The Messy Middle: What Actually Happens if BIP110 Splits the Chain

• The Dates
• The Clean Scenarios
• A Note on Difficulty
• The Switching Threshold
• The Dumper’s Dilemma
• The Exchange Enigma
• The Restricted Holders
• Peer-to-peer Trading
• A Note on Replay Protection
• Incentives + Frictions + Convictions = Complexity

What’s going to happen on August 8th?

According to the BIP110 proponents, miners will switch, the soft fork will activate, and they’ll have essentially won. According to the BIP110 critics, no miners will mine BIP110 blocks, the chain will continue without activating BIP110, perhaps there’s a hard fork in there by the BIP110 proponents, and they’ll have essentially won, with maybe some profit from fork coin dumping to boot.

What neither side is really talking about is what happens if reality lands somewhere in between: a split, two chains with no hard fork and no clean resolution. I think that’s the likeliest outcome, and gaming it out is a good way to understand both what the incentives are and just how complicated everything gets.

Before I start, one framing note that I want up front rather than buried at the end. Everything below is an economic skeleton. It tells you the fee level at which switching hashrate becomes economically rational, what the feedback loops look like, and where the pressure points sit. What it cannot tell you is who is actually positioned to act when those thresholds get hit. Miners are not the profit maximalizing robots that people pretend they are. They have opinions about whether BIP110 is good for Bitcoin, and more importantly they have contracts, hedges, auditors, and shareholders. The incentives establish the bounds. The frictions and convictions determine what happens inside them. I’ll come back to this at the end, because the interaction between the incentives and all those other things is complex and pretty much impossible to predict.

Two pieces of vocabulary, since I’m going to use them constantly. Wipeout means a reorg where one chain overtakes the other and erases its history. A wipeout is possible here because BIP110 is a soft fork, so BIP110 blocks are valid under legacy rules and legacy nodes will follow them if that chain ever gets heaviest. URSF means User Resisted Soft Fork. These are nodes that explicitly reject BIP110-signaling blocks, which turns a soft fork into a permanent split.

§The Dates

August 8th isn’t one event, it’s the first of three

• Block 961,632, roughly August 8. Mandatory signaling begins. Enforcing nodes reject any block that doesn’t signal. This is when the chains can first diverge.

• Block 963,648. Lock-in, guaranteed by the mandatory signaling mechanism.

• Block 965,664, roughly September 1. The rules actually activate. This is when the data restrictions bind.

So there’s a four-week window where the chains split over the signaling bit while the actual BIP110 restrictions aren’t yet in force. During that window both chains accept identical transactions. They’re only distinguished by which block headers they’ll build on. That window matters for the replay discussion below.

And while we’re on the calendar: Paul Sztorc’s eCash hard fork is scheduled for block 964,000, around August 21, but that’s a hard fork, so it’ll split cleanly, much like the hard forks from 2017-2018. The only real question with respect to eCash is which chain they’ll actually fork from if there’s a split. This may be a consideration for those selling BIP110 fork coins should that be available.

§The Clean Scenarios

BIP110 chain gets 0% hashrate. This scenario is what a lot of critics allege where zero blocks are mined and the BIP110 chain aborts before it can come to pass. No mining on BIP110 means no messy split and Bitcoin chugs along with the least disruption.

BIP110 chain gets 50%+ hashrate. This scenario is what proponents allege. The BIP110-enforcing chain outpaces the non-enforcing chain, wipes it out quickly if any blocks are mined that don’t enforce BIP110 signaling, and there’s no permanent split. Bitcoin chugs along after some possible disruption in the form of a wipeout.

BIP110 chain hard forks. The two main possibilities are a proof-of-work change or the addition of replay protection. Either permanently splits the two chains. Two clean chains, straightforward trading between them as in 2017-2018 with the many hard forks of Bitcoin.

URSF gets mass adoption. Similar to the above where two permanent chains that can never merge, but messier. This is because nodes that are neither URSF nor BIP110-enforcing may flip between chains, possibly repeatedly, but clean enough to possibly enable trading.

I think all four are unlikely.

