I recently came across a pretty interesting incident. A former Solana developer reportedly used 11 pseudonyms to stack protocols on Solana and inflate on-chain data, but it was eventually exposed. This incident highlights a question: what exactly does the TVL we commonly look at actually mean? Can it really reflect a project’s strength?

Let’s start with the basic definition of TVL. Total Value Locked is the scale of funds managed by a DeFi project—bigger numbers make it look more impressive. But there’s a catch: TVL itself is a static metric; it only shows the current situation, and no one can be sure what will happen in the future. With how volatile the crypto market is, if a project’s incentives change or the token price drops, its TVL can fall sharply. What’s even more painful is that TVL means completely different things for different types of projects, and on the layer-one blockchain side, double counting can also easily happen.

In DEXs, TVL directly corresponds to liquidity. Uniswap has no mining and no staking—its TVL is simply the liquidity of trading pairs. But projects like Curve and Sushi let you stake governance tokens to earn fees; those staked tokens are listed separately under the Staking category by data providers.

Lending protocols are even more complicated. Compound’s TVL is the difference between deposits and loans—i.e., total deposits minus total borrowings—showing how much liquidity is still available in the protocol. On top of that, Aave can stake AAVE and LP tokens to earn rewards, and this portion is classified separately as well. MakerDAO is different because it lends out DAI issued by the protocol, which does not affect locked funds—so its TVL directly equals total deposits.

What’s most likely to create “bubble” is projects built on top of other protocols. For example, Convex Finance is a yield aggregator built on Curve. Users’ funds are actually being mined in Curve, but they end up being counted twice. Back when Solana’s TVL was only $10.5 billion, Saber and Sunny accounted for $7.5 billion—this is the reason.

Liquidity staking protocols are also easy to trip over. stETH issued by Lido is used as collateral in Aave and also used to provide liquidity on Curve. These use cases are already included in other protocols’ TVL, and counting them again at the layer-one level results in double counting. Last August, DeFi Llama changed its calculation method so that these overlapping parts were no longer included in chain TVL. As a result, the data for many chains dropped noticeably, but it also became more accurate.

“Middleware” tools like Instadapp can run into the same issue: they help users manage funds that are stored entirely in protocols such as Aave and Compound, and then counting its TVL toward the total blockchain amount again leads to double counting.

In the end, TVL is a metric that’s prone to misunderstanding, but it’s not completely useless. At the application layer, it can be used to compare project size horizontally. At the layer-one blockchain level, previous numbers were indeed inflated, and the recent adjustments have made the data cleaner. Understanding the real meaning of TVL in different scenarios is far more sensible than blindly chasing bigger numbers.