Our scheme doesn’t require dynamically adjusting fees, as unconditional fees are intended to be deployed along side other mitigations which: 1) limit the resources that attackers can consume 2) preserve the ability of peers with historically “good” behavior to make payments.
I would have assumed that a dynamic scheme would have baseline low fees (so that honest parties aren’t charged a high rate), and then adjust upwards if traffic increases? Which would mean you do need to adjust quickly. Again, unclear to me what the actual propose scheme is here so it’s difficult to comment
I agree, my original comment was that I don’t think we’ll be using OMs as input into pathfinding.
I’d really like to emphasize that I think that effort here would be best spent on simulation of the proposed solution (or adapting #1280 for OMs) under adversarial conditions. That’s where we learned the most with channel jamming.
Our simulator should be reasonably usable for these purposes. We didn’t implement unconditional fees (just manually calculated them as 1% of total fees post-simulation), but would certainly welcome contributions there!