1) Write a book with power creep.
Depending on how much you give, this approach may make people happy or piss them off. If you're boosting underpowered concepts or giving things a mild, but universal boost (such as the Adventurer's Vault or Oriental Adventures) people are generally happy. If you go too far, though, you'll make people increasingly unhappy--and if you go TOO far with power creep people will just not buy your book people they know that DMs won't allow it at the table. Overpowered options get a LOT more bad press than underpowered ones.
2) Write a book with no or even negative power creep.
When you do this, you need to ensure that the book is designed specifically to appeal to aesthetics on some level, like the Stronghold Builder's Guide or the Monster Manuals. But when we come to player options, this is really, really hard to do. The first one being that basic game theory states that having more options makes your position stronger and things can interact in unexpected ways to create cheese loops. But your chances of this are still lower than intentional power creep.
The biggest drawback to this method though is that these books are usually the least popular of the set. Adventurer's Vault 2 just isn't as well-liked as the Adventurer's Vault. Song and Silence is probably the most laughed-at at the class expansion books. And Martial Power 2 will never hold a candle to Martial Power 1. I'll go a little more why this method is unsuccessful with the third one.
3) Release books with horizontal power creep.
This means that you release options that are just as powerful as the old ones but they don't intersect. This means that you release a Warlock class that's just as good as the Wizard but they have no overlap at all.
The benefit to this method is that it requires the least amount of supervision to write books for. You don't need to worry about introducing the next Incantatrix nor do you need to worry about writing the next Dread Pirate. The downside is that this method sells the LEAST amount of books if your players are already happy with this. Seriously, if people are already happy with four or five of the published classes, you need to give them a reason to use your shiny new one. Humans being humans, you can't just show someone something just as good and expect them to abandon ship; you need to make it clearly better. Which leads to power creep. Sigh.
The worst position to be in is what happened with 3rd Edition D&D. They were locked into a position where increasingly few ways to release books were viable. There was so much material that the new bits had to be special to get noticed, but there was also so much power creep that people were already on a veto-happy spree. There was just no room to go upwards.
So, how do you reconcile these paths? My proposal: at the beginning of the edition, you should first indulge in no or negative power creep. This is when you coast by on materials like the SBH or the MM2 or Forgotten Realms Campaign Setting. But this period should be very short. Like 3 or 4 months. The reason being is that your edition is still shiny and new and people buy the most books during this period. Next you do horizontal power creep. You should have been polling your player base and releasing the bulk of your horizontal options at a steady rate. If the power creep in the game has been mild it makes balancing and writing the material easier to do; you don't have situations where everyone is all
![Bored :bored:](./images/smilies/dozey.gif)
Next, power creep. Power creep should be actively planned for well in advance. People should measure the power of characters across the board at the beginning of the edition and then come up with an expected power level for characters when the edition is expected to run its course. Power creep should follow this curve. After you knock out all of the non/negative power creep options and end up with about the expected number of basic expansion options (races, classes, weapons, the like) you start releasing minute but still measurable improvements for players until the edition dies out.
When the edition does start to die out this is when you go back to horizontal power creep and release your Playtest materials. You examine the BASIC complaints about your ruleset (not complaints like 'the wizard is overpowered', complaints like 'there's not enough non-combat stuff to do' or 'people feel too dependent on magical items') and then release material that starts to address these complaints. The fixes that are well-received, like Book of Nine Swords, you incorporate into your next edition so people don't feel like they've been wasting their time.
Yes, this means that you can't plan your edition like it's going to go on forever. You actually need to determine a date for planned obsolescence. But don't tell the playerbase about it.
Basically, the model should follow 3E's from 3.0 to 3.5E. 3.0E released a lot of options during that four years period of time but characters at the end of 3.0E lifecycle by and large weren't all that more powerful than characters a year or so in (I'd like to say from the beginning, but pre-OA and S&F warriors were in an even worse shape believe it or not). 3.5E's decision to jump the gun by releasing a bunch of negative power creep options and horizontal power creep options wasn't a bad one, since people were still duped by the 'brand new edition direction' deception, but they milked this period for way too long. 3.5E's selective nerf period should've ended after a period of a year rather than going on for as long as it did.
At no point should you make people feel like they've wasted their time or you're wasting their time. This means that after a year and a half you should have released three monster manuals. This means that you also released all of your extra classes and races. But this also means that you can't release options that are so power-creepy that it invalidates materials of the old book. While this is fine to do for one iteration (like Adventurer's Vault 2) doing it twice or thrice (what happened in 3E repeatedly) burns people out and makes them not interested in new books.