[Beginner] — how people actually lose money, which is mostly not smart contract exploits.
Why this note is in a technical folder
Because the losses are overwhelmingly not technical. Reading about reentrancy while getting phished is the field’s characteristic failure, and the properties that make blockchains useful — irreversibility, pseudonymity, permissionless deployment, no recourse — are exactly the properties that make fraud efficient.
This is not a bug in crypto culture. It’s a structural consequence of the design, and it should be taught as part of the design.
Why fraud works so well here
| Property | Why it’s good | Why fraud loves it |
|---|---|---|
| Irreversible | No chargebacks, final settlement | No recovery, ever |
| Permissionless | Anyone can build | Anyone can deploy a scam for $5 |
| Pseudonymous | Privacy | Hard to attribute, easy to vanish |
| No intermediary | No gatekeepers | No fraud department, no one to call |
| Composable | Build on anything | Malicious contracts compose too |
| Global | Anyone can participate | Jurisdictional arbitrage |
The catalogue
Phishing — the largest single cause of individual loss.
A fake site, a fake airdrop, a fake support agent, a poisoned search ad. The victim connects a wallet and signs something. The signature is often setApprovalForAll or a permit — it doesn’t move funds now, so nothing looks wrong, and the wallet is drained later.
“Wallet drainer” kits are sold as a service. This is an industry with products and customer support.
Defences: never sign anything you can’t read → signatures; bookmark real sites and never use search ads; a hardware wallet for anything meaningful; a separate “hot” wallet for interacting with new protocols; and revoke old approvals periodically (revoke.cash).
Address poisoning. The attacker sends a zero-value transfer from an address whose first and last characters match one you use. Later you copy the address from your transaction history — and send to theirs. Always verify the middle characters, and use an address book or ENS.
Rug pulls. The team takes the money. Variants: the liquidity pool is drained (“soft rug”); a hidden mint function creates unlimited tokens; a blacklist prevents everyone but the deployer from selling (“honeypot” — you can buy and never sell); or the contract is upgraded into something that drains it → upgradeability.
Deploying a token costs a few dollars and requires nobody’s permission, so this is manufactured at industrial scale → tokens.
Pump and dump. Coordinated buying, promotion, then selling into the retail bid. Ordinary securities fraud, executed on assets with no securities regulation and no disclosure requirements. Now largely automated on memecoin launchpads.
Ponzi and “yield” schemes. Returns paid from new deposits. Any protocol paying high fixed yields with no explicable source is paying you with someone else’s deposit — Anchor’s 20% on UST was the largest example, and it worked exactly as long as inflows continued.
“Pig butchering.” Long-form romance and investment fraud, often run from trafficked labour compounds in Southeast Asia — a serious organised-crime and human-rights problem, not just a financial one. Victims are groomed over months and shown a fake trading platform with fabricated gains. The largest category by aggregate dollar value in recent years.
Fake support. Nobody legitimate will ever ask for your seed phrase. Ever. Anyone who does is stealing from you, without exception.
Exchange failure. FTX (Nov 2022) was not a crypto failure — it was customer funds being taken and used, which is old-fashioned fraud that happened to involve crypto. Mt. Gox, Celsius and QuadrigaCX are variations. “Not your keys, not your coins” is the accumulated lesson, weighed against the fact that self-custody has its own irreversible failure modes.
Reading a token before touching it
Mechanical checks, most of which take a minute:
- Is the source verified on Etherscan? Unverified is a hard no
- Is there a
mintfunction, and who can call it? - Is there a blacklist, a pause, a transfer fee, or a max-transaction limit?
- Is it upgradeable? Who holds the key? Is there a timelock?
- Is the liquidity locked, and for how long? Unlocked liquidity can vanish in one transaction
- How is supply distributed? A few wallets holding most of it is a countdown
- Does a linked audit exist, and does its commit hash match what’s deployed? → audits
Token sniffers (Token Sniffer, GoPlus) automate much of this. They catch known patterns and miss novel ones — useful as a filter, not as approval.
The base rates
Worth internalising, because they’re the actual expected value:
- The overwhelming majority of new tokens go to zero. Not a risk — the base rate
- Most airdrop farming is unprofitable after gas and time
- Retail leveraged trading loses money in aggregate, with more certainty here than in traditional markets, because of MEV, funding rates and 24/7 liquidations → MEV
- Guaranteed yield is a marketing claim, not a financial instrument
For builders
You inherit a duty here, and most projects discharge it badly:
- Publish and verify your source. Unverified code is indistinguishable from a scam
- Lock liquidity and say where. Timelock privileged functions and publish the delay
- Explain the risks plainly, including what happens if you disappear
- Never ask for a seed phrase, and say in your UI that you never will
- Make signature requests legible — EIP-712, with a plain-language summary in your own interface before the wallet prompt
- Don’t promise yields you can’t source. If you can’t explain where the return comes from, you’re running a Ponzi whether you meant to or not
Key insight
The properties that make blockchains valuable are the same properties that make fraud efficient, and you cannot keep one without the other. Irreversibility is why settlement is final and why theft is permanent. Permissionlessness is why anyone can build and why anyone can scam. Security here is mostly operational discipline — what you sign, what you approve, where your keys live — not knowledge of vulnerability classes.
Related
- wallets — signature phishing
- tokens — why a token means nothing by itself
- regulation — the response
- social engineering — the same techniques, older
Software engineering
Infrastructure & security
Computer science
Hardware & engineering