Re: 9,223,372,036,854,775,808 sha1 calculations
Re: "While true, the likelihood of doing this AND getting a collision is highly improbable"
Aslong as you can pad the document in addition to making required changes AND making a change results in a financial advantage for you of more thanUS$130,000, I would be reluctant to call this highly improbable.
Based on previous hashes, the discovery of collisions has lead to more weaknesses being found, and usually large collision spaces within a hash function that need to be avoided.. What costs ~US$130,000 today will likely cost less than US$10,000 within 5 year.
Does it mean we have to throw away existing SHA-1 hashes? Probably not unless the financial incentive to attack them exists.
Does it mean we need to start patching SHA-1 hash functions to address discovered weaknesses deploying SHA-2 and later hash functions now for verifying important documents or software versions? Definitely - mainly because if we don't it just doesn't happen * stares at MD5 hashes *
Someone mentioned financial websites - all current browsers already require SHA-2 hash functions on certificates since January 2017. The only real exception I am aware of is old and cruddy Java 7 (or earlier) apps that refuse to upgrade or to die.