Merged mining (AuxPoW): one electric bill, two coins

Normal mining: your machine grinds hashes for one chain, and a valid hash becomes a block on that chain. Merged mining (auxiliary proof-of-work, AuxPoW): the same grinding simultaneously produces blocks on a second chain. Same machines, same electricity, same hashes — two paychecks. For the second chain it's transformative: it inherits the first chain's hashrate without paying for any of it.

The terminology: the chain whose blocks are actually being constructed is the parent; the chain riding along is the aux (auxiliary). An aux chain that enables merged mining is effectively renting the parent's security budget for free.

The trick is a commitment. When your pool builds a parent-chain block template, it embeds a hash of the aux chain's block header inside the parent's coinbase transaction — the special first transaction of every block. Your miner then grinds the parent header as usual. If the result clears the parent's difficulty, it's a parent block. If it's too weak for the parent but still clears the *aux* chain's (lower) difficulty, the aux chain can verify — from the embedded commitment — that this hash genuinely committed to its header, and accept it as an aux block.

In practice you never do this alone: pools do it for you. You point your miner at the pool with a merged-mining port; the pool merges the chains, credits you for both. For you, merged mining looks like one stratum connection that pays two coins.

• 2011 — Namecoin↔Bitcoin: the original deployment. Namecoin wanted Bitcoin's hashrate without competing for it; merged mining was invented so BTC miners could secure NMC in passing.
• 2014 — Dogecoin↔Litecoin: the version most miners meet first. When Dogecoin's hashrate collapsed after its first halvings, it enabled AuxPoW on Litecoin — and practically all DOGE hashrate has been borrowed LTC hash ever since.
• Since: dozens of Scrypt-family chains ride LTC/DOGE the same way, and the pattern repeats in other algorithm families wherever a big parent exists.

If you already mine the parent, the aux side is a free lottery ticket: zero extra hardware, zero extra electricity — you're selling the same hash twice. A rig earning its rent on the parent collects whatever the aux coin adds on top. If the aux coin ever pumps, that free ticket pays; if it dies, you lost nothing but the effort of configuring a second payout address.

What it is *not*: a reason to start mining a parent you otherwise wouldn't. The aux income rides on a parent that must stand on its own economics. Work the parent's numbers first; treat the aux as the bonus line, never the base.

• Inherited hashrate ≠ inherited demand. The aux chain gets the parent's *security*, not the parent's buyers. A coin can sit on terabytes of borrowed hashrate and still have no market.
• Difficulty tells the truth. If an aux chain claims huge network hashrate but its difficulty is trivial, the "hashrate" is merge-hash that shows up only when parent pools bother to merge it — it can vanish as easily as it came.
• Parent dependency cuts both ways. If the parent changes policy (pools drop the merge, fees make it unattractive), the aux's security walks away overnight. You are a guest of the parent's incentives.
• Double stale pressure. Two chains means two streams of job updates; stale-rate management (see the pool mechanics lesson) matters more, not less.

“Which side of the + is it on?”

the only question that matters when evaluating a merge coin

A deep-liquidity parent merging a thin aux is the good geometry: aux gets free security, miners get a free second coin. A thin parent claiming merge support it doesn't have is marketing. Before touching any merge coin, find out who actually parents it, how much of its hashrate is genuinely merged, and whether the pools doing the merging publish their numbers.

The tracked set includes both geometries: SOQ is an aux riding Litecoin/Dogecoin-family hash from genesis, while PRL is a parent whose pools pay auxiliary coins on top of PRL rewards. Project pages show each coin's live network and pool hashrate — enough to see merge-hash inheritance in the numbers.