Polygon Nightfall
Block Explorer

Nightfall is an optimistic rollup for companies that can’t let competitors read their ledger. The explorer had to prove the network works without exposing who uses it, so deciding what not to show was half the design. I mapped eight explorers first. All of them covered blocks and transactions. None had a place for what an optimistic rollup runs on: blocks that can be challenged, and the proposers who stake on them.

One session with the Nightfall team: who uses an explorer and why, what the competitors do, then seventy minutes of sketching. We left with five pages on paper — front page, block, transaction, challenged blocks, proposer — and a list of what each would refuse to show.

Deposits and withdrawals touch Ethereum, so they’re public anyway — the explorer shows them in full and links to Etherscan. Transfers inside Nightfall are the private part: you see that one happened, when, and the fee — not how much moved. NFTs get an ID, never a value.

The rule: show what proves the network works, hide what belongs to the people using it.

Challenged blocks: how many were disputed, which ones, who challenged, and who was right. Even the name was a decision — the sketch still carries the sticky note that says so. The proposer page turns good blocks, bad blocks, and stake into a track record anyone can judge.

Exports, tracked accounts, anything behind a login waited for version two. Three weeks were enough for one thing done properly: a public explorer, prototyped and tested. When the design system arrived late, I built from wireframe components and moved the schedule, not the scope.

Five 30-minute think-aloud sessions on a Figma prototype, with people from the Nightfall team. Three tasks: find network stats, check a deposit’s status, look up a proposer.

1. Tester after tester looked for a search button; the bare input read as a filter.
2. Transaction status wasn’t seen at first glance — people scanned the list instead.
3. “Are these L1 or L2 blocks?” came up unprompted more than once.
4. One timestamp raised a question: time of deposit, or time of block?
5. Fees in ETH and MATIC read as one number; nobody could say what “average fee” averaged.
6. “Settled” meant little even to people who work on the network.
7. The proposer page was hard to reach.

Testers also disagreed. One wanted the proposer address on every transaction, another said it didn’t belong there. One wanted the orange pending state gone, because it looked like a network fault.

The testers were network insiders, not end users. They catch vocabulary and structure; they won’t catch what a first-time visitor misses. Next I’d test with people outside the team and score task completion instead of collecting quotes.