AsterScan · Documentation
The Alpha Protocol
How AsterScan ships exploratory features without misleading anyone. The rules we follow before, during and after an experiment becomes part of the explorer.
Why an “alpha” surface at all
Aster Chain is the first L1 to ship privacy-by-default at the order level. Roughly half the trading flow is encrypted on-chain. Every existing block-explorer convention — “list the recent transactions, decode the calldata, render the from/to/value” — breaks against that.
We could pretend that surface doesn't exist (most explorers do). We could publish raw inferences as if they were facts (some dashboards do). The third option is to declare an “alpha” surface where we openly experiment with reading the public metadata around encrypted activity — wallet behaviour, validator participation, buyback timing, position aggregation — and tell visitors exactly what confidence we have in each output. This page is the contract for that surface.
The four principles
Lifecycle: ALPHA → BETA → STABLE
Every feature on the explorer carries one of three status badges. The badge tells you what guarantee we're willing to make about the numbers behind it.
| Stage | Purpose | Contract | Typical duration |
|---|---|---|---|
| ALPHA | Exploration | Methodology disclosed. Accuracy NOT guaranteed. Do not trade. | Until the underlying signal is proven coherent (typically 4-12 weeks). |
| BETA | Hardening | Methodology stable. Accuracy bounded but not yet 100%. Monitoring for regressions. | Until the error bounds tighten enough that the metric is publishable as primary intel (typically 2-6 weeks). |
| STABLE | Public surface | Indexed, embedded in public RSS feeds, available via the public API. Numbers carry the AsterScan reputation. | Forever, unless the underlying chain mechanism breaks. |
A feature can be demoted (BETA → ALPHA, STABLE → BETA) if a regression appears. Demotion is announced in the changelog and never silent.
How to read error bounds
Where we surface a number that comes from an inference rather than a direct read, we attach an error bound. The convention:
- ±5% or tighter — production-grade. We're comfortable building secondary tooling on top.
- ±5–15% — directional only. The shape of the curve is meaningful, the absolute value isn't.
- ±15–30% — exploratory. Look for trend changes, not levels.
- >30% — we shouldn't be showing a number. If you see one, it's a bug — please report it.
Error bounds are computed from the underlying sample size and the known properties of the inference (e.g. for Private Mirror, we use the public/private ratio observed over the same window).
What's in alpha right now
As of this page being last edited:
- Private MirrorALPHA
- Time MachineALPHA
- Insight EngineALPHA
- Wallet ReplayALPHA
- Cross-Date DiffBETA
- Liquidation MapBETA
The list is regenerated by the dashboard rather than hand-maintained — if something is missing here, it's missing from the lifecycle tracking, not from the page.
How to use alpha output safely
A short rulebook for visitors:
- Read the methodology section on each alpha page before you read the chart.
- Treat any single alpha number as the start of an investigation, never the end.
- Cross-check against on-chain primary sources (the Aster RPC, the bridge contracts) — links are provided on every page.
- If the chart looks empty or the number looks suspiciously round, the indexer is likely warming up. Refresh in 30 minutes.
- If you find a discrepancy bigger than the published error bound, file it via the feedback link — you're probably right and we want to know.
/alpha/* are the experiments operating under it. If the two ever drift, the protocol wins.