Imagine a rigged carnival game — the barker shouts about prizes, the crowd rushes in, but the claw machine has been programmed so the grip never closes. You can insert coins all day; nothing comes out. That's what's happening on decentralized exchanges, and this paper is the first to document the mechanism at scale. The committed claim: what Telegram channels advertise as pump-and-dump schemes on DEXs are not pump-and-dumps at all. They are a structurally distinct fraud the authors call "Bait-and-Trap." Organizers deploy honeypot tokens — smart contracts engineered so that buy transactions succeed but sell transactions revert. Subscribers rush in thinking they're riding a coordinated price spike, but they can never exit. The organizer is the only party who can withdraw liquidity. This isn't price manipulation with uncertain outcomes; it's a deterministic theft mechanism embedded in code. The numbers are stark. Across 3,677 events collected from 83 Telegram channels on BNB Smart Chain and Ethereum, organizers profited in 99.3% of cases, extracting over $7 million total. Compare that to classical CEX pump-and-dump schemes where organizers face meaningful execution risk — subscribers can sell before the dump, bots front-run the signal, and liquidity can evaporate unpredictably. Bait-and-Trap eliminates all of that uncertainty by making selling impossible at the contract level. The "pump" is theater; the trap is architecture. Architecturally, this sits at the intersection of smart-contract analysis and social-network fraud detection. The authors combine on-chain transaction monitoring (tracking token deployments, buy/sell success rates, liquidity events) with off-chain Telegram channel surveillance (scraping pump signals, timing announcements, mapping organizer accounts). The detection tool they build uses transaction simulation — essentially dry-running a sell transaction against the token's live contract state before any real money moves. If the simulated sell reverts, the token is flagged as a honeypot. This is the same class of pre-flight verification that MEV searchers use, repurposed as a consumer protection tool. The integrity picture is solid for an empirical security paper. The dataset is large (3,677 events), collected from a real adversarial environment rather than a synthetic benchmark, and the profit figures come from on-chain accounting rather than self-reported data. The main limitation is observational — the authors cannot verify what fraction of all Bait-and-Trap operations they captured, only what flowed through the 83 channels they monitored. The detection tool is validated against the collected dataset but lacks a pre-registered evaluation protocol or independent third-party audit. The milestone question is practical: when does this detection capability become a default feature in wallets and DEX interfaces? The transaction-simulation approach is lightweight enough to run pre-trade, but adoption depends on wallet developers integrating it. The real unlock is not academic — it's whether MetaMask, Trust Wallet, or Uniswap's frontend ships honeypot detection as a default warning. Until then, the tool is a proof-of-concept that protects researchers and sophisticated users but not the Telegram subscribers who most need it. The obvious next experiment the authors didn't run: adversarial robustness testing. A honeypot developer who knows about transaction-simulation detection could write a contract that behaves normally during simulation but activates the sell-block only when specific on-chain conditions are met (e.g., after a certain block number, or when the caller isn't in a simulation context). The authors likely know this cat-and-mouse dynamic exists but chose to establish the baseline detection rate first. Expect the adversarial escalation paper within 12-18 months — either from this team or from the honeypot developers themselves.
