Wow. This is a big topic. As someone who is deeply interested in this cooking movement, let me try and suggest a few things.
First, reading this article would be a very good start. In it, Ferran Adria/Harold McGee/Heston Blumenthal and Thomas Keller (not exactly a man known for avant garde cuisine) put forth the position that in essence there is no such thing as molecular gastronomy, it is merely a logical continuation of cooking techniques that have stood the test of time.
Bearing that in mind, I think what you're asking is 'how do I paint from the new palette that Adria has spearheaded?'
The answer is both simple and complex. At its most basic, MG is about playing with perception, and tries to look at food in a couple of different ways. First is what is widely known as deconstruction: taking a well known dish and reimagining it in order to play up a specific flavour from the dish, provide the same flavours with new textures, or provide a new spin on the dish. Even something as simple as substituting a pancetta cracker on a Caesar salad instead of lardons is fundamentally deconstructive in its approach.
What Adria does is out there at the absolute bleeding edge of cuisine. He views texture and flavour as separate things, and during their working sessions at elBullitaller and in the restaurant (which is, sadly, closing as a restaurant on 31 July of next year) they focus on presenting interesting flavour and texture combinations. Blumenthal and Achatz, by contrast, aim at providing more recognizable food and trying to make the 'molecular' processes in the dish more transparent.
So look at some of the things you make on a regular basis. How can you highlight a specific flavour? How can you change perception of the dish? As an example, I was thinking last night of making bite-size creme brulee; encapsulate the creme using the calcium/alginate method and set with gelatine, make spheres. Modifying Adria's technique for his pumpkin oil sweet (from A day at elBulli), one can then coat the whole thing with a crunchy layer of caramelized isomalt. Alternatively, use his method for adding a layer of caramel to an egg yolk. Voila... reimagined dish that everyone will recognize in flavour, but very different in texture and how it behaves in the mouth. Garnish and so on are then up to you.
But that may be going too far for you. So let's look at some other ideas. A restaurant I used to work at served a sous vide short rib that was actually two pieces of (bone out) short rib meat glued together with transglutaminase, allowed to bond overnight, then cooked sous vide for 24 hours until medium rare all the way through. Beyond that, the rest of the dish was grounded very solidly in classic French technique. To the guest, the 'molecular' aspect of the meal was entirely transparent.
Or consider the classic combination of smoked salmon/dill/lemon/cream cheese. You could make a foam of the cheese (to reduce the claggy mouthfeel while sacrificing no flavour), mix in encapsulated lemon caviar (to provide a textural counterpoint when your guest crushes it with their teeth), and dress the whole with a dill oil. Same flavours, slightly different textures... and it's all 'molecular.' You could go way out there and infuse the cheese with the flavour of smoked salmon, or make a smoked salmon cracker or whatnot, but I think that is going too far for you.
A dish I was working on at the restaurant (and have had to abandon for now, due to switching jobs) was a simple coconut cream pie, deconstructed into a napoleon-esque stack. But instead of pastry, I made coconut crackers with methylcellulose and coconut water. The pastry cream was made with coconut milk instead of cream (note: you need to leave the cream overnight for the flavour to develop properly; also ensure you cook it long enough to cook out the cornstarch). To be served with a toasted coconut macaroon of sorts (coconut bound with barely enough lightly whisked egg white to hold it together, baked until toasted) and a white chocolate and lemon sable cookie to provide the missing pastry flavour, and a little chantilly enhanced with Malibu rum. Again... molecular elements, but largely transparent to the diner. The focus was on providing the most intense coconut experience I could come up with.
At its base, MG is really about finding the truest and purest expressions of flavour. If you look at most MG-centric dishes, they focus on a single flavour with everything else on the plate providing a supporting or contrasting role. Not all that different from classic techniques, is it?
One of the other approaches to creating a new dish that Adria uses is to consider a single ingredient, and what works well with that. Exploring flavour compound databases is a really good way to go about this process. Here is one that I find useful, even if the design is stuck somewhere circa 1995. It is a method that Blumenthal uses extensively, which is how he came up with his salmon coated in licorice (not that far a stretch, given the affinity of tarragon and fennel for fish), as well as the sounds-bizarre-but-is-delicious combination of cauliflower and chocolate. Easy recipe: clean and cut your cauliflower into florets. Dust lightly with cocoa powder, coating each piece. Shake off excess. Season with salt and pepper, roast approx 15 minutes at 375. Sounds weird, yeah? Well, turns out that cauliflower and chocolate both share distinctive flavour chemicals. So they go well together.
At its heart, molecular gastronomy really is just the same classic techniques we are familiar with. The key difference is that it asks three questions:
- How? (do we make this better/enhance this flavour/provide a new texture)
- Why? (do we put these flavours together/cook things this way/follow these steps)
- Why not? (do it differently)
I think I will stop my rambling here. I hope it helps. And I will leave you with three incredible blogs (one of them courtesy of Michael at Herbivoracious).
Playing With Fire and Water - truly spectacular molecular-influenced cooking.
Ideas in Food - Wonderful. Molecular weaves in and out of what they are doing.
Studiokitchen - Questions a lot of sacred cows, aims at making molecular transparent to the guest.
Best Answer
First of all, the term "molecular gastronomy" is almost universally derided by those who practice it and "Modernist cuisine" seems to have become the accepted nomenclature. The book Modernist Cuisine has what I regard as an excellent introduction to the movement and drawing the analogy with other Modernist movements in other artforms. The Modernist movement was about rejecting the lineage of everything that had come before and re-examining basic questions such as "what is perception". Similarly, we can trace a direct lineage from Escoffier to Nouvelle Cuisine and Modernist cuisine seeks to break that lineage.
The best way I've come up with to describe the Modernist philosophy is "the use of technology to gain direct, precise control over temperature, humidity, pressure and texture".
When you're poaching fish in a simmering stock, you want to bring the fish up to 45C by putting it in 80C stock that's heated by a 600C flame. You control point in this scenario is the knob of the stove which is the second derivative of the temperature of this fish. What's more, your feedback loop is entirely perceptual and based on human sensing and control. Modernist cuisine starts with the goal of bringing a fish up to 45C and determines the optimal way of accomplishing that goal which is sous vide. With Sous Vide, you are directly controlling the temperature of the fish and you're doing it precisely because you replace humans with automated feedback loops.
Another example might be reducing a sauce. In this case, you manipulate the vapor pressure of the liquid through the indirect application of heat. What you really want is to drive water off but you end up causing many inadvertent changes as well. With Modernist cuisine, you use a rotovap to control pressure directly and vacuum evaporate liquids without heating.
Same with using a roux or cornstarch to thicken a gravy. Instead of indirectly controlling the texture of a sauce by picking from one of the dozen common, natural thickeners, Modernist cuisine starts from asking exactly what textural qualities you want from a sauce and then figuring out the right combination of starches, hydrocolloids and proteins that accomplish that task.
Modernist cuisine may seem complicated on the surface because so much of it is new and unfamiliar but I find it actually to be radically more simple than traditional cooking once you understand this ethos. Rather than a hodgepodge of tradition built up over millennia, Modernist cooking is simply about starting first from your desired goals of flavor, temperature, pressure & texture and then figuring out the appropriate use of technology to accomplish that goal.