> The issue is that big majority will see this and think "lightning is so bad if my $100 payment fails one out of three tries."

What gets measured, gets improved.  ¯\\\_(ツ)\_/¯

https://stacker.news/items/412708/r/random_

If it's any consolation, payment reliability is high in the pronet. 

I'm not saying it's a bad measure - it's standardized and provides some insight. 
The issue is that big majority will see this and think "lightning is so bad if my $100 payment fails one out of three tries." 

nout

lightning

Amboss CEO On Growth Of The Lightning Network, Tether (USDT) On Lightning

Jestopher_BTC

Hot take but that graph makes lightning look more reliable than it is because it uses multipath payments.

>If you simulate random payments from any node to any node that's by far not matching the reality.

If you restrict the payment pairs to large well managed nodes, you introduce [survivorship bias](https://en.wikipedia.org/wiki/Survivorship_bias). Why shouldn't you measure any nodes ability to pay any other node in the network? 