When faced with a new project and a deadline, how do you know how long to assign to each task if this is a new endeavor? What tricks to the trade are there for putting together a reasonable timeline?
Try to break the task into its component parts. Then figure out if each component will take you an hour, half a workday (ie 4 hours before or after lunch), a whole day or a week.
Alternately, look at each day or half day increment and try to figure out how many component tasks you can complete in each increment.
I’m a scheduler. There are resources for estimating out there, but I’d need to know what sort of project it is before giving you specifics.
My number one piece of advice is to ask the people who are doing the work or someone who has done something similar. Depending on who it is, what they do, and how well you know them determines the size of the grain of salt you should take it with.
msmith537 is right about developing a Work Breakdown Structure (WBS). You can do this initially without regard to duration, just keeping sequence in mind. Again, ask people questions, don’t think you can do this yourself.
ETA: If you have a deadline, you should definitely make sure that’s the last milestone and that all your tasks are linked either to that or a succession of tasks leading to it. This will give you a clean critical path.
An Arky, I wish you’d go over Critical Path Method while you’re up here at the podium. I need a refresher!
Oh, and just for you:
\m/
One big problem I’ve seen with traditional scheduling software is that they don’t handle rework very well. If you are putting on a play or building a building (the typical examples in the tutorials) this isn’t much of an issue, but when you are writing software or designing a processor it is. You may hit your target for the first release, but how long debug and test takes depends strongly on the quality of the design, which isn’t something you know or can plug into a schedule.
A lot of scheduling is figuring out how to assign resources to a task to bring the completion date on track. If it is just you, most of the complexity is gone.
When I managed software, my heuristic was to ask the programmers how long it would take, double it and add two weeks. Worked pretty well.
CPM is where you include all the activities required to complete the project (the WBS), the duration of each activity, and the dependencies between the activities. The scheduling software (hopefully you’re not doing it with pen and ink) will calculate the longest and shortest path to the completion of the project, depending on whether you use the “as soon as possible” or “as late as possible” constraint type for each task, which in turn tells you how much float you have. A task is on the critical path if any delay would result in a delay of the end of the project. A manager can determine if activities can be performed in parallel, increase resources, etc.
Need coffee now? I sure do.
Yup. Of course it’s possible to have multiple critical paths, but that makes my head asplode.
Assigning resources can be tricky, and I advise that you do NOT let the software level your resources for you.
Your heuristic is about right. 
For me, it depends on the size of the project. Really big ones usually need to be broken down into components, with me working on those components for a set amount of time or breaking them down into sub-components. Alternatively, I’ll chunk out all the parts of a project, get the little but annoying bits done first, then strategize on the larger portions. Developing that strategy actually gets me more energized about the project in general and provides me with better focus. If I don’t plan out a project, I procrastinate because I don’t know where to start.