What is a dApp (decentralised application) and why it's beneficial

An app is a dApp when the things that matter most, who owns what and what the rules are, are enforced by code on a public blockchain rather than by a company's database. ethereum.org lists four properties: dApps are decentralised (no one person or group controls the network), deterministic (they behave the same wherever they run), Turing complete (they can perform any computation given the resources) and isolated (they execute in a sandboxed virtual machine).

The heart of a dApp is the smart contract, in ethereum.org's words "simply a program that runs on the Ethereum blockchain" (source). The classic analogy is a vending machine: with the right inputs, a certain output is guaranteed, and nobody behind the counter can decide otherwise.

Decentralisation is a spectrum, not a switch. Most dApps still have a hosted website and, often, admin keys that can upgrade the contracts. What makes them dApps is that the core state and rules live on-chain, so if the website disappears, the contracts, balances and history remain. ethereum.org is blunt about the other end of the spectrum: centralising a dApp's critical parts "eliminates many (if not all) of the advantages" of a blockchain.

A dApp is a stack of components, and only one of them lives on the blockchain.

Sources: ethereum.org on wallets, JSON-RPC, storage and smart contracts, and The Graph on indexing.

Here's what happens when someone clicks a button in a dApp:

1. The frontend builds a transaction, such as "swap 100 USDC for SOL", and asks the user's wallet to sign it.
2. The wallet shows the user what they're approving and signs it with their private key. The key never leaves the wallet.
3. The signed transaction is sent to the network through an RPC node.
4. Validators run the contract or program logic. On Solana, transactions are atomic: "If any instruction fails, the entire transaction fails and all state changes are reverted" (Solana docs).
5. Once the block is confirmed, the new state is public. Indexers pick up the change, and the frontend updates.

Nothing in that flow asks the dApp's operator for permission. That's the point.

A good dApp and a good web app can look identical. The difference is underneath: who runs the backend, and who holds the keys.

The trade is control for responsibility. A dApp removes the need to trust the operator, and hands users responsibility the operator used to carry. As ethereum.org's wallet guide warns, "there's no customer support in crypto" and "transactions can't be reversed".

The benefits are real, but they're specific. They matter most when a product involves value, ownership or rules that several parties need to trust.

Users own their assets

Balances and items are recorded on the ledger against the user's own address, not in a company's database. "Wallet providers don't have custody of your funds," as ethereum.org puts it. NFTs extend the idea to unique items: ownership "is stored on the blockchain for anyone to verify publicly", and other apps on the same chain can use the token instead of it being locked inside one platform (source).

Rules anyone can verify

Smart contracts "can be analyzed and are guaranteed to execute in predictable ways" (ethereum.org). On Solana, verified builds let anyone confirm a deployed program matches its public source by comparing hashes, though the docs note this proves what's running, not that it's secure.

Open access, around the clock

Decentralised finance, in ethereum.org's description, is "accessible to anyone who can use Ethereum", its markets "are always open", and "anyone can look at a product's data and inspect how the system works" (source). At the protocol level, "no single entity on the network can block users from submitting transactions" (source).

Composability

Public contracts behave like open APIs that any other contract can call (ethereum.org). A dApp that expects payment in one token can swap whatever token the user holds on the way in, and a flash loan lets someone borrow without collateral if it's repaid within the same transaction (source). New products stand on existing ones without asking permission.

Programmable payments

Stablecoins, tokens designed to stay at a fixed value such as one US dollar, let a blockchain carry stable, global payments that smart contracts can program (ethereum.org). That's no longer fringe. In December 2025, Visa launched USDC settlement in the United States, with its first banking partners settling over Solana, and reported more than US$3.5 billion in annualised stablecoin settlement volume (Visa). For apps and merchants, Solana Pay standardises payment requests.

Shared treasuries and DAOs

A DAO, or decentralised autonomous organisation, is "a collectively-owned organization working towards a shared mission", with its rules in smart contracts (ethereum.org). Its treasury can't be accessed "without the approval of the group". For grant programs and community funds, "trust the treasurer" becomes "check the contract".

