I sort of touch on that in the slides. If the smart card didn't take difficulty into account, you could bypass it simply by generating a bunch of low level difficulty transactions, with minimal effort.
You would need to also model how much difficulty each proof of work should have. It's a risk, as some major bitcoin collapse could leave you effectively with no ability to decrypt your saved package. The closer you're able to accurately model how much difficulty there will be over the time period, the more secure the system will be. The more lenient, the less power an attacker would need.
Even 5% of the bitcoin block network's hashing power for the duration of the life of the secret would represent a substantial amount of computing power.
You would need to also model how much difficulty each proof of work should have. It's a risk, as some major bitcoin collapse could leave you effectively with no ability to decrypt your saved package. The closer you're able to accurately model how much difficulty there will be over the time period, the more secure the system will be. The more lenient, the less power an attacker would need.
Even 5% of the bitcoin block network's hashing power for the duration of the life of the secret would represent a substantial amount of computing power.