Committing to a cyber attack is a pretty grey area as far as 'Acts of War' versus 'Pure Intelligence Gathering' is concerned. To exemplify this statement, consider a scenario:
US intelligence hears a report that a foreign leader was rushed to a hospital recently, but the government in that country is staying mute on what is wrong. In the interest of intelligence gathering, US Cyber Command decides to snarf hospital records from all capital city hospital networks in an effort to chase down rumored conditions (if a patient was admitted with rumored conditions at the right time, increased likelihood of meaningful report, "Like" the source, or whatever it is US Cyber Command does with good intel).
The trouble with the scenario? Breaking into a hospital's patient record system has the potential to deny service to that system for the hospital. In turn, it could mean slower patient access to care, and could actually cause deaths. If the country in question determines that the US executed the attack, could it be considered an act of war?
Congress, at the very least, would want answers. One of the first questions Congress would ask should be: "What are your rules for engagement and procedures for determining worst-case-scenario in a cyber intelligence gathering mission?"
Cyberwar Doctrine is currently more of a CYA for government hackers so that they can show the purse-string holders they are being responsible. In the end, it's a good thing: better to have too little hacking than too much, and better to err on the side of, "we could kill people by accident, maybe we shouldn't do this."
No doubt various countries are developing their own internal Cyberwar doctrines, and no doubt there will be a few major incidents of power outage/etc before countries sit at the same table and hash out the differences in their doctrines, spell out what constitutes war requiring physical retaliation.
Welcome to the 21st century.