Solana’s block time has dropped again by 17%.

This report from Decrypt dated September 19 has a very straightforward headline—your heartbeat speeds up, but the extra speed gives you “freshness,” not capacity.

These two words are very different, and many people don’t distinguish them.

Freshness refers to how quickly state updates reach you—the prices, balances, and on-chain state you see are closer to what’s happening now. Capacity refers to how many transactions, and how much computation, can fit into each time window. The former is the experience; the latter is the ceiling.

Solana has been following the same path for the past few years: squeezing block time from 800 milliseconds down to 400 milliseconds, and then pushing it further down. After each reduction, the official line is always “it’s faster.”

The problem is that as we get closer to the end, the marginal improvement perceived by users keeps getting smaller.

When you place an order on a DEX, the difference between 400 milliseconds and 200 milliseconds is basically invisible to the naked eye. What truly holds you back is the transaction failure rate when the network is congested, and during peak times, priority fees spiking to a few dollars per transaction.

These are capacity issues, not speed issues.

I looked through market data from the same period, and tokens in the Solana ecosystem haven’t broken out into an independent trend. On CoinGecko’s trend list, Ethena (ENA) is up 13.7% over 24 hours, ranking 47 by market cap; Avalanche (AVAX) is up 14.5%, ranking 28; Zama (ZAMA) is up 32.5%, ranking 183.

What’s going up are other narratives, and it has little to do with Solana speeding up.

But a few numbers are quite interesting. On Binance spot, CELR is up 119.6% over 24 hours, with trading volume of $0.2 billion; ONE is up 70.9%, volume $0.8 billion; and G is up 66.0%, volume $0.5 billion.

These are old coins moving, and the trading volumes aren’t that large—typical capital rotation, not a change in fundamentals.

USDC’s 24-hour trading volume reached $1.8 billion, up or down 0.0%. BTC’s trading volume is $1.0 billion, down 0.7%. Money is staying in stablecoins, not rushing into risky assets.

Back to Solana.

At this stage of the performance narrative, the diminishing marginal returns are obvious. Cutting block times by another 17%—how many new users, new scenarios, and new capital would that actually bring? I’m skeptical.

What you really should focus on is the capacity metrics: the per-block upper limit of compute units, the actual utilization rate of parallel execution, and the transaction failure rate during peak periods. Solana’s official documentation has these data, but very few people bring them up.

There’s also a comparison. On September 18, Zcash announced its November upgrade NU7, reducing block time to 25 seconds and, starting in 2031, setting aside at least 60% of transaction fees to subsidize miners’ rewards.

Zcash’s speedup comes from changes to its economic model; Solana’s current round only adjusts the clock.

The Robinhood Chain angle is even more interesting. A report on September 19 said that the seven-day average fees fell 82%, while the number of transactions only dropped 6%; and there’s still about $1.5 billion moving on-chain every day. Fees collapsed, but volume stayed.

So what does this mean? It means the chain’s throughput hasn’t changed, but the price users are willing to pay is going down. If Solana only speeds up without expanding capacity, it will end up at the same point sooner or later.

Speed is for market makers and arbitrage bots. Ordinary users want it to be cheap, stable, and not fail.

A team running an MEV strategy on Solana is probably doing the math right now: reducing block time from 400 ms to 332 ms shortens their front-running window by another 68 ms—do they need to rewrite the strategy?

Meanwhile, for a regular user opening a wallet, they’re still seeing that transaction pending for 30 seconds, and the priority fee of 0.0005 SOL.

In the eyes of these two people, Solana isn’t the same kind of blockchain.