I should take some notes here... Because I WILL finish a project before the end of this year! :p

It is also a good idea to not keep your browser open if you want to progress a lot on your project one day, unless you need help. Else you might end up procastinating.

What I do is really think through the project first. Is this something you'll have the time to finish to completion? Is it something that's creative and fun? These are often things people forget before the come out with a half-assed project, and I certainly made this mistake many times in my calc programming career.

Then, try programming out a simple engine. What does your game really revolve around? Is it possible given the limited capabilities of the system you're running it on? Do you have the expertise or access to help that can bring that engine to completion? Sadly, I've had quite a few projects that I just didn't know how to program.

Once you've got that down, the hardest part is imminent. Designing your game. Levels, storyline, etc. What I've done in the past is take a laboratory notebook with graphed paper and fill page after page with tilemaps. It's often much easier to plan out on paper, but you can program a level editor if you don't feel like doing that manually.

And then the easy part - optimizing and packing! Do some fine-tuning; make that creation really shine. Lots of games fail because they just don't have that spark. I try to put something interesting in my projects, whether it's the graphics or just some special particle effects. Share it with your friends, so they can tell you what sucks and what's good. Once you know people are definitely going to like your project, post your creation on the interwebs, and hopefully something will come of it!

As everyone is different, each person will have unique things that work better for themselves. I'd personally say that at least some degree of planning (and research) is good. That way it will be easier to reach your final destination. It also helps if you have a fairly clear idea of what you want to make when you set out.

I'd also go on to say do not under estimate the value of letting others test it as you go. It gives you the chance for someone outside of the project to give you feed back. Not only can the help finding bugs, but they can also let you know what doesn't work out gameplay wise or if something just isn't fun.