1. It wouldn't fly with a guaranteed buyback of 10% - to get believers to depositors you'd have to guarantee full value of their investment if the ponzi you were betting against closed down and paid everyone out. Now as long as you didn't touch original invetsors' funds you wouldn't be running a ponzi (as depositors would be being paid from investors cash - not from the cash of fellow depositors). So if the ponzi ran too long and investors funds were near exhausted you'd have to force buy-back at original value all deposits - then they could laugh at you and deposit them in the original ponzi.
Which is why you must keep the ponzi+ considerably smaller than each original ponzi. The ponzi operator has an incentive to maximize his returns, keeping it running so long that he loses 90% of his potential profits by paying almost all of his deposits as coupons doesnt make much sense. The investors in the ponzi have no influence over how long it runs.
2. What happens with investors' funds? Either they sit around in some escrow account earning no interest or you use them to make profit. But if you actually try to use them to make profit then they're no longer reliably there to back your payments.
Investors money would just sit there and slowly be depleted for as long as the ponzi runs to pay for coupons. It wouldnt earn interest obviously, but it would assure depositors and would earn investors a lot more than any other credible investment if/when the ponzi collapses.
3. The scheme if it succeeded could become a victim to its own success. Say some new obvious ponzi starts up - so you run a company paying slightly more. As your company pays more, is transparent AND has secured funds to pay investment there's very little risk - so everyone would invest in your scheme rather the ponzi. The ponzi would thus get little funds in (and so little liability out) - whilst you rapidly depleted investors' funds
Thats why you have to limit its size to a small fraction of the original ponzi. Just dont issue more than x% bonds as the original ponzi. I suggested 10%.
The big problem with it is the sheer quantity of investment you'd need to have to make any impact - investment which couldn't really be utilised for anything if deposits were to be properly secured.
You would need only as much capital as the amount of bonds your are selling to guarantee the buyback at face value if you lose your bet. The sale of the bonds themselves generates the rest, just like in the original ponzi. In fact you would even need less if you assume you are betting against a rational ponzi operator, since he wouldnt pay out 100% of his bond sales back as coupons, but only a small fraction of it.