A public ledger as the source of truth

Sometimes the benefit is simply that nobody has to take your word for it. In Auctioning, communities back businesses with RP, and paid fuel, at $1 per RP, is signed on-chain on Solana, so every paid contribution is logged publicly. Race recaps are written only from what the ledger recorded, never from claims.

Every benefit above has a cost, and good dApps are designed around these from day one.

• Onboarding is still harder. Users need a wallet, sometimes a funded account, and an understanding of what they're signing. ethereum.org lists user experience as a core drawback: "the average end-user might find it too difficult to set up a tool stack" (source).
• Every action has a fee. Ethereum fees are paid in gas, and its layer 2 networks exist largely to make transactions cheaper (ethereum.org). Solana charges a base fee of 5,000 lamports (0.000005 SOL) per signature, plus an optional priority fee that raises the chance a transaction is scheduled ahead of competing ones (Solana docs).
• Keys are unforgiving. Lose a seed phrase and the account is gone; sign a malicious transaction and the funds are gone. Chainalysis counted more than US$3.4 billion stolen across crypto from January to early December 2025, including US$713 million from individuals, with personal wallet compromises making up 20% of the value stolen (Chainalysis). The February hack of the Bybit exchange alone accounted for US$1.5 billion, which shows custodians get hacked too.
• Bugs are hard to undo. Contracts are designed to be hard to change, so a flaw can sit in production with real money behind it. ethereum.org puts losses from smart contract security defects at "easily over $1 billion" and warns that audits "won't catch every bug" (source).
• Transparency cuts both ways. Every transaction on a public chain is visible. Addresses are pseudonymous, not anonymous, and once one is linked to a person, so is its history.
• Upgrades are a trust decision. Whoever holds an upgrade key can fix bugs, and can also change the rules. Immutable contracts remove that risk, and the safety net with it.
• Centralisation creeps back in. A dApp with one hosted frontend, one RPC provider and one admin key can be switched off in practice, even if its contracts can't.

There's no single rule for dApps in Australia. Depending on what a dApp does, the business behind it can be regulated like any other financial or payments business, and the rules moved quickly in 2026. This is general information, not legal advice.

• Is your token a financial product? ASIC's INFO 225, last updated on 30 April 2026, says it depends on the rights and features attached to the asset, assessed case by case. A token might be, for example, an interest in a managed investment scheme, a security, a derivative or a non-cash payment facility.
• Do you hold assets for others? The Corporations Amendment (Digital Assets Framework) Act 2026 brings digital asset platforms and tokenised custody platforms into Australian financial services licensing. It received assent on 8 April 2026 and is due to commence on 9 April 2027, with a low-value exemption for platforms with no more than $10 million in transactions over 12 months (Gilbert + Tobin).
• Do you provide virtual asset services? AML/CTF reforms have extended AUSTRAC regulation to more virtual asset services, and providers of the newly regulated services had to apply to enrol and register by 29 July 2026 (AUSTRAC).
• Do you collect personal information? Australian privacy law can apply to the user data you hold off-chain, and anything written to a public chain can't be deleted later. Keep personal information off-chain.

One design question runs through all of this: custody. Do users sign their own transactions, or does your business ever hold assets on their behalf? Get advice on your specific design before launch, not after.

Both are mature, general-purpose platforms, and both run serious dApps. They make different trade-offs, and those trade-offs show up in your product.

Sources: ethereum.org on contracts, upgrades and finality; Solana docs on programs, accounts, transactions, fees, Sealevel and Alpenglow.

Both chains keep moving. Ethereum's Pectra upgrade (May 2025) let ordinary wallet addresses take on smart contract features such as batching, fee sponsorship and recovery, Fusaka (December 2025) made rollup data more efficient with PeerDAS, and Glamsterdam is in development, listed for Q4 2026. Solana's Alpenglow consensus upgrade was live on testnet and devnet, but not yet on mainnet, when we checked.

