So you'd rather have a definition list of thousands upon thousands of extinct dos viruses to give you a sense of security? Hehe... Cause thats whats in a goodly many of the databases in some of these companies...
Now with that being said, does Avast need improving? I personally feel it needs a lot of improving. It does miss things, its database needs definitions fast for some real threats out there. I'm not happy with the speed they update their database from submissions - but thats the way it works. I like Avast, but i'm waiting to see what 4.5 offers before I climb onboard with it, or send them my credit card number. I'd like to send in more samples of what it misses, but based on the ones i've already sent not being added, I gather they are considerably behind, so i'll give it a few weeks and re-evaluate then.
PS: I think the future isn't in defintions anyway, its in things like Code Emulation, Heuristics, Artificial Intelligence, and smart detections. Theres only a couple products out there now to even come close to offering this, and they can run with hardly any definitions, and detect a godly amount of threats. I'm running one right now as a matter of fact, and its quite impressive to see it in action.
Definitions are great, but I think they need to be best used to back up heuristics or code emulation detections. McAfee has made some nice strides in behaviorial detection with their Enterprise 8.0i beta, which is probably why Microsoft seems to be buying McAfee. Looks like their Microsoft-AV will be RAV+McAfee, revamped. So they will have a huge database of threats, and some behavior based detections.
So who knows, maybe in a year from now, we'll all be running Microsoft-AV, and the point will be moot?