You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
While ProbabilisticScorer in #1227 is likely to be incredibly powerful for always-online nodes that are actively probing or have large payment volume, its much less likely to be helpful for nodes that do not, as their information will simply not be enough to assign meaningful scores. Instead, we should explore tracking how often a channel is unbalanced as a way of assigning "badness" scores to nodes from which we can calculate path probabilities. This would be more akin to the way lnd assigns success probability, where it tracks historical successes and failures at different amounts.
This may also be a good way to support mobile with a good scorer, where we can ship only the "badness" scores to mobile clients for use in routing, without shipping our current estimates of network balance allocations.
The text was updated successfully, but these errors were encountered:
While
ProbabilisticScorer
in #1227 is likely to be incredibly powerful for always-online nodes that are actively probing or have large payment volume, its much less likely to be helpful for nodes that do not, as their information will simply not be enough to assign meaningful scores. Instead, we should explore tracking how often a channel is unbalanced as a way of assigning "badness" scores to nodes from which we can calculate path probabilities. This would be more akin to the way lnd assigns success probability, where it tracks historical successes and failures at different amounts.This may also be a good way to support mobile with a good scorer, where we can ship only the "badness" scores to mobile clients for use in routing, without shipping our current estimates of network balance allocations.
The text was updated successfully, but these errors were encountered: