The price is mostly determined by information asymmetry - you don't know which debtors will be 'dry' and which will pay, but you can buy/sell a bundle of them and assume some average rate. If the sale was burdened with the right of first refusal (on a loan-per-loan basis) to a debtor who has much better knowledge on how much he can pay, then the deal is broken - collectors would be stuck with all the value-below-price loans while the value-above-price loans would get away for the listed price.
Also, it changes the value of your loan pre-sale - if my policies, in effect, tell you "well, you'll have a chance to get away with paying x% of the loan" then you won't pay more than x%, ever. If my policies tell you "I'll sell your loan for x%, but in that case you'll get screwed and pay 2*x% to the collector", then I have a reasonable chance to get more money than x% out of you before the sale.
It's quite likely that a policy "pay $1500 or I'll sell your loan to someone for $1000 who'll harass you until you pay much more than $1500" is rational and economically effective even though theoretically you're leaving $500 on the table; since it affects your decisions on when/if/how much to pay.
I am also not sure how debt is currently priced. But, it would seem that current credit score and other metrics might go into the pricing, such that there is more of an actuarial approach. In this case, the price wouldn't be driven as much by information assymetry.
So, combine a lender who is more informed about how much debtors can pay with debtors who have incentive to pay as much as possible to protect their credit, then add the fact that you're throwing out the middlemen (debt collectors). It might be a win-win for the creditor/debtor, with only the debt buyers losing.
If you're still just delaying and are willing to pay near-100% of the principal - then you might get some deal where "protect your credit" is a factor, but these aren't the sort of pennies-on-dollar deals the original article talks about.