When is there an incident?

I just wanted to post this as a question to those that do read this blog, all 2 of you :P. I had a discussion with a security admin the other day. They wanted me to take a look at their incident handling document. This document outlined the steps that they would take in the case of an incident. Now don't get me wrong, the document was spot on I believe. It was well written and you can tell a proper balance of technical and informational data was found. What this did bring up in my mind is; When has an incident, specifically a compromise, happened that a process like this needs to be put into action?

I realize there is a balance that needs to happen because if we did this same routine for every system infected with a virus, management would probably start to not trust things are going well (little boy crying wolf). What about a bot though? If you are not familiar with bots, you can read previous posts I have put on this blog or just Google the term and you should run into a ton of information. Long story short is that bots are used to control systems. The problem that I see is that a lot of companies downplay the significance of a bot. Just because at this time that bot is only popping up ads on your PC doesn't mean the attacker has any less than full control of your system. In my mind, a party outside of your network, often unknown to you, has full control of one of your systems. That sounds like a compromise or incident to me. It only takes one update from the bot's command and control center to turn it into something much more horrifying.

Now there are controls in place like IDS and IPS systems which can often block and alert of the existence of such a software. This is a good thing. The question is though, should this be treated like an incident of compromise or should it be quietly removed and cleaned up because it was caught so early? I guess a third option would be to have a non management alerted incident handling process in place as well. Not that we want to cover these tracks, but for the security admin to keep track of but possibly release at some quarterly meeting saying "we had x many major incidents and y many minor incidents". It's an interesting thought to find that balance. I would love to hear some opinions.