Personally, I think there are numerous reasons to avoid extrapolating. There are good reasons to know the characteristics of each node you’re aiming to add to a subnet (not just the current state, but the recent history and performance details, which are only recorded for nodes that belong to a subnet). See proposals 134486 and 134491 as example cases.
Hey @sat, it’s great to hear that this is being considered. Has there been much movement on this?