ICP numbers are probably incomplete

[brson]

ICP is using https://rosetta-api.internetcomputer.org for its data. The numbers appear to reflect "ICP transactions" on https://dashboard.internetcomputer.org/, but is nowhere near the ~4800 tps that website claims. Probably these numbers aren't accounting for cannister messages.

Previous discussion: #41

[domwoe]

+1 These are only the ICP token transactions - a very small part of the messages on the Internet Computer.

[sasa-tomic]

@brson since you're already trusting https://rosetta-api.internetcomputer.org/ for its data, why not trust https://ic-api.internetcomputer.org/ ?
The metrics you are looking for are present there today: https://ic-api.internetcomputer.org/api/v3/swagger

I do agree it's a bit "centralized" in a sense that you do have to trust internetcomputer.org that it provides you valid data, but we are actively working on adding more metrics in a trustworthy and decentralized manner. It’s a tough problem so may take a bit of time. But in the meantime you should be able to use the metrics endpoint above and then switch to the new decentralized metrics when they become available.
WDYT?

[brson]

@sasa-tomic This project is structured to load blocks, count the transactions in them, and average them over time. That is the single thing this project does.

While I would prefer to always go to a node that participates directly in the network to get that information, secondary aggregators like Rosetta, if they can provide blocks, timestamps, and transactions, are ok at the moment.