How we score · version 1.1

How we score networks

CriterionWeight
Decentralization25%
Security track record20%
Speed15%
Cost to use15%
Reliability15%
Wallet and app support10%

A network (or blockchain) is the system your crypto moves on: Bitcoin, Ethereum, Solana, the Lightning Network and so on. We score networks on how well they serve the people who use them. We never score coins as investments, and a high network score is not a reason to buy its coin.

In the setup finder, we re-weight these criteria toward what you said matters most: low fees, speed or security. The overall score on this site uses the standard weights in the table above.

For networks built on top of another network (“layer 2s”, such as Base or Arbitrum), we also check whether users can withdraw to the base network without the operator’s permission, using public risk analyses such as L2BEAT.

Payment-channel networks such as Lightning work differently. Payments settle between the people involved and only reach the base network when a channel opens or closes, which can be months later or never. For these networks we score speed as the time until a payment settles and can’t be reversed in practice, and cost from the fees that routing nodes advertise, converted to US dollars at a dated bitcoin price. The review says when these stand in for the standard measures below.

Decentralization

How hard it would be for a small group to halt, censor or rewrite the network. When the network’s Decentralization Index can be calculated, this criterion uses it directly: seven measures, including the Nakamoto coefficient, node counts, client diversity, token concentration and insider allocation at launch.

Some networks can’t be measured that way. A payment-channel network such as Lightning has no block producers and no shared ledger of its own, so most of the index’s measures don’t exist for it. For those networks we score this criterion from sourced evidence against the descriptions below, and the review says the index couldn’t be calculated. A network that is only missing data waits until it can be measured, and we don’t publish its score before then.

  • 10: No small group could halt or censor it. Thousands of independent operators run it on home computers, no one program runs most of it, and no single party can change the rules.
  • 5: A handful of parties could together halt or censor it, or one program runs most of it. Upgrades are coordinated by a core team, but operators choose whether to adopt them.
  • 0: One party or a very small group controls it, or can change the rules or pause it on their own.

Security track record

  • 10: No successful attacks, consensus failures or forced rollbacks in five years, and attacking it would cost a fortune.
  • 5: A past incident, fixed openly. Or a lower but still substantial cost to attack.
  • 0: Recent rollbacks, deep or damaging chain reorganizations, or “51%” attacks. A short reorganization that the network settles within minutes, with no double spend, is normal for proof-of-work networks and does not count.

Speed

Time until an ordinary transfer is final and can’t be reversed in practice. For layer 2s, we use the time until the transaction is final on the base network, and note the faster “soft” confirmation.

  • 10: Under 5 seconds.
  • 7: Under 1 minute.
  • 5: Under 15 minutes.
  • 2: About an hour.
  • 0: Several hours or more.

Cost to use

Median fee for a simple transfer over the last 30 days, in US dollars.

Some networks have no published 30-day median for simple transfers. For those we use the 30-day median for all transactions as a stand-in. The score note says so, and the network profile leaves the typical fee blank. Where no 30-day median in dollars is published at all, we estimate one from the median fee rate, the size of a standard transfer and a dated exchange rate. The score note shows the working, and when the estimate sits close to a line we take the lower score.

  • 10: Under $0.01.
  • 7: Under $0.10.
  • 5: Under $1.
  • 2: Under $5.
  • 0: $5 or more.

Reliability

  • 10: No outages or halts in the last 24 months.
  • 5: One outage or halt lasting under a few hours.
  • 0: Repeated outages, or halts lasting many hours.

Wallet and app support

  • 10: Supported by most major wallets, hardware wallets and exchanges, with a broad range of apps.
  • 5: Supported by some major wallets and exchanges.
  • 0: Needs special software, with few wallets or exchanges.

Figures we publish

Each network profile shows a typical fee, time to finality and Nakamoto coefficient, with the date we measured them and the source. These change with network activity, so we re-measure at least every three months.