I'll have to add my words of congrats to DBS - well done on this thread. It's an excellent read.
I'll add in my thoughts on recipe generation, especially my APA/AIPA.
I've pretty well done recipes from the start of my AG career with a look at other recipes for inspiration. I've only brewed one recipe, which is Tony's Monteiths' OA clone. Excellent recipe.
Having said that, as a rule I'm fairly conservative as a brewer, style wise, and my more out there brews (good and bad) have been either flukes (such as my Golden Strong Pale Lager, Smaragd hops blind, an overboil I forgot to dilute and US at about 14 degrees - which placed in QABC) and other more interesting brews (my Rauchroggenweiss which has a 19.5 point muddle mess).
My experience? Stick to a style. AABC style guidelines are the best guide there is. Brew with these in mind, and you won't have a bad recipe (though there will always be some improvements).
If you want to do something experimental - have a clear picture of what you want to do, and why. My Rauchroggenweiss is a point in question. With the QABC, you have to write what you were achieving in a category 18 beer. It was a waffle. Good indication that I wasn't doing any good.
Contrast that to a black IPA - someone wanted to brew an IPA, but a bit of roast for colour and to dry it out a bit. Winner. American Brown? One wants to make an English style brown but one only has American hops. So one adjusts for that.
But - You are unlikely to invent a new, you beaut style of beer. So brewing a hybrid beer or a 'new style' just to create one will leave you quite likely with a batch of very average beer. Some want to push that - as a type of experimental art and are happy with a high risk of epic fail. Most want to brew a nice beer and will take a less 'out there' method.
As for recipe formulation (apart from the above advice on experiments) - I have never gone wrong with 3 or less malts. One base, 2 specs (or a small amount of base used as spec) to add the dimensions necessary. One base one spec does well. I do find smash beers a bit one dimensional, but don't go to the other extreme of putting 20 varieties of malts in a beer - it just ends up a muddled mess. 2 or 3 is great. My ratios are generally 90/10 split on base to spec (for 3 varieties) and 95/5 split on 2 varieties.
Hops. Again, I keep it simple with additions. If the AABC style guidelines say "focus on early bittering addition, rather than aggressive late hopping", then it's a 60 minute addition with maybe a small 30 minute. If it's an APA, then a 30 minute bittering addition (15IBU) and a 10 minute addition (15-25IBU). AIPA, the 30 minutes becomes 60 minutes. I have pretty much got the ratios and numbers down to a formula.
Jury is out on me whether you can have too many hops. Not enough - well that depends on style. If it's an English Standard bitter, one hop is fine, keep it in style. With an APA, I've done up to 7. Sometimes 7 is better than 4, sometimes 2 is better than 4. It all depends.
Whether it works or not is a complex art - it can work or it can be a waste of perfectly good hops. Apollo, Belma, Cascade, Citra, Amarillo, CTZ and Mosaic are in my current brew. It works and isn't muddled. It could have been though. But sometimes I'm better off with 2 or 3 (say Nelson Sauvin, Cascade and maybe a touch of Galaxy). With my APA/AIPA - I find I need at least one 'dank', 'spicy', 'pine' or such American variety to contrast the frutiness of the usual C hops, Amarillo and the newer fruit varieties.
Again, this is gleaned from use, but sticking to my pre-determined style formulas gives me some boundaries to play in, whilst ensuring I never make a bad beer.
Sorry for the crapping on, it's the best I can do turning a qualitative thing into some sort of quantitative definition.