You can think of increasing the collateral requirement like decreasing the leverage and thus the ROI.
There are two ways to adjust short supply: provide constant leverage (2:1) and variable cost (interest rate) or provide variable leverage and constant cost (0 interest rate). There comes a point where the leverage gained by going short is not worth the risk.
If you want the interest rate model to work you require two prediction markets working at the same time. This approach is potentially viable because any attempt to "short BitUSD out of existence" would end up in sending the interest rate prediction market through the roof.
The demand for BitUSD is highly correlated to how well the peg is holding up. If it does not hold up well and has a wide spread then the demand will be low. If it holds up very well then the demand will be very high because it is a proxy for the dollar.
The "dual market" approach may be the only viable solution that can operate without a price feed. It is very challenging in deed.
Given a price feed we can prioritize shorts very effectively by collateral.
Why not a third, variable leverage and variable cost.
Have the collateral base the effective range of what interest a short would be required to pay. Rank the shorts by the amount of interest they are willing to pay. Shorts willing to pay higher interest take priority. higher leverage requires higher interest.
Average the % interest from actual trades and this is your base interest requirement.
I'm very much more in favor of the proposed interest rate models around Gulu's line, and agree that having a working client is by far the most effective, most needed work.
So I was proposing a fixed up front fee, rather than interest, since it is easier for a person to understand what their costs would be up front. However, from a game theory perspective it shouldn't matter.
Regardless, the problem with letting both things float is that you end up with a "two dimensional" market, where shorts and longs have to agree on two parameters instead of one. This adds complexity, and provides a greater opportunity for shorts and longs to not agree and sit out of the market. That said, the more I think about it, the more I think this is required.
The entire point of the peg is that we want price of btsx / $ to always equal btsx / bitUSD. So we really don't want shorts and longs competing with each other on price. They need something else to compete with. The feed encourages them to stay near the peg, but it won't respond to volatility in the btsx / $ market fast enough. People just sit out of markets if they think they can't make money, and all of our attempts so far with feeds and only letting shorts trade in a range is just making people sit out of the market.
I think we need to let the fee or interest paid float, as well as the price. Then the consensus of the market will encourage the fees to vary, rather than the price straying from the peg, but it still gives somewhere for the "market pressure" to go without forcing them out of the market entirely.