Valid point.

I won't explain https://en.wikipedia.org/wiki/Occam's_razor just read the linked article.

Regarding our case, in cryptography all security assumptions are reduced to the minimal set of assumptions. The foundation of Byteball security is based on assumption that majority of the witnesses can be trusted. An obvious way to implement an oracle in such system is to require majority of witnesses to sign an oracle's statement. The link to the BTC-oracle tells that cryptographic proofs based on Merkle tree are used. This is overengineering, because if the quorum of witnesses states that a Bitcoin transaction is confirmed then we don't need to do Simple Payment Verification (which utilizes Merkle tree approach).

Even more, if we decide to create an oracle for a PoS-based coin then we can't use Merkle trees, we need to verify the complete history of payments. In other words PoS coins cannot use SPV described in Satoshi's paper. So, instead of using Merkle trees for PoW coins and something else for PoS coins we can just combine everything if we rely on the witnesses.

I hope I explained why I called BTC-oracle "an overkill". I tried to be as ELI5 as possible.