Miners Can Steal From Drivechains. Here's Why They Won't.
Updated: 6 hours ago
Yes, miners can steal from a drivechain. Paul Sztorc, who invented drivechains, says so himself.
The design gives miners the job of approving withdrawals. That includes bad withdrawals. Nothing in the code can tell an honest payout from a theft.
The real question is whether they would. A drivechain theft takes weeks, happens in public, and costs the thieves more than it pays. Below is the strongest case against drivechains. Then we answer it with the rules in BIP 300 and the numbers behind them.
eCash (ECX) is the first coin running drivechains with real money. So this is no longer just a debate. It is a test anyone can watch.
Drivechain security in short
Drivechain security is economic, not cryptographic. Miners vote on every withdrawal from a drivechain, one block at a time.
On eCash, a withdrawal needs more than 6,048 yes votes within 12,096 blocks. At ten minutes per block, that is six to twelve weeks. Stealing is possible, but it is slow, public and costly. A miner who wants profit earns more by leaving the coins alone.
Last verified: November 13, 2026.
How a drivechain withdrawal works
Getting coins into a drivechain is quick. You send ECX to the drivechain's slot on eCash, and it shows up on the sidechain.
Getting coins out is slow on purpose. Here is the process, from the BIP 300 spec and eCash's mining software:
The sidechain gathers everyone who wants out into one batch, called a bundle.
The bundle becomes a single payout transaction on eCash.
Each new block can vote yes on the bundle, vote no, or stay quiet. The count moves by one at most.
On eCash, the bundle pays out once it has more than 6,048 yes votes. It gets 12,096 blocks to do it.
If it can no longer reach 6,048 in the blocks left, it is thrown out at once.
BIP 300 describes it as a giant multisig. Instead of keys, each block is a signature. On eCash it takes more than 6,048 of 12,096 to sign.
eCash uses a shorter window than BIP 300
BIP 300 as written asks for 13,150 votes in 26,300 blocks, about three to six months. Most older articles, including Paul's FAQ, quote those numbers.
eCash's mining software, version 1.0.0, cut the window to 12,096 blocks and the bar to more than 6,048 votes. The vote still needs half the window, so the security math below is the same. Only the clock is shorter.
Why blocks, not days
The window is counted in blocks, not on a calendar. At ten minutes a block, 6,048 blocks is 42 days and 12,096 blocks is 84 days.
eCash block times can run fast or slow, so the real time can stretch or shrink. Our article on how eCash changed its difficulty adjustment explains why. Either way, a theft can never be quick.
Can miners steal from drivechains?
Yes. Miners choose where a withdrawal bundle pays out. If enough of them back a bundle that pays themselves, it goes through.
A miner majority can already attack Bitcoin itself, by rewriting recent blocks. Drivechains add a second thing a majority could do.
So the honest claim is not that theft is impossible. It is that theft does not pay. The rest of this article tests that claim.
The same bet Bitcoin already makes
Bitcoin already runs on this kind of promise.
Nothing in Bitcoin's code stops a miner majority from rewriting recent blocks. They do not do it, because honest mining pays better.
Drivechains make the same bet. Miners could approve a bad withdrawal. The design is built so that it pays worse than staying honest.
The two are not identical, and the critics are right about where they differ. A reorg lets miners double spend their own coins. A drivechain theft takes coins that belong to other people.
The answer to that gap is time. A reorg can be over in hours. A drivechain theft needs a majority for weeks, in public, with every vote on the record.
The case against drivechains, at full strength
The best way to argue for drivechains is to state the critics' points better than they do. Here are the five strongest.
1. The bribe math flips
A drivechain could hold far more value than it will ever pay in fees. Then one big theft might beat years of honest income. The more a drivechain holds, the louder this worry gets.
2. Rented hashrate does not care about the future
Miners who own warehouses of machines care about next year. Someone renting hashrate for a few months may not. They have no hardware at stake and no fee stream to protect.
3. Theft becomes a feature, not a crime
Greg Slepak raised this on the Bitcoin developer mailing list in 2017. He argued drivechains let miners steal coins, which they cannot do today. He called it a clear weakening of Bitcoin's security.
4. Miners could use theft as a weapon
Peter Todd made a sharper point in the same debate. A miner who checks sidechains better than its rivals might let a theft happen on purpose. The fallout could give it cover to push weaker miners off the network.
5. eCash's hashrate is for rent
This one is specific to eCash. eCash has far less hashrate than Bitcoin, and the machines that mine it can also mine Bitcoin. During testing, rented hashrate poured in and out of the chain in big waves.
A majority of eCash's hashrate can be rented. A majority of Bitcoin's cannot.
