In Bitcoin We Trust

Who Pays to Mine Bitcoin’s Play Money? Inside the Launch of Testnet5.

Who Pays to Mine Bitcoin’s Play Money? Inside the Launch of Testnet5.

Twenty minutes after Bitcoin’s newest test network produced its first block, a short email appeared on the Bitcoin development mailing list. A contributor signing his name simply “Neill” said two Libbitcoin nodes had joined the network. Their names were Wormer and Carmine. He included links to their block explorers so anyone could see what the machines were seeing.

What I like about that email is how ordinary it sounds. Someone got two computers talking and figured other people might want a look. Behind it was a distinctly odd bargain. Miners had switched on ASICs, the specialized machines used to mine Bitcoin, agreed to pay for the electricity, and pointed them at a chain whose coins are supposed to be given away.

Why bother? Because if you’re about to ship a Bitcoin wallet, you’d much rather discover on a Saturday morning that it sometimes credits the same deposit twice. The expensive way to find out is a Monday morning email from a customer.

The morning Bitcoin supplied its own newspaper headline

On November 9, developer Anthony Towns gave miners a starting signal they could all verify. They would wait for Bitcoin’s main network to reach block 975,744, its next scheduled difficulty adjustment. Testnet5’s first block would have to contain the hash of that Bitcoin block, or of a later one. November 13 was the estimate. The block took until November 14.

The mainnet block carries a timestamp of 05:02:50 UTC. The Testnet5 genesis block carries 05:03:13 UTC, just 23 seconds later on the two headers. Its embedded message identifies mainnet block 975,744 and gives its full hash. Hashing the published 80-byte Testnet5 header reproduces the genesis ID ending in 4a91d497.

In 2009, Satoshi Nakamoto put a headline from The Times about a bank bailout into Bitcoin’s first block. A reader could date the newspaper. Seventeen years later, Bitcoin itself supplied the piece of information nobody could have known the day before. Even the developers who wrote the launch code had to wait for that particular mainnet hash before they could mine the agreed chain.

Towns went one step further. He published an OpenTimestamps proof tying the completed Testnet5 block back to Bitcoin by mainnet block 975,749. The two commitments bracket its creation. The 23-second figure comes from the block headers. The order of events is something a reader can check independently.

For the launch, Towns offered a public solo-mining pool with a zero fee. Anyone with compatible hardware could point it there. A miner with more computing power still had better odds. What the public pool and the unpredictable starting signal bought everyone was a chance to show up on time. The genesis reward itself was deliberately unspendable.

Thirty-six blocks and nothing to spend

The pool published a bundle of its first 36 blocks, starting at zero.The header at height 35 reads 05:18:29 UTC, just over fifteen minutes after genesis. Towns’ plan for this stage called for blocks containing only the transaction that pays the miner. A developer could take that little bundle, feed it to a new node, and check whether the node arrived at the same tip as everybody else. A block explorer could already show a growing chain. A wallet user had a simpler question.

When could somebody actually send a coin?

Bitcoin requires an ordinary mining reward to mature for 100 blocks before a transaction can spend it. The reward earned at height 1 could first be spent in block 101. At height 35, the winning miner could open a wallet, see the reward, and still have nothing available to send a friend. That would be the correct result. It’s an excellent early test for any wallet that uses a single big number at the top of the screen.

Which number did its designer promise the customer they were looking at?

The 20-minute favor that made confirmations hard to trust

Testnet4 inherited a rule that made perfect sense if you pictured a developer sitting alone with a laptop. Suppose a large miner went offline. A long wait for the next block could wreck an afternoon of testing. Under the old rule, a sufficiently late block could qualify for minimum difficulty once its timestamp was more than 20 minutes past its predecessor’s. A small miner had a way to get the chain moving and acquire some coins.

The network had already survived one version of this idea going badly wrong. On Testnet3, Jameson Lopp documented 12,682 blocks in a single day in 2015. Bitcoin’s usual ten-minute target works out to about 144. Racing through block heights also meant racing through the halvings that reduce new coin issuance. By the time developers proposed Testnet4, people were buying and selling coins that newcomers were supposed to get for free.

Testnet4 curbed those block storms. Its particular headache was a succession of short reorganizations. Several easy blocks could compete near the tip. A node would accept one recent history, then switch to a competing one with more accumulated work. A payment you had just watched confirm could lose its confirmation.

One such reversal makes a fine wallet test. The wallet should notice, wind back whatever it credited, and credit the payment exactly once if it confirms again. When reversals keep interrupting ordinary tests, a developer trying to diagnose an unrelated payment failure first has to reconstruct what happened to the chain. The afternoon the shortcut was meant to save disappears anyway.

