IMO we can start with a lossy (spread) conversion and if it turns out that everything is working great but the only problem we still have is that he peg has only a bit too much slack (which would still be such as huge improvement from the status quo obviously) then we can tweak it again to tighten/remove the spread. That is a a relatively small change.
You are viewing a single comment's thread from:
Yeah, I do like taking steps to make it safer if and when we do implement something. Paying a small price to do the conversion may be a good way to introduce it. It's still possible some systemic risk exists we haven't yet seen and would only find once we see it in production (unless we develop ways to model this ideas effectively beforehand).