The security of Bitcoin depends on financial incentives, specifically the assumption that the participating parties try to maximize their financial gain and that they act rationally to achieve that goal. The security can fail in cases where that assumption does not apply. A double-spend via a 51% attack can never be dismissed as impossible.
Specifically financial stability or long term financial plan. Double-spend via 51% attack could have good
short term financial incentives if :
1. Network hashrate is relative low
2. Attacker have access to 51% or more hashrate (compared with network hashrate), either by own or rent mining hardware.
3. There's vulnerable exchange/service with relative low minimum confirmation and fast time to process withdraw request/send digital goods.
51% attack on BTG is the prime example.
To be even more clear:
Defrauding people by re-org attacks on blockchains
is not double-spending because it doesn't inflate the coins in circulation, it is just a fraud!
- Collided miners/pools could be spotted and being put behind bars!
- Smart people should wait for enough (weeks maybe) confirmations before releasing large amounts of assets in exchange for digital tokens as long as they are concerned about such a threat.