Pol Espinasa and Fabian Jahr, the authors of BIP95, chose a fresh network. Patching the old one would mean coordinating incompatible rule changes across the miners and software already using it. Starting again let them drop the 20-minute exception altogether and raise the difficulty floor from roughly 1 to 1,048,561. That higher starting point curbs how many blocks ASICs can churn out before the first regular adjustment. It still leaves room for a home miner with suitable hardware.

What does a millionfold increase feel like to one miner? Suppose a machine can try one trillion hashes a second, or one terahash per second. At Testnet5’s starting difficulty, that machine would find a header meeting the target about once every 75 minutes on average. At 100 terahashes per second, the same calculation gives about 45 seconds. Both estimates assume the difficulty stays at its opening level. A lucky miner can find a block on its next try, and an unlucky one can wait much longer. This is why a small machine can make a useful contribution even though a roomful of hardware can still make opening day look unusually fast.

Who pays when the reward won’t cover the electricity?

The first person paying is whoever owns the machine or pays for its power. A rig drawing one kilowatt for a day uses 24 kilowatt-hours. At an illustrative 15 cents per kilowatt-hour, that’s $3.60 a day before cooling. An operator who would otherwise mine Bitcoin gives up a chance at a mainnet payout as well. Two miners can face the same difficulty and have very different reasons to stay plugged in.

For a mining software team, the expense can make sense. A block rejected on mainnet can cost the miner its subsidy and transaction fees. Finding a mistake with a small rig on Testnet5 can be a cheap day at the office. P2Poolv2 already has a use for those coins. Its developers say their system for tracking miners’ contributions and assigning payouts is running on Testnet5. They publish a dashboard where other people can examine the pool. For a team writing payout software, seeing whether the right miner actually gets paid is worth something.

Consider what a pool actually promises a small miner. While everyone waits for a full block, the pool records easier pieces of proof of work called shares. They establish that a machine contributed. P2Poolv2 links these shares in a chain other participants can inspect. When the pool finds a block, a miner can compare the eligible shares attributed to its machine with the payout in the block’s reward transaction. Give two small miners different payout addresses, and each can check that the right wallet received the money.

Here’s a realistic reason to bother. Imagine a rounding error leaves a small contributor short. Bitcoin nodes could accept the block just fine. They check whether the total reward is allowed and whether the block obeys consensus. They don’t have a copy of the pool’s promise to that particular miner. Finding a discrepancy like that on Testnet5 would pay for quite a few days of electricity.

Things look different from the student’s laptop. Practically speaking, the higher floor puts the early mining work in ASIC territory. A student can still run a node, verify blocks, build a wallet, and receive test coins. Their ability to get started now depends more heavily on a willing miner, a colleague, or a faucet, a service that gives small amounts away. A millionfold increase in mining difficulty does precisely nothing to refill an empty faucet.

What if half the miners go home?

José Edil Guimarães de Medeiros raised the harder objection in the original design discussion. What happens when a large contributor points its machines back at Bitcoin? Difficulty normally adjusts every 2,016 blocks. If a lot of computing power leaves just after an adjustment, the remaining miners still have to produce the intervening blocks at the old difficulty. Waiting 20 minutes no longer gets them an easier one. The price of getting rid of the shortcut is that a genuine slowdown can last.

Suppose the network has settled near ten minutes per block. Just after an adjustment, half its mining power leaves and stays away. At the same difficulty, the expected block interval doubles to twenty minutes. A full 2,016-block period at that pace takes roughly four weeks. Actual blocks arrive unevenly, but the distinction matters to anyone planning a launch. A demonstration booked for next Tuesday runs on the calendar. The software waiting for its hundredth confirmation runs on block height. Nobody moving the meeting invitation can make those clocks agree.

That gives Testnet5 two very different constituencies. Miners need a good reason to stay. Everyone else needs the coins to stay easy to get. The protocol can’t set a resale price of zero. Giving coins away freely gives people less reason to buy them. If obtaining practice funds starts to require knowing the right person, a chain can keep producing perfectly valid blocks while failing some of the people it was built for.

Think of a teacher with twenty students. Give each student 0.001 test bitcoin and the lesson needs a total of 0.02, plus whatever fees the distribution transaction pays. At the end of class, the students can return what remains after fees for the next group. One modest supply can teach a lot of people. If the teacher has to spend two evenings persuading a stranger to send that first 0.02, however, the next class may learn on a private simulator. The telling measure of access is how long it takes a newcomer to make a first payment. You won’t find that number in a block header.

