
Testnets: how to practise self-custody without risking anything
Every irreversible system needs somewhere to make mistakes. Aviation has simulators. Surgery has cadavers. Crypto has testnets, and almost nobody uses them.
That's a shame, because the specific thing that makes self-custody frightening — that a mistake is permanent — is exactly the thing a testnet removes. The mechanics are identical. Only the consequences are missing.
What a testnet actually is
A testnet is a fully working blockchain with the same rules as its mainnet counterpart and one deliberate difference: the coins are worthless. They're handed out free by services called faucets, they have no market, and nobody accepts them for anything.
Everything else is real. Real blocks, real confirmations, real fees, real signatures, real congestion, real irreversibility. A transaction you get wrong on testnet is just as unrecoverable as one on mainnet. That's the point — it teaches the same lesson without the same bill.
There's a design detail that makes this safe in a way that isn't obvious. Under BIP-44 and its descendants, every chain has a coin type in its derivation path, and coin type 1 is reserved for all testnets. That means testnet keys are derived from a structurally different branch of your seed than mainnet keys. Same seed, different subtree, different keys, different addresses.
SSP encodes exactly this: a chain is testnet if its SLIP-44 coin type is 1, and every testnet it supports — Bitcoin testnet, Bitcoin signet, Flux testnet, Sepolia, Polygon Amoy, Solana devnet — carries that marker, while no mainnet does. It's a single-property check, which is the right shape for something the whole app keys off. This is one of the more elegant consequences of BIP-48 style derivation: the separation between play money and real money is structural, not a UI setting.
The ones SSP supports, and what each is for
Bitcoin testnet — the long-standing Bitcoin test network. Genuinely useful, with a caveat: its mining is erratic, so you'll sometimes see hours without a block and sometimes hundreds at once. Deep reorgs happen. That unpredictability is itself instructive, but it makes it a poor place to learn what "normal" looks like.
Bitcoin signet — the more civilised option. Blocks are signed by a known set of participants, so they arrive on a predictable schedule and the chain behaves itself. If you want to practise sending and receiving without fighting the network, use signet.
Flux testnet — for Flux, including node-related workflows.
Sepolia — the main Ethereum test network. This is where you'd rehearse anything EVM: contract interactions, token approvals, gas behaviour.
Polygon Amoy — Polygon's test network, useful for the cheaper-L2 experience.
Solana devnet — Solana's development network, and the right place to meet token accounts and rent before they surprise you with real money.
By default SSP hides all of them. Testnets are opt-in and stay hidden until you switch them on, which is the correct default — nobody should stumble into test coins by accident and mistake them for real ones.
What's genuinely worth rehearsing
Reading about self-custody and doing it are different skills. These are the things worth doing at least once where it costs nothing:
A full recovery. The single highest-value rehearsal there is. Write down your seed, wipe the wallet, restore from the seed, confirm your funds reappear. Almost everyone who has lost crypto to a "backup" discovered the backup didn't work at the moment they needed it. Testing recovery is the only way to know, and doing it under real conditions — with real money — is a terrible way to find out you wrote something down wrong.
The two-device signing flow. With a 2-of-2 setup, walk the whole path: build a transaction on one device, approve on the other, watch it confirm. Do it until the rhythm is boring. Boring is what you want when a real payment is in front of you.
Sending, receiving, and change. Send to yourself. Look at the transaction on an explorer. Find the change output and recognise it as yours rather than as something alarming — reading a transaction page is much easier to learn when the stakes are zero.
Fee behaviour. Deliberately underbid and watch a transaction sit unconfirmed. Then replace it. Experiencing a stuck transaction on purpose is far better than meeting your first one during an actual payment.
Contract interactions. Approve a token allowance on Sepolia. Revoke it. Look at what the approval screen actually said. This is the cheapest possible way to build the instinct for what a signing prompt means.
The honest limits
Testnets are a rehearsal space, not a simulation of reality, and they differ in ways that matter:
Testnet coins are free, so your emotions aren't real. The hardest part of self-custody is behavioural — the pause before approving something that looks slightly wrong. You cannot practise fear when nothing is at stake, and that gap is genuine.
Fee markets aren't real. Testnets don't have the sustained congestion of a busy mainnet, so fee estimation behaves differently and the tier labels mean different things.
Testnets get reset. They're infrastructure for developers, and they're occasionally wiped, replaced, or deprecated. Ethereum has retired several over the years. Your testnet balance is not meant to persist.
Faucets are unreliable. Getting test coins can be irritating — rate limits, social-login requirements, empty faucets. This is a rite of passage rather than a real obstacle, but it does put people off.
Some things simply don't exist there. Liquid markets, most real dapps, the specific token you actually hold. You can practise mechanics; you can't always practise your exact situation.
The one thing to take away
If you do nothing else, test your recovery.
Not the seed phrase you wrote down and stored carefully. The restore. Actually take that phrase, put it into a fresh wallet, and confirm it produces the same addresses. It costs an evening. The alternative is finding out it doesn't work at the exact moment you have no other option, which is a category of discovery worth paying almost anything to avoid.
Everything else on this list is useful. That one is the difference between having a backup and believing you have one.


