That one's always struck me. At least if you squint, that would imply that blockchain's fraud-prevention strategy hinges on nobody ever using the blockchain to handle anything too valuable or important.
Yes, there is a limit to the security that a cryptocurrency can provide (like with anything else). The level of security provided is based on the market price: The higher the price, the higher the rewards, which increases the profitability of mining, thus leading to more miners, which then increases the amount of competition you'll have if you want to post wrong information. Since there is a fixed rate of new data which can be added, that's what defines the cost of inserting data to the blockchain.
An attacker with a lot of resources (>50% of current mining power consumption) could use their resources to post wrong information, but it would need to be profitable enough to make up for the immense costs. It's unlikely there would be such a situation where an attacker could recover the costs for such an attack. At best they would be spending a lot of money to cause a temporary network disruption as the "fair players" move to a forked coin or similar.
So, I can see that argument following for something like Bitcoin. Though I think that part of that is because cryptocurrency doesn't actually have any intrinsic value. So an attack on, say, the Bitcoin network would also have the effect of devaluing the very thing you're trying to compromise.
But it does seem to be something you need to contend with when looking for new uses. Most new ideas people are thinking up have to do with something that exists outside the blockchain itself. And that, not Bitcoin, is what Azure Blockchain Service was really about. The merits of cryptocurrency as an application of blockchain are largely irrelevant to this discussion.
My sense is that the general pattern for most other uses is that the number of potential legitimate participants in the blockchain system is likely be inversely proportional to the overall importance of the thing the blockchain is trying to manage. Everyone can get in on something like CryptoKitties, but there are probably very few natural participants in a blockchain network for managing nuclear fuel resources.
You can use it to handle anything whose value to an attacker is less than the cost of attacking the network as a whole right? May not be following though.
An attacker with a lot of resources (>50% of current mining power consumption) could use their resources to post wrong information, but it would need to be profitable enough to make up for the immense costs. It's unlikely there would be such a situation where an attacker could recover the costs for such an attack. At best they would be spending a lot of money to cause a temporary network disruption as the "fair players" move to a forked coin or similar.