Four proposed Bitcoin fixes get a public rehearsal

From block 1, Testnet5 enforces version 1.0.0 of BIP54, the consensus cleanup proposal by Antoine Poinsot and Matt Corallo. Its four objectives sound remote from the everyday business of sending bitcoin until you ask who bears the cost when one goes wrong.

Who sets the clock?

Bitcoin uses block timestamps to adjust mining difficulty. BIP54 tightens the rules around adjustment boundaries so that manipulating the reported times is harder. Honest miners have an interest in a difficulty calculation that reflects the work actually being done.

How long must your node spend checking someone else’s block?

Certain unusual transactions can demand an extraordinary amount of signature-checking work. The cleanup limits that worst case. If you verify Bitcoin on modest hardware, the time your machine can be made to spend matters more than a benchmark from a fast server.

Can a lightweight wallet trust a payment receipt?

Some wallets use a compact proof that a transaction was included in a block. That proof comes from a branching structure called a Merkle tree. One unusual transaction size can be mistaken for an internal part of the tree. BIP54 rules out that ambiguous case, making those proofs easier to rely on.

Can two mining rewards have the same transaction ID?

Early Bitcoin history left behind duplicate coinbase transactions, the special transactions that pay miners. The cleanup ties each new coinbase more tightly to its block height, giving nodes a simpler way to enforce uniqueness as the chain grows. This is the kind of rule a pool building its own blocks needs to get exactly right.

The date on one of those duplicates is worth a second look. On November 14, 2010, mainnet block 91,842 repeated the transaction ID of a mining reward from thirty blocks earlier. Several hours later, on November 15, block 91,880 repeated a different one. Testnet5 started exactly sixteen calendar years after the first of those two duplicates, with rules designed to keep future mining rewards distinct. Anyone syncing a Bitcoin node today still has to be able to follow the early chain people actually used. You can look up both historical exceptions in BIP30.

Those early blocks still shape the code that will run at height 1,983,702. Under the current rules, nodes would then have to resume an explicit duplicate check. That’s over a million blocks beyond the one that anchored Testnet5. On the new chain, a pool can try the proposed fix with the software that builds its next reward transaction. A separate reviewer can check compatibility against Bitcoin’s historical blocks. One has tomorrow’s mining job on screen, the other has records dating back to 2010, and both can help answer whether the cleanup is ready.

Bitcoin Inquisition had already exercised these rules on Signet, where block production on the public network requires an authorized signature. Testnet5 gives independent miners a different opportunity. They can build a candidate block with their own software, mine it, and see whether other implementations accept it. The Bitcoin Core proposal for BIP54 explicitly leaves mainnet activation for a separate decision. A successful rehearsal gives the people making that decision better evidence.

Twelve minutes shy of Taproot’s fifth anniversary

November 14 turns up again in a block familiar to anyone who followed the Taproot upgrade. The mainnet block that activated Taproot, number 709,632, is timestamped 05:15:27 UTC on November 14, 2021. Testnet5’s genesis clock reads 05:03:13 on November 14, 2026. Five years apart, twelve minutes and fourteen seconds shy of the exact anniversary. Towns had chosen a block height for the launch. Bitcoin supplied the coincidence.

In May 2021, a person running Bitcoin Core could read the released Taproot activation code and learn what would happen next. If at least 90 percent of blocks in a qualifying 2,016-block period signaled readiness, the rules would lock in and take effect at the specified height. An exchange maintaining a mainnet node had a concrete block number to plan around once the change locked in. A team testing BIP54 today can likewise plan around Testnet5 block numbers. The published Bitcoin Core proposal leaves the corresponding entry in the exchange’s mainnet calendar blank.

What a Testnet5 miner can contribute to that conversation is more specific. A pool operator can describe which version it ran, which change it made to its reward transaction, and whether an independently operated node accepted the result. If the experience uncovers a bad assumption and sends a proposal back for revision, the electricity has bought something worthwhile. A delay can be a perfectly good outcome for people whose savings would have to live with the rule.

A miner that could stumble over the number three

One of the best examples turned up before launch. Towns reported a quirk in firmware used by Bitaxe miners. Bitcoin Core and the mining software did not read the small numbers at the beginning of a new chain the same way. For block heights 1 through 16, the firmware could misread the height in the transaction that pays the miner. Towns said he appeared to lose some effective mining power as a result and submitted a fix.

