Where blockchain SHOULD NOT be used

In my first year in Web3, I was so excited about smart contracts that I wanted to put everything on-chain. I designed a chat app that stored messages on-chain, a school grading portal that stored test scores on-chain, and an attendance log that stored every click on-chain.

They were all disasters.

They were slow, they cost thousands of dollars in gas to test, and they leaked private data to the entire world. I had made the classic rookie mistake: treating blockchain as a general-purpose cloud database.

It isn't. It is a highly specialized, very slow, and very expensive trust primitive.

1. The Litmus Test: The Shared-Write Database Check

Before deciding to write a single smart contract or build a dApp, you should run your idea through a simple set of questions (originally formulated by computer scientist Wüst & Gervais):

1. Do you need to store state? If no, you don't need any database.
2. Are there multiple writers? If only one entity writes to the database (e.g. only you, or only a single company), use a standard database like Postgres. It is infinitely faster and cheaper.
3. Is there a trusted third party? If you have a coordinator that both sides fully trust (like AWS, or a mutually agreed escrow company), use a centralized database managed by them.
4. Are the writers untrusted or hostile? If multiple entities need to update the database but they do not trust each other, this is where blockchain becomes viable.

If your project doesn't pass all four steps, do not use blockchain.

2. Layman Explanation: The Public Chisel

Imagine you want to keep a personal diary.

• Do you buy a notebook and write in it with a pen (centralized)?
• Or do you hire 100 sculptors to follow you around, chiseling every sentence you say onto giant granite blocks and storing them in the town square (blockchain)?

For a personal diary, chiseling granite is ridiculous. It takes hours, costs thousands in sculptors' salaries, and everyone can read your secrets.

You only use the granite chiseling team if you are writing the Town Constitution — a document that no single person should be allowed to erase or alter, and that must be visible to every citizen forever.

Most application data is like a diary. It is private, changes constantly, and doesn't need to be chiseled in granite.

3. Technical Comparison: Blockchain vs. Database

Let's look at the numbers of running a transaction on a standard database versus a blockchain:

| Metric | PostgreSQL (AWS RDS) | Ethereum Mainnet | |:---|:---|:---| | Writes per second | 10,000+ transactions | ~15–30 transactions | | Read latency | < 1 millisecond | ~12 seconds (average block time) | | Cost per write | Less than $0.00001 | $2.00 – $50.00+ (varies with congestion) | | Data Privacy | Encrypted, access control limits | 100% public to anyone running a node | | Storage cost | ~$0.10 per Gigabyte/month | ~$20,000.00+ per Gigabyte (gas cost) |

Staggering differences. If you store a 1MB profile picture on Ethereum, it will cost you tens of thousands of dollars in gas.