Most developers (and all should) use version control.

It’s especially easier if they are strong in GIT(http://git.or.cz/), which allows for easy merging of branches. In theory, this could mean that you could allow for different version of the same app, but as you find bugs and exploits and patch them up, you can fix multiple different versions.

Of course, the overhead is huge for the developers.

* Brent has mentioned, in the video from An Evening at the Adler, that he uses Frameworks to keep both NetNewsWire and NetNewsWire Lite much easier to manage.

Something I hope is that all application developers will want to be as “nice” as you’re stating.

Mike: I definitely agree that most users will want to take advantage of bug fixes, security patches, and the like. The biggest worry isn’t about getting new features or the application getting refinement - the biggest worry is when features disappear (like the example I gave with iMovie).

About Viddler: This is exactly why I wrote this post. We have a fairly unique position that a portion of our “application” or service is seen by people that never visit the site. The player. Our player is under constant development. Without checking my guess is that the current build that is being seen right now is just over 400+. We’re constantly refining the underpinnings, and we sometimes adjust the interface as well. So far, at least not to my knowledge, we haven’t lost any features. But I worry about the day that we do, and so this question popped into my head!

Perhaps we’ll do some experimentation with our awesome beta team on Viddler, who tests new things for us, and I will report back here once we see the results of that.

On the other hand, when it’s a mature site I can see the point you’re asking about. What I think happens, if a web app/service is going to change as dramatically as something like iMovie (which really should have been given a different name in ‘08), is the branching off into a separate site. It’s so easy to EOL a website by just stopping any further development. When an idea changes so drastically, I seem to recall that most developers will simply let the site languish — or sell it — and move on to another site/service. Then it’s the choice of the visitor/member/user if they want to join the new site or stick with the current one, just like you’re suggesting.

Why don’t you check the viability with Viddler? Is this something you guys think you could do?

It’s a matter of feasibility and providing “what’s best” for your users.

Firstly, at least for social applications, to provide different code bases and database configurations for different types of users would likely not even be possible in the world of traditional DBMS’s, let alone what the hassle would be to run to different uniquely optimized code bases.

Secondly, does it really happen that often that web applications change so much that they alter how you fundamentally use the web service? Your two examples would be such examples, but people don’t do that. And if they did (in the case of Microsoft Office 2007 or the re-vamping of Google Reader), it would be at the disgression of the application provider to decide what they deem a best choice for the user. People will upgrade to new versions of software no matter how different they become because thats how we are used to reacting to new versions of things. If users of web applications were provided the option to upgrade in these extreme situations they would likely choice the upgrade path not because they think that’s best but because it’s what they think of as “default” in their mind. It wouldn’t make sense to impose such a question on the majority of users because likely the majority of the users that were afraid to upgrade would simply be because of laziness and fear.