Writing Playbooks

I've written a few playbooks now, and I've looked through a lot of really promising builds for playbooks and some that are just atrocious. Having scoured the Barf Forth Apocalyptica forums and committed to memory many of the random details from Vincent about how he wrote the original 11 playbooks, I feel pretty confident in writing this step by step process for writing a playbook.

Step 1: Have an open-ended concept
This is key. Your concept is everything, it helps define what you're going to write about the playbook. But you can't cling to it like a bible. Your vision of the archetype can't be constrained and must be open to interpretation. The best playbooks have a concept which is clearly understood and open to enough interpretation that it doesn't force the character to behave a certain way. The archetype molds to the character, not the other way around.
Look at the core playbooks and you can see the archetypes clearly: the Brainer is a weird psychic, the Angel is a competent medic, the Driver is a car obsessive, etc.

Step 2: Balance isn't everything
Every playbook has a stat block that usually includes a "best" stat. In almost every instance, the stat blocks add up to a total of +3 (+2 +1 -1 +1 +0 = +3). In some instances a character starts with two stats at +2 and those break the mold, but those stat lines also have a consistent math of adding up to +2 (+2 +2 +0 -1 -1 = +2), the price you pay for starting with two +2s is one less beneficial point.
Also, every playbook should start with three moves, or things, that make it distinct. These don't need to be balanced in power with other playbooks, as long as the character simply has 3 things. Look at the core playbooks and they all have three things to start: the Gunlugger starts with a choice of 3 moves, the Driver has their car (1) the no shit driver move (2) and then a choice of a second move (3), and the Savvyhead has their workspace (1) and a choice of 2 more moves (2 and 3).
Again, there are some playbooks that break the mold, but these tend to be balanced within themselves:
the Battlebabe only chooses 2 moves to start, but they have a Cool +3 and that's their third thing
the Hardholder only has their hold and a gang, with moves that accompany each, however these are bigger than the holds and gangs other playbooks can get as improvements, plus the Hardholder explicitly never needs to spend money for food and lodging
the Operator gets gigs and moonlighting, plus one more move, but the Operator's gigs outnumber what other characters can get as improvements

Step 3: Scarcity is the name of the game
One of the key things when writing moves and improvements for your playbook is to remember the theme of scarcity. In other words, you can't get all of the options! Nothing is perfect in Apocalypse World and you're always left wanting a little more, this should apply to character improvement as well. If your playbook has 5 moves to choose from and you can choose 3 during character creation, then you should think about adding a 6th move, maybe even a 7th move, or limiting the "get a new playbook move" improvement so that there's one move you can't get.
Again, look at the examples with the core game:
the Skinner starts with 2 moves and can get 2 more through improvements, but there are 5 Skinner moves total
the Gunlugger starts with 3 moves and can get 2 more through improvements, but there are 7 Gunlugger moves total

Step 4: Peer review
Share your ideas with your gaming friends and ask them for criticism. Share your ideas in the forums and ask for criticism. There is no better resource for criticism than a gaming community, just remember to not let all criticism affect your work. When I first shared the Wolf some people noted the villainous nature of the archetype and said they didn't think it fit with the game, I ignored that criticism because FUCK YOU! Don't tell me how to play the game! Just because it wouldn't be fun for you or your group doesn't mean it won't be fun for me and mine. Those kinds of criticisms about the theme and tone of your work are useless. In a way, they are just an outsider reaction similar to that lack of trust one displays at the gaming table when somebody else is narrating the action.
But when somebody says "Hey, it looks like your math is wrong." or "Isn't this move just a weaker version of going aggro?" Those are valid. Listen to them.
Scratch that! Listen to everybody. Just keep your bullshit detector on a high frequency.
And don't wait until your playbook is finished to share it either. As soon as you have something to work with you should be talking about it and sharing it to see where the cracks are. Poke and prod at your work until you think it's solid.

Step 5: Play it!
Once you think your work is solid, you have got to play it. Get other people to play it if you can. Play it, play it, and play it some more. Because you will never have played it enough and somebody might still find something wrong with it.

That's all I've got. Good luck!