In BBR's solution, the linear advantage that full nodes get does not solve any scalability issue (especially so for thin nodes) and introduces the trust / security model problem that is in no way a trivial pedanticry. Commiting the ring signatures with an additional "full" hash for each block would alleviate that problem, but still trusts "anybody" stores the signatures.

And to add to all that, open source is not the holy grail in code vetting. I will name three issues from recent memory that had different direct causes, but the same primary cause:

1. Heartbleed
2. Shellshock
3. Block 202612

The primary cause is "just because anyone can do it doesn't mean anyone will do it", both in terms of open source vetting and in terms of storing the signatures. It is a tragedy of the commons.