Advice to Aimless, Excited Programmers

I occasionally see messages like this from aimless, excited programmers:

Hey everyone! I just learned Erlang/Haskell/Python, and now I'm looking for a big project to write in it. If you've got ideas, let me know!

or

I love Linux and open source and want to contribute to the community by starting a project. What's an important program that only runs under Windows that you'd love to have a Linux version of?

The wrong-way-aroundness of these requests always puzzles me. The key criteria is a programing language or an operating system or a software license. There's nothing about solving a problem or overall usefulness or any relevant connection between the application and the interests of the original poster. Would you trust a music notation program developed by a non-musician? A Photoshop clone written by someone who has never used Photoshop professionally? But I don't want to dwell on the negative side of this.

Here's my advice to people who make these queries:

Stop and think about all of your personal interests and solve a simple problem related to one of them. For example, I practice guitar by playing along to a drum machine, but I wish I could have human elements added to drum loops, like auto-fills and occasional variations and so on. What would it take to do that? I could start by writing a simple drum sequencing program--one without a GUI--and see how it went. I also take a lot of photographs, and I could use a tagging scheme that isn't tied to a do-everything program like Adobe Lightroom. That's simple enough that I could create a minimal solution in an afternoon.

The two keys: (1) keep it simple, (2) make it something you'd actually use.

Once you've got something working, then build a series of improved versions. Don't create pressure by making a version suitable for public distribution, just take a long look at the existing application, and make it better. Can I build an HTML 5 front end to my photo tagger?

If you keep this up for a couple of iterations, then you'll wind up an expert. An expert in a small, tightly-defined, maybe only relevant to you problem domain, yes, but an expert nonetheless. There's a very interesting side effect to becoming an expert: you can start experimenting with improvements and features that would have previously looked daunting or impossible. And those are the kind of improvements and features that might all of a sudden make your program appealing to a larger audience.