Signaling has been volatile and low. Cumulative signaling since May sits well under 1%; but recent days have had 4% of blocks signaling. Take whichever number you like, but take the range seriously, and note that even a few blocks means somebody will get some trades in. At the upper end of that range you get roughly six blocks a day on the BIP110 chain, which is enough to rule out scenario #1. The same range is nowhere near enough for scenario #2. 

Scenario #3, though egged on hard by the anti-BIP110 side, still seems unlikely, at least for a while because of how hard forks have done in the past. That said, I count this as the most likely among these four because it only takes one actor to make this change, whereas the other scenarios require mass mobilization.

Scenario #4 is unlikely given current URSF node adoption.

So we’re left with somewhere between 0 and 50% of hashpower on BIP110, with no wipeout protection on the non-BIP110 side. Some blocks on BIP110, more on the other chain, two chains coexisting.

We have never had this in Bitcoin. A minority chain holding the specter of a wipeout over the majority chain, indefinitely, as long as blocks keep getting produced on the minority chain.

§A Note on Difficulty

The BIP110 chain inherits the last difficulty and holds it for 2016 blocks as it starts on the mandatory signaling period. At 4% of network hashrate that’s about 6 blocks a day, which means the first retarget arrives in roughly a year without any hash rate changes. Of course, the hash rate is very likely to be volatile and it’s not easy to know even which direction the hash rate will go for the reasons listed below.

It’s entirely possible that the one-year soft fork won’t get to enforce the ruleset even once without a hard fork because of how slow the blocks are.

§The Switching Threshold

The minority chain will have higher fees than the majority chain, simply because block space is more scarce. Lower hash rate and less block space along with some demand for that space in the form of people wanting to trade it means higher fees. At some point fees may reach a level where mining that chain becomes rational which depends on the price ratio between the two coins.

The missing input is that price ratio, and with no liquid futures market we have very little to go on. The only market I know of is Predyx, which is a slightly different bet than a real futures market and has almost no liquidity. That market, extrapolated to futures, currently trades around 10 BIP110 coins to 1 non-BIP110 coin.

At 10:1, you need roughly 28 BTC in fees in a BIP110 block to make switching rational on purely economic grounds. That’s about 2,800 sats/vbyte, though in a chain whose coin is worth a tenth as much, it looks 10x more expensive to the person paying. At 100:1, you need around 309 BTC, or 30,900 sats/vbyte, which again is roughly the same in majority-chain terms once you adjust for the price ratio. At 5:1, it’s 12.3 BTC, or 1,230 sats/vbyte. The pattern holds such that in terms of majority-chain purchasing power (roughly 300 sats/vbyte in majority chain terms), the switching threshold is roughly constant. It’s the nominal fee on the minority chain that explodes as the coin gets cheaper. But again, the price itself, if it exists in any universal way (which for reasons listed below may not exist) is likely to be volatile.

One subtle consequence is that the higher the price ratio, the more of a whale you have to be to dump the minority chain, just because the loss in fees make the trade much worse for smaller UTXOs. Heck, even if you’re a whale but have lots of small UTXOs, it may not be economical to dump. But that’s not even the end of the incentives at play here.

§The Dumper’s Dilemma

Everyone who wants to trade the split tokens has to get into the few blocks BIP110 miners are producing. And yes, there may be a lot of people who want to dump. The question is whether they’re willing to pay the fees to get in.

And here’s the irony: the fees they pay are exactly what motivates more miners to come to the BIP110 chain, which increases wipeout risk on the chain they’re betting will never be wiped out.

If BIP110 is as bad as its opponents say, the moral thing to do is to not dump BIP110 coins (adding fees and increasing the hash rate), while the economically rational thing to do is to dump them as fast as possible. I don’t know how that tension resolves. I do know that it would be a strange position for the anti-BIP110 side to be in, which is applying moral pressure on whales to not act in their economic self-interest. The dumpers will want more hash rate so their transactions get confirmed, but not so much that it destroys their trade.

There’s a lock-in effect on the miner side too. Fees earned on the BIP110 chain require 100 blocks of coinbase maturity before they can be spent. At 2 blocks a day, that’s a month and a half. At 6 blocks a day, that’s 16 days. A miner who switches for a fee spike can’t cash out for weeks, which gives them a strong reason to keep mining the chain rather than defect back, if only to mature their coinbase reward faster. Switching hashrate is stickier than it looks, in both directions and this may require more fees to offset or less depending on the miner’s calculations.

