Browse Source

update readme

master
rra 5 years ago
parent
commit
15d19e0435
  1. 21
      README.md

21
README.md

@ -1,22 +1,29 @@
# directorate for applied fediverse research
# doing frequent applied fediverse research
* independently tries to verify fediverse statistics
* hands-on approach to verify statistics on fediverse from fediverse.network and the-federation.info
* draws conclusions from that
## methodology
Currently the script starts from <https://post.lurk.org> and queries `/api/v1/instance/peers` to find servers it is peering with. For each of the peering servers it hasn't seen before it does the same and in addition it tries to query `/api/v1/instance` for meta data.
### Mapping the network
Currently the script starts from <https://post.lurk.org> and queries `/api/v1/instance/peers` to find servers it is peering with. For each of the peering servers it hasn't seen before it does the same. This from the assumption that getting peer lists from Mastodon & Pleroma gives enough of a view of 'known fediverse' to work with.
This method is a bit lacking because providing `/api/v1/instance` is voluntary and specific to later versions of mastodon/activitypub fediverse. We should study the methodology of [fediverse.network](https://fediverse.network/info) for better results.
> initial peer list
>> all peers of the initial peers
>>> all peers of the peers of the inintial peers
------------------------------------------------- +
the known fediverse?
When the request fails on a given instance it just logs it as 'error' now.
### Instance metadata
We try to query `/.well-known/nodeinfo` for instance meta-data such as software type etc. This is what both fediverse.network and the-federation.info do
When any request fails on a given instance it logs the raised `Exception`, if it is a HTTP error instead we currently log the answer.
Latest scrape results can be found in `instance_scrape.json`
## TODO FIXME
* add detailed error message to json when we get one
* ~~add detailed error message to json when we get one~~
* ~~abstract the functions so we can multithread them~~
* find a way to also scrape for instances that don't announce themselves

Loading…
Cancel
Save