8

In Lightning, the sender can never be sure that a multi-hop payment will go through, as intermediary nodes don't announce the distribution of funds in their channels (only the total capacity). Is there any publicly available data on how many attempts it takes on average in practice to route payments of different sizes? Something like this (the numbers are my wild guess):

Amount - Average paths tried

100 sat - 1.8

1000 sat - 3.3

10000 sat - 5.9

1000000 sat - 13.2

There is a similar question Lightning network routing failure rate and a 2018 article linked from there, but it addresses the success probabilities based on capacities only (this can be calculated from a network snapshot). I'm interested on the data from real-world transactions.

Improve this question
2
  • 3
    It seems to me that a ‘generalized network average’ will not be extremely useful, as I’d expect the standard deviation to be very large. That said, I do think this is a measure that could be meaningful, but it would need to be standardized against the liquidity of your channels and of your neighbouring peers, or something along those lines. Is there a term for the quantification of a LN node’s liquidity within the network graph? Affluence? Ha. Good question, anyways.
    – chytrik
    Commented Oct 10, 2019 at 22:13
  • I suggest to create histograms or a box plot. While not a single measure like a (generalized) average it will still be easy to read and probably as expressive as it gets
    – Rene Pickhardt
    Commented Oct 11, 2019 at 15:21

1 Answer 1

Reset to default
7

I don't think there is a Public dataset online with an answer to your question. This would also be extremely hard to create as the results could change a lot depending on which node is doing the experiment.

I know people are probing payment channels you could do the same. Just do payments to random payment hashes and collect the data for your node. You will get the routing failures and attempted tries in that way you can easily collect those statistics yourself.

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.