§The Exchange Enigma

Surely more trades mean more commissions for exchanges, so offering the split trade is a no-brainer? Not exactly.

Wipeout risk makes splitting terrifying to intermediate. As an exchange, you’d have to put up a warning explaining that if you split coins, you’ll always have the BIP110 coin, but the non-BIP110 coin is subject to wipeout. In this case withdrawals made after August 8th may be reversed. Worse, deposits may be reversed too. And that risk exists whether or not you offer the trade. You cannot opt out of it by declining the product.

An exchange might decide wipeout risk is effectively zero and operate as if two chains are permanent. That’s an option, but a less riskier path is to credit a URSF chain, so the split is clean and the accounting is unambiguous. But then you get scenarios where the longest chain is neither the BIP110 chain nor the URSF chain, and now there are three chains, two of which can wipe out the third. See what I mean when I say this can get really complicated?

The settlement math doesn’t help either. Other than pre-August 8th coins split at the exchange more or less artificially, every other BIP110 coin takes many hours, perhaps days to arrive. At 4% hashrate, 3 confirmations is around 12 hours on a chain with elevated reorg risk. Six confirmations is a whole day. At lower minority chain hash rates, it’s much longer. Every exchange offering this trade has glacial turnover on one side of the book, meaning the supply side of the BIP110 coins may be severely restricted, creating weird trading dynamics and probably weird prices, where two exchanges will have very different prices of the same coin.

§The Restricted Holders

A large fraction of Bitcoin now sits in ETFs and institutional custody, and those filings generally disclaim rights to forked assets. IBIT alone held around $45 billion worth as of early July.

A meaningful share of supply being held by such entities may not be able to claim the minority coin, nor sell it, and in some cases not even acknowledge it exists. Which means the marginal seller of BIP110 coins is wildly unrepresentative of the marginal holder of Bitcoin. Whatever price the split coin finds, it will be set by the subset of holders who are small enough and unencumbered enough to act, trading in venues which may have little depth. Every number in the fee analysis above depends on a price that will be discovered by a relatively small sample, which you can argue is good (plebs get to determine this!) or bad (not everyone is acting according to natural market forces!). I’m not saying it’s either, but it’s a reality that needs to be acknowledged.

§Peer-to-peer Trading

If exchanges mostly won’t touch this, which very well might happen, trading goes peer-to-peer. Maybe people start dumping BIP110 tokens on-chain through something like the smart contract Rob Hamilton designed.

Those transactions will have to pay hefty fees on the BIP110 side, and those fees motivate the BIP110 side to add hashrate. At a certain point, the BIP110 side can simply pay miners to switch for a while, possibly even just long enough for one wipeout, which would at minimum be very disruptive.

But there’s a worse version, and I want to spell it out step by step because it illustrates just how complicated this can get with just one actor pushing things in a particular direction.

Suppose a dumper wants out of 1,000 BIP110 coins, and a miner is willing to buy at 10:1.

1. The dumper sends 1,000 coins to the miner on the BIP110 chain.

2. The miner sends 100 coins to the dumper on the non-BIP110 chain.

3. The miner now holds 1,000 BIP110 coins and is out 100 non-BIP110 coins. Trade completes, both sides are satisfied.

4. The miner then spends those 1,000 BIP110 coins on fees on the BIP110 chain, or adding hashrate themselves or recruiting hash rate of other miners.

5. If that pushes the BIP110 chain past the non-BIP110 chain in accumulated work, non-BIP110 nodes reorg onto it.

6. The non-BIP110 chain’s history is wiped out, including step 2. The miner’s 100 coins revert to the miner.

The miner ends up with the 1,000 BIP110 coins, the 100 non-BIP110 coins, and if they spent less than 1000 BIP110 coins on the hash rate recruiting, they profit. The dumper ends up with nothing, neither the 100 non-BIP110 coins from the trade which were wiped out, nor the 1000 BIP110 coins which they traded away. The capital required to attempt the wipeout is refunded by the wipeout itself. The scenario is also reinforcing. The more miners that do this, the less risk it is for the next miner to do the same, and it only takes one wipeout for every such miner on the BIP110 side to profit.

That’s just one scenario with a single actor. There are many others that include hedging, futures, cartels and the like which are much more complicated and are similarly economically rational. What damage they cause and how the ecosystem deals with it is an even more difficult question to answer, and frankly impossible to predict.

§A Note on Replay Protection

There’s been a lot of noise about replay protection. I don’t think it’s a practical issue, though I want to correct something first, because I had this wrong.

The obvious trick is to add an 84-byte OP_RETURN, which exceeds BIP110’s 83-byte cap and is therefore only valid on the non-BIP110 chain. But that doesn’t work until block 965,664, which is after when the fork actually happens. For 4032 blocks on BIP110, 84-byte OP_RETURNs are valid, but afterwards are not. It’s the block header, specifically the signaling bit that is required on the BIP110 chain, but not required on the non-BIP110 chain.

But the fee disparity does the job anyway, and does it with economic incentives.

Broadcast a transaction at a low fee rate (under 5 sats/vbyte). On the non-BIP110 chain, with roughly 95%+ of the hashrate and abundant block space, the transaction is confirmed in 20-30 minutes at the most. On the BIP110 chain, where fees are likely running in the thousands of sats per vbyte, a low-fee transaction simply never gets mined. It sits in the mempool. Then you spend the same inputs on the BIP110 chain, to a different destination, at a fee that BIP110 miners will actually take. Those transactions conflict, and BIP110 miners will mine the one that pays them. 

I want to be honest that this is a race rather than cryptographic protection. Your low-fee transaction remains technically valid on the BIP110 chain and could theoretically get mined. But with a hashrate ratio somewhere likely north of 20:1 and a fee ratio potentially in the hundreds, the race isn’t close. And if by some miracle the low-fee transaction gets mined on the BIP110 chain, then the would-be coin splitter can always try again.

My read is that the demand for replay protection is mostly people who want a clean hard fork instead of the messiness of a contested soft fork. That’s an understandable, not to mention more profitable and less risky, preference. It’s just not a technical necessity, and both sides know how a hard fork has played out before, which is probably why replay protection won’t happen. I’d note the push for a PoW change is being made by the BIP110 critics and won’t happen, at least right away, for the same reason.

§Incentives + Frictions + Convictions = Complexity

So back to where I started.

Everything above assumes miners respond to economic incentives. They partly will. But the frictions are asymmetric, and that asymmetry is as informative as the incentive itself.

A publicly traded miner likely cannot redirect hashrate to the BIP110 chain no matter how good the fees look. They have hosting agreements, hashprice hedges, lender covenants, an audit committee, and disclosure obligations. Directing hashrate to a chain whose coin has no liquid market, no institutional custody, a 100-block lockup, and an unresolved legal characterization is not a decision a CFO makes on a spreadsheet. Large institutions will have to go through a long discussion period along with legal review before they’d even consider.

A small private operator can switch in an afternoon. For example, the operators already signaling such as Barefoot Mining face essentially no friction at all.

So the fee incentive doesn’t recruit uniformly across the hashrate distribution. It recruits the already-sympathetic and the unencumbered, which is largely the same population that would switch on conviction anyway. That means a move from 4% to 8 or 12% is entirely plausible with zero major pools participating. But getting past 50% requires Foundry, which controls about a third of the hash rate, or Antpool and those big players are the most friction-bound actors on the network. I’ll also note F2Pool has already refused outright to mine BIP110 blocks. 

There are institutional reasons for the hash rate on the BIP110 side to have significant difficulty getting over 50%. The fee mechanism has a soft ceiling built into it, and the ceiling sits right in the middle of the range where nothing resolves.

That’s the conclusion. Incentives matter and they tell you where the pressure will build, but they push hardest on the actors least able to move, and they push on a population of small operators whose behavior is driven at least as much by what they think Bitcoin is for.

Do you understand now why I say this gets complicated fast? 

How much moral pressure will BIP110 opponents put on whales not to pay high fees? How many small miners switch when fees spike? How do they price coins they can’t touch for 100 blocks on a chain that won’t retarget difficulty for who knows how long? How does an exchange price wipeout risk it can’t opt out of?

There are enough of these questions about human behavior that nobody knows how this plays out. Anyone telling you otherwise is modeling one of the four clean scenarios and hoping reality cooperates.