Price feeds do TWO things:
1) They prevent market manipulators from "selling a BitAsset to 0" and stealing all value from the BitUSD holders.
2) They prevent market manipulators from "buying a BitAsset to infinity" and triggering a short squeeze / margin call or "black swan" event and thus stealing all of the collateral.
From this I believe you are inferring that price feeds do serve an essential purpose for setting some market limits. So regarding the key question of my OP, would you agree they are a permanent feature of the current model?
The secondary question to be considered then is how trading engine limits are positioned around this price feed...
The critical limit in my mind is the price limit for expiring shorts, as this sets the "contract settlement terms" (continuing my OP analogy) on which bitAssets are valued. I wasn't clear from your post on your view on this, but I think the price limit for expiring shorts should be at exactly 100%. Otherwise it leads to one of the two following outcomes - lack of assurance to bitAsset owners that shorts ever need to pay full value, or conversely, the ability for market manipulators to steal collateral from shorts to the extent the limit allows, as per this...
You can define these three properties (call price limit, short price limit, and turn over interval) and the system will work with very loose limits and very long intervals because market makers can charge for liquidity to bridge the gap. In fact I would argue the limits should be very wide ( +/- 25% ) and the interval very long (1 year +) which is different than what BitShares has today. I think that these settings should be tunable by delegates because the parameters that make sense in a immature market do not make sense in a mature market.
BM, for clarification, are you suggesting that expiring shorts might be forced to cover up to a bitAsset premium of 25% over the feed price? Or is the +/- 25% just referring to the price at which new shorts can be established?
The concern I have with expiring shorts being forced to cover beyond the feed price is they are at the mercy of longs. For example, a large player or several in collusion could buy every unit that comes up for sale below the limit (even at premiums), and force shorts to buy them back at the limit. This would make shorting very risky and subject to attack. That's why I think expired shorts should only have to pay to the price feed and never more.
A 100% limit is still also required on new shorts so that they can't undercut sell orders and roll their shorts indefinitely, at least on the current bitAsset model where we have individual shorts.
One has to be careful with the turnover interval, because the longer it is, the larger the bitAsset discount that will be allowed before arbitragers step in. Market-makers will charge for this liquidity in the form of a return on capital over such period.
What do you think of this analysis?
[Edit: edited original comment on price limit for opening shorts, which might suggest less scope to widen limits.]