High Peer Count Blocks

Networks:
Ethereum Mainnet
Time range:
Start 2026-01-14T00:00:00Z
End 2026-01-21T23:59:59Z
block-propagation peer-count network-performance xatu
Published January 22, 2026 at 01:21 AM UTC

Question

Does having a massive peer count (6k+ peers) give a node a significant advantage in seeing blocks first?

Background

The Ethereum p2p network propagates blocks through a gossip protocol. In theory, having more peers should help a node receive blocks faster - more connections mean more chances to hear about a new block quickly.

We tested this hypothesis using utility-mainnet-prysm-geth-tysm-003, a Prysm/Geth node configured with ~6,000 peers (compared to the typical 50-100 peers for most nodes).

Win rate measures how often this node sees a block first among all nodes in the Xatu network (typically ~180 nodes). A random win rate would be about 0.56% (1/180 nodes).

Percentile ranking shows where this node falls in the distribution of block arrival times - e.g., 87th percentile means it sees blocks faster than 87% of the network.

Investigation

When Competing for First

How often does our high-peer node see blocks first compared to the rest of the network?

View Query: high_peer_win_rate_daily
Loading...

With ~180 nodes competing, random chance would give a win rate of ~0.56%. Our high-peer node achieves about 0.33% on average - actually below random chance. This suggests that while the node is consistently fast, a small group of exceptionally well-positioned nodes consistently beat it to first place.

When Looking at Speed Ranking

A better measure than raw wins is percentile ranking - where does this node fall in the distribution of block arrival times?

View Query: high_peer_percentile_hourly
Loading...

The high-peer node typically ranks in the 75th-95th percentile, with a median around 85%. However, there's notable variability - P25 can dip to ~50% while P75 reaches ~95%. This node is usually fast, but not consistently so.

When Comparing Timing Over Time

How does our node's timing compare to the network median and fastest nodes over time?

View Query: high_peer_timing_hourly
Loading...

Our high-peer node consistently tracks between the network fastest and network median. The gap between our node and the median shows the advantage of having many peers.

When Looking at Delta from Median

How much faster or slower is our node compared to the network median? Negative values mean we saw the block faster.

View Query: high_peer_delta_hourly
Loading...

The consistently negative delta shows our high-peer node sees blocks ~20-40ms faster than the network median on average.

Takeaways

  • A 6k+ peer count does provide a measurable speed advantage - seeing blocks ~30-40ms faster than the network median
  • The node typically ranks around the 85th percentile - usually faster than most of the network, but with notable variability
  • However, the win rate is actually below random chance (~0.33% vs 0.56% expected) - a small group of exceptionally well-positioned nodes consistently beat our high-peer node to first place
  • Diminishing returns: Having thousands of peers helps maintain consistently good performance, but doesn't overcome the advantage of optimal network positioning and infrastructure that some nodes have