That fear got a live test in November 2026. Most of eCash's hashrate left within days, and rented machines filled part of the gap.
Each of these is a fair point. Here is why we think none of them wins.
Why miners won't steal
The heist runs in slow motion, on live TV
A theft cannot start in secret. The bad bundle sits on chain with its payout address in plain view. Its vote count ticks up block by block, for weeks.
Paul's drivechain FAQ compares it to shoplifting on live national television while moving in ultra slow motion. Everyone can see which miners vote yes. Users can leave through faster swap routes. Honest miners can vote no.
Honest miners can make theft much harder
Most explanations skip this part. Staying quiet and voting no are not the same thing. A quiet block adds nothing to a bundle's count. A no vote takes one away.
Under BIP 300, a yes vote for one bundle also counts against every rival bundle on that sidechain. A miner who backs the real withdrawal is voting against the fake one.
That changes the math. If honest miners stay quiet, a thief needs just over half of all blocks. If honest miners vote no, the thief needs three quarters of all blocks, held for the whole window.
Here is the working. Say the thief mines a share of blocks called T, and everyone else votes no. On average, each block the count rises by T and falls by the rest.
Over 12,096 blocks the count must net 6,048. So 2T minus 1 must reach one half, and T has to be at least 75%.
This is our own calculation and it assumes steady block shares. By default, eCash mining software votes yes only on a bundle its own node proposed or passed along. It stays quiet on the rest. Voting no takes a setting the software calls alarm, or a yes vote for the real bundle.
Theft kills the goose
Drivechains pay miners. Each sidechain block wins its spot by bidding for space in an eCash block. That bid goes to the miner. A theft ends that income from the chain it hits.
The vote rule is the same for every drivechain. Paul's FAQ says the number is fixed partly so all sidechains stand together after a theft. A theft on one chain is a warning to every user of every chain.
The loot loses value before it arrives
What do the thieves get paid in? ECX, the coin a public theft would hurt most.
The timing makes it worse for them. The market sees the theft weeks before it pays out. Holders can sell, and the price can fall, long before the thieves collect a single coin.
To be fair, one classic argument is weaker on eCash. On Bitcoin, miners stand to lose the value of machines built for one coin. eCash miners can point the same machines back at Bitcoin. That is a real gap, and the slow window is what covers it.
Users get the last word
If miners do not back down, users can refuse any block that includes the theft transaction. Paul points out this needs no new code. The theft transaction is known weeks ahead, so users only need to block that one transaction.
If the market values the honest chain more, miners follow the money. Paul is careful here, though. Drivechains do not rely on this. It is a backstop, not the design.
Answering the critics, point by point
The bribe math
Paul's model rests on two things. The first is the fees miners earn from sidechains. The second is the extra value ECX has because it can move into sidechains at all.
A theft puts both at risk across every chain, not just one. The bribe has to beat all of that, paid in a coin the market has spent weeks marking down.
There is a real limit here. If one chain held huge deposits but paid tiny fees, its safety margin would shrink. Fee levels are the number to watch as deposits grow.
Rented hashrate
Renting a majority for an afternoon is one thing. A theft needs a majority for weeks on end, paid for the whole time, in public.
Honest miners and anyone else can rent hashrate too, and vote against it. Rentable hashrate makes eCash a harder test than Bitcoin would be. That is exactly what makes it a real test.
Theft as a feature
Users opt in. No one's ECX sits in a drivechain unless they put it there. Coins in your own eCash wallet are never part of a withdrawal vote.
Paul's FAQ puts it plainly: whether a sidechain is worth the risk is each user's call. Today, the alternative for most of these features is an altcoin or a custodian.
Theft as a weapon
Paul answered Peter Todd directly in 2017. Miners can stay quiet on any bundle, which takes no side. And the whole process is public. No miner can quietly let a theft through and claim it did not know.
Two newer edges found in testing
eCash tester Michael Blowes documented two lesser known cases in his drivechain guide. Both were read from the mining software's code and shown on a private test network.
Replacing a live drivechain. Miners can hand a slot to a new sidechain, and the coins in that slot stay with the slot. That is another road to the same coins. On eCash it needs the same 6,048 votes over 12,096 blocks as a withdrawal, so it is the same slow, public heist in a different costume.
Contested empty slots. If two new sidechains propose for the same empty slot, mining software backs whichever proposal's code sorts highest, not the first or most popular one. A rival can tweak a proposal until it wins. That is a nuisance for new chains, not a way to take deposits.
eCash is the experiment
For about a decade, drivechain security was argued on mailing lists. On eCash it runs with real coins, real miners and real money at stake.
If the theory is wrong, someone will try a theft and it will work. The market will price drivechains to match. If the theory is right, bundles will keep paying the right people, block after block.
