I've often seen the malware situation described as a "war", and conventionally, wars are fought over territory.
What territory has been lost to malware?
Consider various integration points that are now routinely defended against usage, on the basis that the only things likely to use these, are malware. These OS "features" are now effectively "owned" by malware, in that legitimate software will trigger defence alerts if they are used.
Consider a number of ill-advised features that are designed to allow arbitrary material to automate the system, e.g. MS Word auto-running macros, auto-running scripts in HTML email "messages", \Autorun.inf processing on USB flash drives, etc. Today, these will typically be disabled, because the most likely use will be by malware. So Malware "own" that, too.
Consider several business models that involve messages, attachments or links sent by the service's site, such as email greeting cards. As malware can arrive via forgeries of such messages, usage is limited to those who are too dumb to know the risk they are expecting the recipient to take, which is a smaller and more limited demographic than when such services were first started. Effectively, these kinds of businesses and practices have been killed by malware.
Should we scorch and abandon some of this territory? For example, remove OS integration points that are hardly ever used by anything other than malware?
Should we assess likely future "ownership" before creating new technologies and features that are likely to be swamped by malware?