How we score · version 1.0
How we measure network data
Every number on the Measure pages comes with a source, a date and a label saying how we got it. This page defines each figure so you know exactly what you’re comparing.
Definitions
| Figure | What it means | How we measure it |
|---|---|---|
| Transactions per second | How busy the network is | Transactions over the last 30 days ÷ 2,592,000 seconds. User transactions only: consensus votes that some networks record as transactions are excluded. We don’t show theoretical maximums. |
| Daily active addresses | How many addresses use it | Unique addresses that sent or received a transaction each day, averaged over 30 days. Addresses aren’t people: one person can have many, and bots inflate counts. |
| Typical fee | What a simple transfer costs | Median fee for a simple transfer over 30 days, in US dollars. |
| Time to finality | How long until a payment can’t be reversed | For layer 2s, the time until a transaction is final on the base network. We note faster “soft” confirmations separately. |
| Nakamoto coefficient | How few block producers could disrupt it | The smallest number of entities controlling enough block production to halt or censor the network. We use a third of stake for every proof-of-stake network, whatever thresholds its own consensus protocol uses, so figures compare like for like; and half of hash power (by mining pool) for proof of work. Validators or pools run by the same company count as one entity. A staking pool whose validators are run by several companies, such as Lido, is split among those companies. When public data can’t link all stake to the company that runs it, we take each company we can size at the most it could run and record the lowest figure that gives. The note says how the unlinked stake could change it. |
| Block producers | How many produce blocks | Active validators, miners or mining pools that produced blocks in the last 30 days. |
| Full nodes | How many independent computers check every rule | Reachable full nodes reported by public crawlers. The true number is usually higher. |
| Largest client share | Software diversity | Share of nodes running the most common client implementation. |
| Hardware to run a node | Whether ordinary people can verify it | Home computer, powerful desktop or data-center servers, from the network’s published requirements. |
| Held by top 100 addresses | Ownership concentration | Share of circulating supply in the 100 largest addresses, excluding known exchange, bridge, burn and contract addresses where the data allows. Exclusions are listed in the notes. |
| Outages, last 24 months | Reliability | Times the network halted or stopped finalizing blocks. |
| New coins per year | Issuance | Coins created per year as a share of supply. |
| Who can change the rules | Control over upgrades | Whether node operators choose upgrades, a team coordinates them, or admin keys can force them. |
The two scores are explained on their own pages: the Decentralization Index and insider allocation.
How we get the numbers
Each figure is labelled with one of three methods:
- Researched: gathered by our team from block explorers, network dashboards, official documentation and published research, with the source linked.
- Data feed: pulled automatically from a public data provider on a schedule.
- Our node: measured directly from a node we run ourselves.
Today almost everything is researched. Over time we’ll move figures to automated feeds, and then to our own nodes and block explorers, so you can verify the data without trusting anyone, including us.
Freshness
We re-measure every network at least every three months. Each figure shows the date it was measured. Network behaviour changes quickly, so treat older figures with care.
Corrections
Spotted something wrong? Email us with a source. We’ll check it, correct it if needed and note the change.
See these scores in action: compare networks side by side.