I want to make really good websites.
Pet projects do really help in exploring blank spots and consciously mastering aspects you want to acquire.
I'm happy with the process and the result but what to do next with this product? Users start to use it and I don't want to support them because I want to continue building my skills, not end user products. Hosting for all these web apps cost some money too.
Should I just code it till I find it interesting and technically challenging and then just junk it and move to another one? Or release everything as Open Source and don't care about support?
The reason why I don't want to do support is because I want to specialize. There is too wide topic already so I don't want to wide it even further.
I heard about the idea that you shouldn't care a lot about what you working on during your early years in development because pretty much everything will be junk. So just try more. Is it the way I should follow?
What is more effective?
You don't say how technical your websites will be - are you looking to build pure HTML, more dynamic DHTML or web applications with server scripts?
Regardless, you should look at taking on pet projects that challenge you in a variety of areas. As you get more experienced with web projects, you'll find that you end up doing many tasks over and over for different projects.
A good start would be to take on pet projects to tackle 'common' problems (layouts, styling, user logins/sessions, persisting data etc) and then look to abstract your work to a series of components that you can reuse in future projects. This way you will build up a library of reusable 'widgets' that means you don't have to scrap everything and start again each time.
As you get more experienced, set yourself tougher challenges, and before long you'll have a considerable arsenal of sample code, if architected well it will be mostly reusable, and at the same time you'll expand your experience.
Good luck!