Skip to main content

Chain Variable Changes

· 3 min read
Evan Vigil-McClanahan
Coco Tang
Rahul Garg

Today we're going to issue a transaction to change several chain variables. The high level summary:

  1. More high scoring Hotspots will have a chance to be part of the consensus group.
  2. More transactions will be allowed per block.
  3. We'll widen the range of RSSIs that are allowable for PoC witnesses, to address community complaints.
  4. Old witnesses will get get cleaned up over time and no longer contribute to Proof of Coverage path generation.

Details​

  1. election_selection_pct will be changed from 20 to 1. Since we now have a large number of high scoring Hotspots and the top of the list is currently dominated by Hotspots in suspect configurations, we're spreading out consensus membership more, in the hopes of disincentivizing bad behavior by lowering the returns to cheating your way to a high score.

How will this affect Hotspot owners?

  • More high scoring Hotspots will have a chance to be in the consensus group instead of the select few now. This, in conjunction with a change we made a few weeks ago to geographically distribute consensus group Hotspots will further disincentivize unwanted behavior.
  1. batch_size will be changed from 200 to 400. We lowered this a few weeks ago due to some gigantic blocks that caused some issues with gossip. It looks like we lowered it too much, as we're only able to sustain about 150 transactions per block. We currently have a target of 250 transactions per block, so we're raising the limit in order to hit that target. Expect more adjustments in the future.

How will this affect Hotspot owners?

  • While this may not have a direct affect on Hotspot owners, you may see reduced pending transaction times.
  1. poc_good_bucket_high will be changed from -80 to -70 and poc_good_bucket_low will be changed from -118 to -130. These change what recorded signal strengths are allowed for witnesses. This is an anti-cheating measure, but it looks like we've been too restrictive, and people with high gain antennas are having issues with Hotspots witnesses nearby and valid Hotspots too clearly.

How will this affect Hotspot owners?

  • This means those with non-stock antennas will no longer be penalized and bucketed as a poor witness. Hotspot owners with high gain antennas may see more challenges succeed.
  1. witness_refresh_interval will be added to the chain at 10080 (approximately one week of blocks), and witness_refresh_rand_n will be added at 1000 (an implementation detail that nevertheless must be on the chain for the sake of determinism). The former will probabilistically age out witnesses over the course of a week, leading to more valid and recent paths. This should allow the network to adjust to seasonal variations and the slow churn of spots as people add, remove, and move them around.

How will this affect Hotspot owners?

  • Currently witnesses never expire, which means old and weak witnesses are often used to create paths. By refreshing witnesses on a weekly interval, we can purge bad Hotspots and create more meaningful PoC paths. We can expect fewer challenges to spuriously fail with this change.