I disagree with some answers
Q) Why is there a problem implementing pegged assets?
A) The primary answer is that there can be no redeemability to the real asset on chain, for obvious reasons (since fiat/gold/silver/oil are not digital in the first place).
Yes, that is the challenge
Q) Why is bitshare's current solution inadequate?
A) Those who borrow bitUSD from the system are at much higher risk than those who buy it from the market because the borrower must actively maintain his collateral and can get margin called by the system if the feed price moves enough
What you are describing is actually fine and not an inadequacy. The problem, as I see it, is the lack of symmetry in the arbitrage opportunities. The blockchain provides an arbitrage opportunity in the case of an oversupply of bitUSD (forced settlement), and rewards the trader that pushes the market closer to peg while reducing oversupply. There is no opposite action. The equivalent would be the blockchain offering all collected fees for sale at settlement +10% or something like that so more bitUSD are automatically put in to circulation when there is an under-supply. A long knows they can always sell at the feed, but shorts have no such assurance as to at what price they can buy.
Q) What effect does this have on the price?
A) The borrower is forced to price his risk into the sale price of the bitUSD he borrows from the system - this leads to a situation where the price of bitUSD will always trade at a premium compared to the feed price, this damages the viability of the product as a whole.
This is fine, it is designed to be at a premium. The problem is that it is impossible to price this premium since there is no way of predicting what the premium will be in the future since there is no explicit reason (no arbitrage opportunity) for shorting.
It's helpful to think about what the "price" of a smartcoin should be. The price occurs at the balance point where longs are no longer willing to pay for over-valued smartcoins and when shorts are no longer willing to accept more risk. Since the shorter assumes both "normal" risk (variance in BTS:fiat markets) and a new network risk (will the premium for smartcoins improve or worsen in the future?), the short's risk is very difficult to quantify. Having an network backstop will help shorters to quantify risk, and thus keep the price of a smartcoin closer to peg.
Q) Is there another design which doesn't have the same biased risk profile, or is this just a natural consequence of not having redeemability?
A) Discuss
Bias risk profile is fine to have. The problem is when there is no equilibrium in risk. Since there are no arbitrage opportunities for shorters, equilibrium is difficult to quantify or obtain. If fiat:bitFIAT on/off ramps exist, that would enable arbitrage, or if the blockchain provided a backstop (automatically shorting at settlement +10%), that would help too.