Choose Ethereum, or one of its layer 2s, when your product depends on assets, protocols or users already in the EVM ecosystem, or your team's experience is in Solidity.

Choose Solana when your product needs many small, fast, low-cost actions against one shared state: consumer apps, games, payments and anything where a user signs several transactions in a session. One layer 1 also means one place for composability, without bridging between networks. That's the shape of the products we build, which is why our on-chain work is Solana-native.

Building a dApp is closer to building fintech than building a website. This is the sequence we follow.

1. Decide what truly needs to be on-chain. Ownership, value transfer and rules several parties must trust go on-chain. Profiles, search, analytics and personal information stay off-chain, where they're cheaper, faster and private. A smaller on-chain surface is a smaller attack surface.
2. Choose the chain and the custody model together. Decide early whether users sign their own transactions or you hold anything for them. That choice shapes your security model and your regulatory position.
3. Design accounts, permissions and upgrades. On Solana, plan the account layout and program derived addresses before writing code. On either chain, avoid a single all-powerful admin key: ethereum.org recommends multisig control for admin actions, and timelocks or voting for upgrades (source).
4. Build wallet flows like checkout flows. Show users what they're signing in plain language, simulate transactions before asking for a signature, and remove steps. Account abstraction enables sponsored fees and batched actions on Ethereum, and Actions and blinks let a Solana transaction be offered anywhere a link can be shown.
5. Plan the read path. Load pages from an indexer or your own queryable store, and treat the chain as the source of truth that store is checked against.
6. Test harder than feels necessary. ethereum.org says unit testing alone "is minimally effective" for security, and recommends adding static analysis, fuzzing and, where the stakes justify it, formal verification. On EVM chains, frameworks such as Foundry and Hardhat compile, test and deploy contracts. On Solana, Anchor is a framework for building secure programs, and LiteSVM runs fast, in-process tests before you rehearse on devnet.
7. Audit, then keep auditing. Get an independent audit before mainnet, run a bug bounty scaled to the funds at risk, publish verified builds and re-audit after significant changes.
8. Launch with monitoring and a plan. Watch your programs and key accounts from day one, and decide in advance who can pause what, and how users will be told.

Spectrum Web Co is a Software Engineering Collective in Sydney, and Solana-native products are one of the four kinds of software we build. Our Ship a Product service takes a dApp from an idea, or a stalled build, to launch:

• Discovery decides what belongs on-chain, agrees the first shippable slice and explains the architecture before any build starts.
• Build runs in short, visible cycles, so you see progress every week.
• Launch means real users on a real domain, with monitoring from day one.
• Cadence keeps releases coming every month.

You own the code, the accounts and the documentation throughout, and you can try our own Solana product, Auctioning, today.

Not sure your product needs a blockchain at all? That's a good first conversation. Book a 30-minute intro call or send us a message, and we'll reply with a clear monthly scope.

Nothing in this post is financial, legal or investment advice, or a recommendation to buy, sell or hold any digital asset.

• ethereum.org: Introduction to dapps, smart contracts, composability, smart contract security, wallets, JSON-RPC API, decentralised storage, development frameworks, DeFi, NFTs, stablecoins, DAOs, layer 2, account abstraction, roadmap, blocks and single-slot finality
• Solana: programs, accounts, transactions, fees, verified builds, Actions and blinks, LiteSVM, Alpenglow and Sealevel
• Anchor documentation, The Graph documentation and the Solana Pay core overview
• Visa launches stablecoin settlement in the United States, Visa, 16 December 2025
• 2025 crypto theft reaches $3.4 billion, Chainalysis
• INFO 225: Digital assets: Financial products and services, ASIC
• Corporations Amendment (Digital Assets Framework) Act 2026, Federal Register of Legislation, and Australia passes highly anticipated digital asset regulation, Gilbert + Tobin
• Register as a remittance or virtual asset service provider, AUSTRAC