It was a big mistake to design version 5 only for later OS's
Well, maybe you should start your own antivirus company/business if you know best.
The fact is, however, that it wasn't really an option to design version 5 for both Win9x and NT. Either there would be no version 5 at all (i.e. only small changes would be made, let's call it 4.9) - or two separate products would have to be created and maintained, which would be
very expensive (money / time / effort...). I certainly don't believe that the current type of avast! 5 GUI would (if compiled as ANSI and without any missing dependencies) run on Win9x - with its notorious problems with GDI recources. We couldn't even move to a new version of the development tools (MSVC) - which is somehow more secure, generates slightly better/faster code, and makes it possible for us to work on a 64bit version. Even if you write a simple "hello world" application in new MSVC, it won't run on Win9x.
Saying no one uses old OS's is incorrect. It alienates customers / potential customers to drop support for old OS's, so in the end, money is lost, not gained
I'm afraid the math is quite simple, and it doesn't support your conclusions.
The ratio of Windows 95/98/ME users of avast! in the past 7 days was... 0.3%. I hope you undertand that bringing more features and more protection to 99.7% of users gains more money (because more users will like and use the product) than supporting the 0.3% minority.
If it were possible to support Win9x painlessly, we'd certainly do that - but it's not the case, the price would be very high, in all respects.
It is a fact that I write a ton of software that works on everything from 95 to the newest Windows OS's and it is not difficult to do so for the vast majority of applications.
What kind of software do you write? Low level applications with their own drivers? I doubt that.
Sure, if you write simple user-mode tools, then it may seem "easy" for you... but for a low level application, with parts of code being updated almost daily as part of the virus definitions (so you should test all changes on all the supported operating systems)... certainly not.
I would ask the company to keep the virus definition updates going
Virus definition updates will certainly go on until there are corresponding vertion 5 replacement for all the editions (server, network, ...)
But the pressure will certainly be to end the support reasonably soon after that - because even supporting v4 virus definition updates is quite expensive; and I'm not necessarily talking about money, but about the time and effort needed - which could be spent in a better way. Simply said, if v4 virus definitions wouldn't have to be maintained, avast! 5 would have better detection today, detect more malware and provide more protection - because the virus lab could focus on that, instead of spending their time on duplicating parts of their work.
So again, the question stands: do we want to provide better protection for 99.7% of our users, or for the other - 0.3% - part?
(and I'm sure the ratio will grow more and more).