Curriculum·S205 On-Chain and Fundamental Data·about 30 min
Metrics that carry signal
By the end of this lesson you can
- →Apply the cost test to any metric, and rank a dashboard by how expensive each number is to fake
- →Explain why address-based metrics are the cheapest to manufacture and the most quoted
- →Identify what an incentive program does to every metric measured during it
- →Select a small set of metrics that are expensive to fake and relevant to a stated thesis
Sophomore · enrolled learners
This lesson opens with LayerZero's sybil identification, May 2024.
- What happened
- Ahead of a token distribution, LayerZero ran a two-stage process to identify addresses controlled by single parties operating at scale. A self-reporting phase closing on 18 May 2024 offered participants 15 percent of their intended allocation in exchange for declaring their own sybil addresses, followed by a bounty phase from 18 to 31 May paying reporters 10 percent of the intended allocation for each address identified, requiring submissions of at least 20 addresses with a stated methodology. The published list contained 803,093 addresses flagged as potential sybils, narrowed from around two million identified initially, against approximately six million addresses eligible for the distribution, which is about 13 percent. The bounty process was paused after an influx of reports.
- The decision point
- Active addresses is among the most quoted on-chain metrics and is treated as a proxy for users. Here a protocol, using its own data and a paid incentive to surface the truth, established that at least 13 percent of its eligible address base was manufactured. The figure is a floor produced by an adversarial process, not a measurement, and every address-based metric published about that ecosystem during the qualifying period was computed on the inflated denominator.
What you will be able to answer
- →What is the cost test?
- →Why are address counts the weakest metric?
- →What does an incentive program do to metrics?
- →Name three metrics that are expensive to fake.
Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.
It is free. We do not sell the list and there is nothing to buy at the end of it.
Sources and review
- https://cointelegraph.com/news/layerzero-concludes-sybil-self-reporting-phase
- https://www.theblock.co/post/295274/layerzero-labs-ceo-announces-pause-of-sybil-bounty-hunter-process-after-influx-of-reports
- https://www.sec.gov/comments/sr-nysearca-2019-01/srnysearca201901-5574233-185408.pdf
- https://docs.llama.fi/
Confidence medium·Volatility high·Reviewed 2026-08-06·Owner unassigned
Contested
The 803,093 figure is the output of an adversarial identification process combining self-reporting and paid bounties, narrowed from a larger initial set, and it is a floor rather than a measurement. Some flagged addresses were disputed. Present it as what a protocol established about its own base, with the method attached.
Sybil activity is not uniformly illegitimate; multiple addresses have legitimate uses including the tiering in F104-02. The metric problem is not that people are cheating, it is that an address count was never a person count and is routinely read as one.