Simulating a Decentralized Lightning Network with 500,000 payments, 0.01% fee per hub and 10 Million Users: 100% success (99.9986%)

I repeated Diane Reynolds' simulation, as described here, using her program. The only change I made was that instead of having "11 different fee policies evenly distributed among users", each node simply asks for 0.01% fee (= value x 0.0001) to route a transaction. Some of the 11 fee policies were quite complicated, like "trying to make channels more balanced".

Here are the main results:

Payments attempted: 500,000

Fee as a percentage of the payment 0.175

Routing failed: 7

Routing failed for bigpayments: 3

Routing failed for midsizedpayments: 1

Routing failed for micropayments: 3

Average lengths of routes: 17.5

So the payment failed for only 0.0014% of the payments. Note that for the last 100,000 payments, routing failed only once (1 midsizedpayment) and from 100,000 to 500,000 only 3 times (1 big, 1 midsize, 1 micro). The average total fee for a payment makes sense, as 0.01% x 17.5 hubs = 0.175% total fee. I didn't realize the program stopped after 500,000 payments and will now try to run it for more payments.

Submitted December 30, 2017 at 03:06AM by sumBTC http://bit.ly/2Cmv9GF