So... what do you want to accomplish?

Let's assume that you get enough energy to create a card game from start to finish, and you are eager to sit down and start writing cards in your notebook. That's a conceivable scenario, and in fact, almost everyone I know started this way, including myself.

Now, although that's a perfectly valid way to start a game, you might want to ask yourself beforehand if that's what you really want to do. You might be wondering why. Hadn't I assumed at the start that you wanted to create a game, afterall? Well, this is because there's something that you might not have considered yet: while you wanted to design a game, the question is, which one?

Before going deeper on the question, I'll discuss some concepts about game design. Games are composed by a set of rules, wich form the game's internal structure. These rules define what the game resources are (be them be cards, points, dice, or any other with which the designer comes up), and their meaning. The logic conclusion is that in order to start designing cards, you need to abide to a certain set of rules.

Now, most people don't spontaneously decide to create a card game. What it happens is that they find a card game that they enjoy, and this is what gives them the idea of making their own. Because of that, they are inclined to adopt the set of rules of the original game, with maybe some small changes to adapt it to their personal tastes. The problem with this approach is that it limits too much what can be done with the game, so the end result will be very similar to the original source. Is that what you want to accomplish?

When working with a source reference, you can make changes to it at different levels of abstraction. The higher the level, the more impact the change has in the game, and the more different they will be in a subjetive scale of likeness. Each type of change has its advantages, and it's more adequated to a different kind of project. Here's a brief description of each one that I have improvised, ranging from most concrete to most abstract:

Type of change: Tweak.
Description: This is the most basic (low-level) change that can be done to a game. Tweaks include adjusting values to cards to maintain game balance, flavor changes, polishing the wording, and adding and removing cards to the game. Changes to the game rules are minor (if at all), and no new zones, resources, nor fields are added or removed. The nature of these changes is so basic that the result is still perceived as the same game, but they are nonetheless important. They can be the difference between a fun game and an unfun one.
Good for: New editions, new attempts at getting a game published, expansion sets, clones.
Not so good for: compilations, new games.
Skills required: Common sense, attention to detail, experience with the game.

Type of change: Variant.
Description: Tweaks played according to the game rules, variants play with the game rules. Variants tipically change the starting configuration, and can also add or remove game zones. No new systems are created, but sometimes some systems (for example, a "combat" system) work different. Still, the source game is still easily recognizable in the final product. Because they require little work and knowledge, most designers start either with a tweak or a variant of their favorite game.
Good for: New game modes, mods, spin-offs, compilations.
Not so good for: Expansion sets, showing your talent as a game designer.
Skills required: Out-of-the-box thinking. Ambition.

Type of change: Redesign.
Description: Tweaks left the game rules untouched. Variants changed some rules, but the systems remained constant. Redesign, likewise, changes the systems, and only abides to the fundamental concepts of the source. That way, it takes advantage of all the elements that made the original so good, without tying the designer to the old design.
Good for: Look-alikes. New games. Sequels. Publishing lines.
Not so good for: Spin-offs.
Skills required: Creativeness. Design skills. Abstract thinking.

Type of change: Departure.
Description: At this level of abstraction, changes can be made to the fundamental concepts of the referenced game (some examples of them are bluffing, diplomacy, or tactical battle), so that the game doesn't have to use the same. In fact, the only limits here are the fundamental aspects of card games. Working at this level, you don't even need a reference game, so the only practical reason to have one is to take away from it the concepts that don't work, or to create something completely different to it. Of course, this new game will have to stand on its own, and it's not easy working at such high levels of abstraction. A mistake here means that the game will be broken beyond repair, as it will have a fundamental flaw that cannot be corrected.
Good for: Basis for follow-ups. Avoiding comparisons. Showing your talent as a game designer.
Not so good for: Sequels.
Skills required: Abstract thinking. Vision of the global picture. Talent.

And that was all! I didn't thought it would take so much text, hope the next entries are much, much shorter.