From the 10,000 word post:
"The narrative ... will grow of its own accord, as events follow events in the player's imagination. They will, as they understand it, move from the town to the road to the wilderness to the dungeon to the treasure vault and back again in a smooth, imaginable manner, as each element is described as fully and accurately as the DM is able."
Let us try to get into the nitty gritty of this. I have been describing the process of dungeon mastering in broad strokes. I would like to spend a post speaking of the particular process of applying the setting to the player's action.
No matter what the place, you have certain stock events which you can present to the party: battle, to give them experience; friendlies, to supply information; civilization, to enable them to resupply; clues, to stoke curiousity and get them motivated; scarcity, to encourage resupply and greater party efforts towards their own stability; opportunity, to stand off scarcity; fame, as it strokes the party's ego; and wealth, as a means to exploit opportunity and fame, stave off scarcity, investigate clues, gain power over civilization, increase the number of friendlies and better fight in battles.
Scarcity and battle diminish party security; wealth and fame increases it. Civilization exists as a resource; opportunity is the means to exploit that resource or fail trying. Opportunity increases wealth and fame, and sometimes contributes to scarcity. Fame increases friendlies, who in turn supply opportunities. Clues suggest the manner in which scarcity may be avoided, or opportunities be found.
All parties want to be secure. The game is based on directly threatening security (battle); suggesting means by which it may be obtained (friendlies); locating it (civilization); increasing its uncertainty (clues); denying it (scarcity); promising it (opportunity); rewarding it (fame) or enabling it (wealth). So long as you're doing one of these things in regards to the player's security, yours will be a good game.
Obviously, the greatest threat to security is dying; lesser threats include losing levels, losing limbs, disease, extreme aging, loss of precious items or irreplaceble combat gear, associated friendlies, the good will of the local community (along with exile) or deity (along with excommunication), damage to reputation or appearance and finally poverty. Not all these things automatically lead to death; some might be imagined to be worse than death; and of course the list is not finite. But as a dungeon master these are all things you can play with while establishing your campaign. Threatening a party with any one of them will create tension, an urge to rally the party's action towards combating the insecurity ... and either the thrill of victory (we are more secure now) or the devastation of failure (whatever we did not want to lose, we've lost).
Every player must be encouraged to view their security as something for which they have the responsibility. If the player has time to make choices; gather supplies and make plans to carry forth those choices, with all the risks they imply; and the character ultimately fails and then dies, the player will not find fault with the DM or the campaign. In point of fact, IF the player has had the opportunity to carefully make measured and meaningful plans, even if the character dies the player will pronounce the campaign to be a good one.
If, however, a player's security is treated casually by the Dungeon Master, who tosses the character into the fire cheaply and without hope for survival, then there is no tension. Conversely, if a character is so well supplied that no insecurity can possibly exist, again there is no tension. At best such campaigns may encourage a certain breed of player who either does not care to invest any emotional regard for their characters, or players who must have everything their way to satisfy their enormous egos, or else pout and stamp on home. Such campaigns are well known to long-time players, who avoid them. Such campaigns feed a base sort of individual who deserves more pity than disdain.
Thus, if you will build tension, your setting must include the possibility of the player dying without that being a guarantee; and the possibility that the player may lose any other object or characteristic if the player does not win the battle, gain the trust of a friendly, solve the clue, return to civilization, refill the scarcity and make the best of their opportunities. Should they do those things, fame and wealth should be forthcoming.
Each battle and friendly and clue and so on is therefore a brick in your campaign's wall. Each must be meshed into the overall scheme like the instruments of an orchestra, playing its part. The battle leads to a clue, which draws the player from civilization, which asserts an opportunity, which cannot be exploited without returning to civilization or making due somehow, the success of which leads to another battle which reduces something else the party needs (hit points or weapons or their own confidence, always in short-supply), again driving them back to civilization, except that they meet a friendly that resupplies them and makes them ready for another battle, which threatens the friendly, whose disappearance leaves behind another clue that promises wealth greater than the party's lack of confidence, and so on and so forth in a great circle that does not end, ever.
If, as a DM, you will think in terms of how each opportunity, clue or other event leads to the next; or how the failure to exploit each opportunity, clue or other event forces a retreat in order to resupply and rest; then you will, as the quote which started this post asserts, build your campaign from town to road to wilderness to dungeon to treasure vault, exactly as seamlessly as I promised. It only takes practice.