The first real withdrawal, November 11
The first withdrawal bundle in eCash history was proposed on November 11, 2026. It came from Thunder and paid two small withdrawals.
It taught us two things the theory does not tell you.
The first is who votes by default. We read the mining software at version 1.0.1. Out of the box, a miner votes yes only on a bundle its own node proposed or passed along. It stays quiet on everyone else's.
That is what happened. The first yes votes all came from one small miner whose own node had carried the bundle. He said in the developer group that he had not known he was voting.
The second is who checks. The main chain cannot tell whether a bundle is honest. It checks the format and counts the votes. It knows nothing about balances on the sidechain.
Checking takes a person. The operator of the largest pool explained how. His team runs its own Thunder node and compares each bundle against it.
This one paid exactly the two withdrawals it should, so the pool began voting yes. He also said the pool refuses by default any bundle it cannot check.
What that means for both sides
Critics get a fair point. Checking is manual today, and it rests on a few pool operators choosing to do it.
Defenders get one too. Silence is the default, so a stranger's bundle earns no free votes. And the largest pool checked before it voted.
One number cuts against comfort. On November 11, a single pool had mined about 92 percent of the last 144 blocks. That is above every threshold in this article.
A pool is not one miner. The machines behind it can leave, and in November most of eCash's hashrate did leave within days. Still, pool share is the number to watch.
Only one bundle per sidechain can succeed at a time. Sztorc confirmed that a rival bundle can replace a pending one, but two cannot pass together.
How to watch it yourself
Every withdrawal bundle and its vote count is public on the eCash chain. Drivechain explorers show pending bundles, their payouts and their votes.
The mainnet view is at explorer.blockfab.org/drivechain. On November 13, 2026 it showed Thunder carrying one pending bundle, the first withdrawal in eCash history.
It also shows how small the prize is today. Every coin locked across all five live drivechains added up to about 22 ECX. Thunder and Flush held nearly all of it.
That is the honest state of the test. The argument above is about what miners do when drivechains hold real money. They do not hold it yet.
A bundle that does not match a sidechain's real withdrawals would stand out weeks before it could pay. You can start with the chains in our drivechain guide.
We think Paul is right. Watch and see.
Drivechain security FAQ
Can miners steal from drivechains?
Yes, but only through a public vote that lasts weeks. On eCash a withdrawal needs more than 6,048 yes votes within 12,096 blocks. Any theft is visible long before it pays out.
How long does a drivechain withdrawal take?
On eCash, at least 6,048 blocks and at most 12,096 blocks. At ten minutes per block, that is about six to twelve weeks. BIP 300 as written uses a longer window of three to six months.
What share of miners does a drivechain theft need?
Just over half of all blocks if other miners stay quiet. About three quarters if other miners vote no.
Is my ECX at risk if I never use a drivechain?
No. Withdrawal votes only control coins deposited into a drivechain. ECX in your own wallet is not part of any vote.
Can drivechains on eCash touch my Bitcoin?
No. eCash drivechains run on eCash. Your BTC is on a separate chain and is not affected.
What happens if a theft attempt starts?
It appears on chain with its payout address. Users can leave through faster routes, miners can vote no, and users can refuse the theft transaction.
Do I have to wait weeks to leave a drivechain?
Usually not. Most people swap with someone else for coins on the main chain. The first such swap on eCash was on November 13, 2026. Ten coins on Thunder were traded for nine on the main chain.
Who invented drivechains?
Paul Sztorc wrote BIP 300 and BIP 301, which define drivechains. His company, LayerTwo Labs, builds the software eCash runs.
Watch the experiment with us.
The first withdrawal vote is running now, and it will take weeks.
We will email you if it pays out, stalls or gets challenged. One short email a week, only when something real happens on eCash. No price talk.
Related reading
Sources
BIP 300: Hashrate Escrows, for how bundles and votes work
LayerTwo Labs enforcer PR 685, merged October 28, 2026, for eCash's 6,048 of 12,096 withdrawal and slot replacement rule
Drivechain FAQ by Paul Sztorc, updated December 2025, for the security model and the user fork backstop
Drivechain: Request for Discussion, the 2017 bitcoin-dev thread with Paul Sztorc and Peter Todd
Greg Slepak's replies in the 2017 bitcoin-dev drivechain follow-up thread
Michael Blowes' drivechain guide, for slot replacement and contested slots, checked against the enforcer source
Block share math: our own calculation from the BIP 300 vote rules
LayerTwo Labs enforcer source at version 1.0.1, read November 13, 2026, for the default withdrawal vote policy
The first withdrawal bundle, pool share and the first fast swap are as reported in the public eCash developer group, November 11 to 13, 2026
Comments