It's pretty impressive that you can create a model with no risk to lenders.
The only thing I wonder is whether we would be better off introducing the widely popular, currently used model but with more conservative initial and maintenance collateral settings? (Rather than a more complex, unfamiliar and also perhaps less favourable/uncertain for borrowers model.)
The lending prices you see on Poloniex and BitFinex are probably taking into account an exchange failure risk of >20% per annum. So while our Black Swan risk to lenders might be higher, our DEX failure rate is much lower so prices may end up being fairly similar to exchanges.
They do bear risk of leaving an order on the book. No different than a bank taking risk of you mailing in the keys rather than payment.
If even one person is left holding the bag for a default it makes the entire block chain look bad. Block chains must have higher standards.
It appears to be setting higher standards for one party in the trade at the expense of the other (poorer terms for borrowers) which might make it hard to create a successful market.
Similar to BitAssets forced settlement at 1-1. While great for providing 100% certainty for the longs, the burden & uncertainty this creates to shorts is part of the reason we see such a high BitAsset premium? Lowering forced settlement considerably and introducing a maintenance collateral requirement level above which you couldn't be force settled would still provide the majority of the benefits but create a more successful, active market imo. (The current setting also allows longs to profit when the market moves slightly in their favour at the expense of shorts or even gain a few % by manipulating BTS price fairly cheaply at forced settlement time.)