This game isn't online yet, but a video of a 100% playthrough can be found at the bottom of the page.
Success to Continue
In this game, play continually exhausts a limited resource, but successful play generates more of that resource. Every Extend is the first game that drew my attention taking this approach so explicitly. In Every Extend, a player's only attack is killing themselves by exploding, but their explosions cause chain reactions. Long explosion chains earn extra lives. But this idea is at the heart of several old arcade games, where big scores, earned by risky play, give extra lives. See Pac-Man or especially Robotron.
Long-Term Goals and Shops
Games keep players hooked best, in the long term, when they provide clear short, medium, and long term goals without overwhelming players. Many important game design tools work towards this end. Leveling in RPGs, like in Fire Emblem, promises long-term progress and gameplay evolution. Well-crafted narratives, like in Suikoden 2, work at short, medium, and long time scales; wanting to know what happens next (in the small and in the large) is an implicit goal. Empty slots in a menu interface, like the screen for souls in Castlevania: Aria of Sorrow, or the pokemon list in Pokemon, can be a very powerful implicit long-term goal, urging completion. Side passages in levels that can be seen but not reached from early play, like in Super Metroid or Legacy of the Wizard, work organically as mid or long term goal, especially when combined with changing player mobility options. Delayed resolution of tasks, as happens in Animal Crossing, provides a different way of hooking players, where seeing the results of completing a goal becomes its own time offset goal. Simply watching your player traverse a zoomed out, iconic, linear map, as in Ghost and Goblins or Castlevania, and getting a frame of reference for progress, can be a subtle long term goal. More immediate as a goal is actively filling in an incomplete map, as in Castlevania: Symphony of the Night or Super Mario World, with seeing the full map being the ultimately long term goal. Getting access to levels that are visible, but not accessible, from a menu interface, as in Tony Hawk Pro Skater or Angry Birds, or Guitar Hero, serves as an obvious long term goal structure. Another organic long term goal structure puts players in open world spaces where some regions are much, much more dangerous than others, and then requires players to learn how to get powerful enough to survive those dangerous regions, as in the overworld in The Bard's Tale, the overworld in Dragon Warrior, and the general game flow of Everquest./* achievements. player abilities. Sims/SimCity/Civ*/
An approach widely adopted by flash games for long term goals is having a single, global shop. Structurally, players alternate between a gameplay section, where they earn money, and the store, where they upgrade their capabilities. Crucially, from the very first moment of play, the store offers goodies and upgrades players can't afford or haven't unlocked. As play continues, players earn money and slowly acquire all the goods in the shop. This, in turn, let's the achieve more in the gameplay section. In addition to providing short, mid, and long term goals, this structure also balances game difficulty for varying player skill levels. Hedgehog Launch 2, Learn to Fly 2, and Burrito Bison are examples of this approach in the Flash gaming scene. I also use this approach in phoenix ashes, along with Racing Comrade and The Fairview Horror.
A problem any linear skill-based game faces is the situation of providing a challenge that is too difficult for a player. If a player encounters a challenge that is too easy, they'll breeze by it and forget it. That's not an optimal experience, but it's forgivable, or even forgettable. But if a player encounters a challenge that halts their progress, and they fail a few times, they'll often quit playing.
A recent design style, most easily seen in Angry Birds, is to make any challenge that prevents progress quite easy, but then add tiers of optional achievements to provide challenges for players who can handle it. So, a novice player might finish many Angry Birds level with only one star out of three, but a good player might consistently get three stars on every level.
Angry Birds gets a second benefit from this design style. If a player does get stuck on a level, rather than simply quitting, they have the option of returning to levels they've already beaten with one star, to see if they can get two or three stars. This at once keeps them playing, gives them a break from the challenge that is frustrating them, and gives them practice that might help them play better when they return to the level that had stopped their progress.
It's quite a tidy design, and it allows Angry Bird to provide hard challenges without loosing players in the process. The only real downside of this technique, which is a great technique, is that it doesn't generally make sense for games that are oriented around a narrative. Being able to instantly replay previous challenges rarely fits in the context of a story.
Outside the mobile / flash context, variations of this approach can also be seen in games like Kirby's Canvas Curse, New Super Mario Brothers, Yoshi's Island, and Super Mario 64, each of which provide easier normal paths and then a large amount of harder optional goals for replay and extra challenge.
In my Long-Term Goals section, I mention that goals be fractal, with short, mid, and long time range goals being juggled. In my mini-essay on level design here, I claim the arrangement of paths and space in levels should be similarly structured. Thinking of change at different frequencies, for all sorts of game assets and structures, is an essential skill much more broadly and often the easiest way to tell good games from bad ones.
Phoenix Ashes takes this approach to the spacing of bosses and minibosses as well. 4-6 normal levels pass before a boss or miniboss encounter. Usually, 2 miniboss encounters precede a big boss encounter. In the narrative, the characters are focused on the approaching boss encounters (of which there are 4), while noting in passing the miniboss encounters.
Additionally, some powerups are more interesting than others, particular new player ships. Those powerups are unlocked after miniboss or boss encounters. The less interesting powerups are unlocked on regular levels.
As one extra beat of change, new enemies are revealed every several levels. Weak enemies are introduced quickly in the first 25 levels, upgraded enemies are introduced more slowly over the next 25 levels, and nasty combinations of upgraded enemeis are introduced over the last 25 levels. This is high frequency, constant change over which the slower introductions of miniboss and boss play out.
The goal here is this: players should always be noticing noteworthy change and variation. But they should also be anticipating larger scale change and challenge, which should happen less frequently or rarely depending on its scale and telegraphed accordingly before it happens.