Think about how long that assumption could survive on equipment pointed at mainnet. Bitcoin passed block 16 in 2009. A machine can spend its entire working life mining real bitcoin without anyone asking it to read block number three. On Testnet5, block three was timestamped 82 seconds after genesis.

An exchange can use the same public network to run a deposit from a customer wallet, through its indexer, into its balance ledger. A Lightning team can open a channel between two independently run nodes and check that both peers mark it ready at the expected confirmation depth. A wallet team can ask a second implementation what it saw when a payment disappeared from a recent block. Each is testing a handoff between systems whose operators can make different assumptions.

For a repeatable bug, a developer can still use regtest, a private chain where they control when the next block arrives. Signet gives different teams a shared environment with more predictable block production. Testnet5 earns its electricity when the experiment needs a miner outside the room, making decisions the person running the test cannot script.

When only one wallet thinks the Lightning channel is open

Picture a mobile wallet opening an ordinary Lightning channel to a node run by another team. For this test, the receiving peer asks for three confirmations. Suppose the funding transaction lands in block 500. The first confirmation is block 500 itself. Blocks 501 and 502 bring the count to three. There is room for a customer-visible mistake in those three integers. At height 501, the funding transaction is plainly visible on a block explorer. If the app enables its “Pay” button too early, a customer can see a confirmed deposit and a failed payment on the same screen. Guess which one they email about.

Now let the phone go to sleep after the first confirmation. The other operator stays online as the next two blocks arrive. When the phone wakes up, somebody can try an actual payment and observe whether it goes through. A channel needs its funding on the chain and an exchange of readiness messages between the peers. The explorer can help settle the first question. Only the two Lightning implementations can tell their developers what happened to the second. A screenshot from the phone would have left an engineer guessing.

When those teams compare notes, they have a block height, a record of what each peer received, and the words the customer saw. The bug report can be as specific as “After waking up, the phone showed the third confirmation but never sent its readiness message.” That’s a hypothetical fault, but imagine how much faster it would be to fix than a ticket that simply said: “Lightning didn’t work.”

What changes for the bitcoin in your wallet?

Testnet5 has its own ledger. A Testnet4 balance belongs to Testnet4 even if the same test address looks valid on both networks. An exchange or wallet team moving its tests across needs fresh test funds and back-end services pointed at the new chain. As a bitcoin holder, you can keep using your wallet as usual. Your fees still depend on demand for space in mainnet blocks. A successful test can show that a wallet knows how to raise a fee. The price of getting a real transaction confirmed on a busy Friday is a different question.

Somebody gave the network a name even a tired support engineer can remember. The four bytes identifying Testnet5 network messages spell FIVE. The address a tester pastes into a wallet, on the other hand, can look exactly like an address on Testnet4. On the wire, the software has a way to identify its conversation. On the screen, a familiar-looking address by itself gives the tester very little to go on.

Say a company moves its test wallet to the new chain on Friday and leaves its support dashboard connected to the old one. A developer sends test funds, sees a confirmation, and opens a ticket when support reports a zero balance. Both screens could be accurately reporting the chains they follow. Asking each back end for the hash of block zero settles which ledger it is looking at. That little diagnostic can save a team hours of investigating a signing bug that never existed.

For developers, launch day already brought more than a block hash. Libbitcoin reported two public nodes enforcing BIP54 and shared their explorers. Towns submitted the mined parameters to replace the placeholders in BIP95. Bitcoin Core’s Testnet5 integration was still under review on November 15. Day-to-day use also depends on reliable peer discovery, including DNS seeds that help a fresh node find its first peers, and faucets that put coins in strangers’ hands.

The milestone I’d watch comes on a much less interesting morning. A developer nobody on the mailing list knows installs a Lightning wallet, connects to a peer, gets a few free coins, and opens a channel with someone running different software. Something goes wrong. They can both look at the same blocks, figure out whose code needs fixing, and try again before lunch.

That developer may never learn who paid for the electricity. The person whose real deposit goes through after the fix ships may never hear of Testnet5. That’s a pretty good return on a coin you can give away.


In Bitcoin We Trust Newsletter

Did you enjoy this article?


ssaurel
ssaurel Verified Member

Entrepreneur / Developer / Blogger / Author.


In Bitcoin We Trust
In Bitcoin We Trust

In Bitcoin We Trust is a place where Bitcoin believers share their ideas about the upcoming revolution. Blockchain and cryptocurrencies are also covered in this publication.

Publish0x Publish0x

Reward the author with $0.01 in crypto, and earn yourself as you read!

20% to author / 80% to me.
Rewards are FREE. Publish0x pays them, not you.

Page not displaying correctly?