This is not revolutionary................the German site that analyses HijackThis logs has been using the principle for ages.

I am currently testing Online Armor, which invites you to feedback what you think is safe. I think that Process Guard and Mamutu do much the same?

it harness its power with your friends, family and the broader community by collecting information on what is safe and not safe from its community.
Sorry, I have seen far too many infected computers belonging to friends, family and the broader community to place much credence in what they think or do

This method of detecting malware will revolutionize the way virus detection and prevention is done because it reduces the publishing delay to zero.
No it won't...............all this stuff does is check downloads and processes (if you activate that bit) against listings of malware. It does not detect 0-day items or anything not on the listings. Traditionally this kind of stuff isn't much good against trojans, backdoors and the like..........particularly if the user is "clicky clicky" and loves their P2P and Warez sites

A lot of software would require analysis to determine if it is, in fact, malware. That takes time, so there is a publishing delay.

The major problems as I see them are:

1. Malware authors change their products almost daily.
2. The stuff I have used mostly comes back with "we don't know" assessments.
3. Users..........you just cannot automate their intelligence.

If someone in the community encounters a threat, everyone else will get protection against that threat.
Only if that threat is identified, and who or what identifies the first instance?

While you are online, your system always queries the cloud to determine whether something is malicious before making a decision.
And how would they or it make such a determination?

Using the internet will doubtless get warnings to people sooner, but this sure as hell isn't a silver bullet.

My personal feeling is that the traditional AV/Malware outfits need to sharpen up their acts. Basically they not only look at detection and analysis but they also look for cleaning and/or repair. That takes way too long, so an approch that prevents bad stuff happening first is preferable. OK, I know that goes against their traditional approach of not providing detection without removal/repair, but I consider that approach to be obsolete these days.

Hey, if you don't get infected you don't need to clean or repair?

Another problem I could see would be if you were on dial-up or capped broadband.

I can see this maturing, but I think that it has a way to go yet.