Loads of fake peers advertised on bitcoin network

[Coding Enthusiast]

Even worse is that sometimes a big percentage of the addresses returned by DNS seeds are not valid.
For example last time I was testing about 20% of addresses returned by seed.bitcoin.sipa.be were dead ends

[1714874809]

1714874809

[1714874809]

1714874809

[1714874809]

1714874809

[1714874809]

1714874809

[1714874809]

1714874809

[matthias.kit]

We run a monitoring (https://www.dsn.kastel.kit.edu/bitcoin/) of the Bitcoin P2P network at Karlsruhe Institute of Technology (KIT) and noticed the invalid addresses, too. Based on the findings posted above, we analyzed what the spamming could be useful for. We found that the propagation of addr messages can be used to estimate the number of neighbors of public peers running Bitcoin Core and to match multiple addresses to the same public peer.
You can find our report at https://arxiv.org/abs/2108.00815.
We do not know who the spammers are but this might suggest that one of their objectives is learning information about the Bitcoin P2P network, esp. about its topology.

[dkbit98]

We do not know who the spammers are but this might suggest that one of their objectives is learning information about the Bitcoin P2P network, esp. about its topology.

It smells (read stinks) to me that it could be some government agency, and this is happening in same time when they are pushing hard for some crazy bill regulations for Bitcoin and all crypto.
I can't prove anything but it could be they are collecting information for some tracking and surveillance of all nodes and transactions.

The conclusion interest me. Bitcoin have 12775 nodes

So is this number of nodes from Bitnodes more correct than Luke Dashjr version that is showing much more nodes?
https://luke.dashjr.org/programs/bitcoin/files/charts/software.html

[pooya87]

So is this number of nodes from Bitnodes more correct than Luke Dashjr version that is showing much more nodes?
https://luke.dashjr.org/programs/bitcoin/files/charts/software.html

They are both correct (ignoring the errors coming from their methodology). Bitnodes shows nodes that are listening for incoming connection (among spy nodes) but the other is showing all nodes in existence including nodes that don't listen for incoming connections.

[pooya87]

@piotr_n mention Luke's DNS seed down for a while, so i wonder how accurate is it.

The two don't necessarily have to be related. The DNS server may be hosted somewhere else or use an entirely different setup specially since DNS seeds report listening nodes not all nodes. Also dnsseed.bitcoin.dashjr.org is currently up and running right now. Here is an online to tool to quickly dig DNS: https://toolbox.googleapps.com/apps/dig/#A/

[matthias.kit]

Sorry for the late reply.

The conclusion interest me. Bitcoin have 12775 nodes (according to https://bitnodes.io/ which exclude node which don't accept incoming connection), so DoS cost is quite expensive and it probably only reduce propagation speed to whole network. However, it's major concern for altcoin which have very few full node count.

I agree that a DoS attack against the whole Bitcoin P2P network would probably be very expensive. However, attacking the most connected nodes might be a more cost-effective attack strategy than attacking random nodes. Therefore it is desirable to protect the identity of well-connected nodes.

They are both correct (ignoring the errors coming from their methodology). Bitnodes shows nodes that are listening for incoming connection (among spy nodes) but the other is showing all nodes in existence including nodes that don't listen for incoming connections.

See https://luke.dashjr.org/programs/bitcoin/files/charts/historical.html which differentiates between listening and non-listening nodes.