I plan to jot more down on this, but recently I've been running a Cypher System campaign using one of my "D&D" settings....specifically Realms of Chirak, but powered by Cypher (with Godforsaken as a primary resource). It has proven to be interesting.
Early on, for the first couple sessions in fact, there was a heavier than usual focus on D&Dish things with lots of combat and map exploration. This was useful in that it helped the group get acquainted (or reacquainted) with the Cypher System mechanics. However it was clear to me by session 3 that Cypher is better when you use it exactly as it is intended, and not as a substitute for straight D&D.
For those of you unfamiliar with Cypher System, the simplest way to put it is this: combat in Cypher is not really any different, ultimately, than any other objective or goal within the game. Because Cypher operates as a system designed around a risk pool mechanic, pretty much every action a player takes will be centered around deciding whether or not the consequences of an action are worth an extra cost to reduce the chance of failure on an action. Part of Cypher's advancement includes finding ways to actually negate the cost/risk to the point where lower difficulty actions become automatic successes, for example. Meanwhile, the GM can throw any level of threat at the players, so long as it is understood that the high level threats will sometimes require resource expenditures just to survive, and low theats could be trivial or ultimately time wasters.
The way you make threats more interesting is to treat them as encounters that are not merely fights. You need to insure that any given conflict has something more interesting going on than just "beat up the bad guy" because in Cypher beating up the bad guy can start to feel a little repetitive; the system gives you enough mechanical oomph to make fights interesting on occasion, but the game system is really not built to make fights a primary focus. As the game itself says in it's GM advice, the focus of the game is exploration, interaction and discovery; fights should be useful components of interesting obstacles or conflicts, but not a goal in themselves.
So, to put it another way: when I decided to adopt my long-running campaign (which has appeared as a setting in D&D since 2nd edition, Runequest since the Avalon Hill edition and GURPS (both 3E and 4E) I needed to shift gears on how to plot the conflicts and focus of the game. The last couple sessions have resulted in most combat-free political and social interactions, emphasis on the character arcs chosen by the players, and a metric ton of intrigue and clue finding. An occasional fight happens, but the goal of any such fight needs to be more than "this monster is in your way." So the last fight of the most recent session involved protecting a woman from the vengeful ghost of her dead sister, and establishing some mechanism by which the ghost could be persuaded to not want her death. The results were entertaining and far more significant than just re-murdering the ghost.
I hear criticism about the risk pool mechanic and combat of Cypher System at times, and realized that with my adoption of a D&D setting (in which I initially started plotting and thinking about it as if it were going to run in D&D) was actually tripping me up, in so far as that the game itself just does not really want to be played that way. When it came to players, realizing their focus was on playing with a risk pool that they needed to evaluate and as GM realizing that combat should rarely be the point of any fight helped a great deal toward restructuring my thoughts on this.
Ironically, this chain of thought never cropped up when I was running the original settings I devised for Cypher, so the organic process of how things worked was never an issue. It was only once I tried moving a campaign world over to Cypher which had previously operated on "D&D logic" that I noticed the contrast.
I am now thinking of some new adventure and campaign ideas to explore in Cypher System, particularly using the Stars are Fire resource, as I think a far future high-concept exploration and discovery campaign in space might be more interesting to me than a more conventional Traveller type campaign. Armed with that thought in mind, the idea is: what can Cypher System do, if untethered from any more conventional (e.g. Traveller proceduralism) approaches to SF? I'll post more as I explore this angle.