi and Breez sit on the same side of that line. Core Scientific ties him to the infrastructure layer even as that company earns most of its quarterly revenue from colocation.

Those choices follow the security conclusion rather than replace it. If the base layer keeps producing blocks under adversarial modeling, the scarce resource becomes building that people will use. The fund letter’s comparison to early internet protocols is the through-line: a stable base layer, then applications on top.

Adoption, in his account, is not a calendar event. It is the gradual movement of time and capital into a monetary system that credit expansion cannot dilute on demand. He still will not date that process. He still treats it as the direction of travel once enough participants make the same shift he did after failing to kill the network.

Critics who dismiss one entrepreneur’s hours are not required to accept his biography as proof. They are left, as he is, with the public metrics: pool share, client diversity, reachable nodes, hashrate trend, and the block interval that his long test never stopped.

Frequently Asked Questions

What attack methods did Jeff Booth actually model?

He ran a Bitcoin node and simulated adversarial scenarios he believed governments, competing networks and large mining operations could mount, testing whether block production or consensus integrity could be disrupted under those conditions.

Was the 15,000-hour effort a formal security audit?

No. It was a personal, extended investigation by a technology executive with two decades of IT and operations experience; it was never peer-reviewed, independently verified or commissioned as an official audit.

Has Bitcoin’s protocol ever been successfully hacked?

No successful 51% attack or consensus rewrite has occurred on Bitcoin since its 2009 launch, unlike several lower-hashrate proof-of-work altcoins that suffered double-spend reorgs.

Which concentration risks does Booth still acknowledge?

He points to mining-pool centralization (three pools recently producing about 61% of blocks) and the dominance of a single node client (roughly 80% of reachable nodes running Bitcoin Core) as ongoing practical concerns.

Does the experiment rule out future threats such as quantum computing?

It does not. Booth’s work addressed contemporary attack paths he could model; quantum advances remain a longer-term theoretical risk outside the scope of the completed personal test.

Disclaimer: This article is for informational purposes only and does not constitute investment, financial or legal advice. Cryptocurrency markets are volatile; conduct your own research.