It can be done as a soft-fork. There are other things, which are more likely to result in a hard-fork, for example running out of timestamps in 2038 (or later, because block time is sometimes handled as int32, and sometimes as uint32). Of course people will want to do that. But as usual, when you want to reach consensus even on soft-forks, there are many things, which are not implemented, because of lack of agreement. So, I guess people will want to do a lot of things during hard-fork, but consensus will be reached only for very minimalistic changes, and only that will be implemented in practice (and also it will be very limited, to imitate soft-fork as close as possible, and have as little points of incompatibility, as possible). It is very simple: as long as total supply is never infinite, by checking the supply in block number N, you can do the mapping between finite and infinite model. Which means, that if you change proportions accordingly, then you can reach the same economy in both. The difference can be simplified into this: in finite supply, users can see single satoshis, taken out of their accounts, and allocated for future mining rewards. In infinite supply, exactly the same thing is done, but coin amounts are preserved, like in fiat currencies, but they are worth less, than they were in the past, according to the level of created inflation. Of course we can. As long as only ECDSA is broken, and not SHA-256, every OP_CHECKSIG could require more conditions, than it requires today.
It can be done as a soft-fork. There are other things, which are more likely to result in a hard-fork, for example running out of timestamps in 2038 (or later, because block time is sometimes handled as int32, and sometimes as uint32). Of course people will want to do that. But as usual, when you want to reach consensus even on soft-forks, there are many things, which are not implemented, because of lack of agreement. So, I guess people will want to do a lot of things during hard-fork, but consensus will be reached only for very minimalistic changes, and only that will be implemented in practice (and also it will be very limited, to imitate soft-fork as close as possible, and have as little points of incompatibility, as possible). It is very simple: as long as total supply is never infinite, by checking the supply in block number N, you can do the mapping between finite and infinite model. Which means, that if you change proportions accordingly, then you can reach the same economy in both. The difference can be simplified into this: in finite supply, users can see single satoshis, taken out of their accounts, and allocated for future mining rewards. In infinite supply, exactly the same thing is done, but coin amounts are preserved, like in fiat currencies, but they are worth less, than they were in the past, according to the level of created inflation. Of course we can. As long as only ECDSA is broken, and not SHA-256, every OP_CHECKSIG could require more conditions, than it requires today.

It can be done as a soft-fork. There are other things, which are more likely to result in a hard-fork, for example running out of timestamps in 2038 (or later, because block time is sometimes handled as int32, and sometimes as uint32). Of course people will want to do that. But as usual, when you want to reach consensus even on soft-forks, there are many things, which are not implemented, because of lack of agreement. So, I guess people will want to do a lot of things during hard-fork, but consensus will be reached only for very minimalistic changes, and only that will be implemented in practice (and also it will be very limited, to imitate soft-fork as close as possible, and have as little points of incompatibility, as possible). It is very simple: as long as total supply is never infinite, by checking the supply in block number N, you can do the mapping between finite and infinite model. Which means, that if you change proportions accordingly, then you can reach the same economy in both. The difference can be simplified into this: in finite supply, users can see single satoshis, taken out of their accounts, and allocated for future mining rewards. In infinite supply, exactly the same thing is done, but coin amounts are preserved, like in fiat currencies, but they are worth less, than they were in the past, according to the level of created inflation. Of course we can. As long as only ECDSA is broken, and not SHA-256, every OP_CHECKSIG could require more conditions, than it requires today.

It can be done as a soft-fork. There are other things, which are more likely to result in a hard-fork, for example running out of timestamps in 2038 (or later, because block time is sometimes handled as int32, and sometimes as uint32). Of course people will want to do that. But as usual, when you want to reach consensus even on soft-forks, there are many things, which are not implemented, because of lack of agreement. So, I guess people will want to do a lot of things during hard-fork, but consensus will be reached only for very minimalistic changes, and only that will be implemented in practice (and also it will be very limited, to imitate soft-fork as close as possible, and have as little points of incompatibility, as possible). It is very simple: as long as total supply is never infinite, by checking the supply in block number N, you can do the mapping between finite and infinite model. Which means, that if you change proportions accordingly, then you can reach the same economy in both. The difference can be simplified into this: in finite supply, users can see single satoshis, taken out of their accounts, and allocated for future mining rewards. In infinite supply, exactly the same thing is done, but coin amounts are preserved, like in fiat currencies, but they are worth less, than they were in the past, according to the level of created inflation. Of course we can. As long as only ECDSA is broken, and not SHA-256, every OP_CHECKSIG could require more conditions, than it requires today.