The question "should gamers finish games they're reviewing" pops up from time to time, kind of like the "games as art" fiasco. The answer is simple: It depends what's meant by "finish."
As in a multiple-choice test? Coding the encryption subroutines for a virtual-private networking application? Calculating the average air-speed velocity of an unladen swallow (African or European)?
Or indeterminately, as in "trimming the hedges," "scrubbing the latrine," or "examining the body in the morgue"?
You might argue finishing a game means finishing its story, but then you're just referring to a particular sort of game with a "story mode." Remember, plenty of games don't formalize the latter. "Finish it" flag-bearers don't really mean stuff like Bejeweled, for instance, or Tetris. I've reviewed versions of the latter over the years, but never "finished" them in the sense that I passed some final, curtain-dropping level. (Does Tetris even have a finite number of levels? You tell me.) So there's the usual categorical problem: When you say "game," you're referring to a range of activities. Finishing Lucasarts' Grim Fandango--an adventure game with negligible branching--is world's apart from saying you've "completed" Lionhead's highly adaptive, morally mutable world in Fable 2 (PCW Score: 100%).
Factor in online play, competitive or cooperative, and miscellany like achievements and collectibles with unlockable bonus items and your responsibilities get even more obscure. No critic plays "perfect" games, so what amounts to "enough"? Bring up the start menu in most games and you're looking at a range of options--call them "features," like the extras on DVDs. Do they demand thorough investigation?
You can watch the Criterion transfer of Citizen Kane and say you've "seen" the film without once invoking film critic Roger Ebert's illuminative commentary. I've read helpful reviews of DVDs by respected critics who leave aside some of those extras entirely. Deciding what readers need to know depends on your priorities, something that's ineluctably subjective. In the case of the two-disc special edition of Citizen Kane's case, Ebert's commentary probably warrants critical attention, but the "rare footage from Hearst's San Simeon estate and Welles Historic War of the Worlds broadcast"? Maybe not.
Audience matters, of course, and used to be a helpful dictate in print where you could survey and react accordingly. Gauging that audience at a time when your story's as liable to wind up linked in a rabidly enthusiast forum as a politely casual one means your feedback's going to be mercurial. There's no right or wrong way to react to the Internet (okay, the wrong way would be trying to please everyone) so it's probably safest to stick with seasoned advice: "Write what you know."
That said, how much time's enough testing Age of Empire III's multiplayer mode on Ensemble Studios Online? Do you have to play as every civilization (of 18 total) in Civilization IV before putting pen to paper? Should you have fully-pimped level 80 versions of all the possible race and class combinations in Blizzard's World of Warcraft before scoring it?
Common sense applies. If a game has a story mode, it's best to play until the credits roll. If it depends on branch-play, as in Star Wars: Knights of the Old Republic, the Fable series, or Sucker Punch's inFAMOUS (PCW Score: 90%), play probingly. You don't have to work these games over twice or thrice. Readers don't want your academic blow by blow. They just want to know if the system works, and why (or why not). In a massively multiplayer game like World of Warcraft, it's enough to pick a path and document your journey. Attempting to probe it all is madness, and readers who expect as much from a single reviewer are asking for the sea in a spoon. Readers have responsibilities, too. One of those is recognizing when a game's too complex to cover unilaterally.
Rules of thumb:
If the game is limited in scope, like Mirror's Edge or Dead Space, you play to cessation. These games have hard limits, and it's simple enough to reach around and find their edges.
If the game leans toward freeform play, like Grand Theft Auto IV or Fallout 3, you approach it like halo-dropping into a war zone and report back accordingly. You can't cover all the bases, so don't. Don't bother ticking off the bullet points either. Readers can find these on their own.
If you didn't finish the game, say you didn't. Nothing necessarily wrong with that. Some critics are sprinters, others marathon runners. What's important is full disclosure for the benefit of your readers, who can adjust their take-away accordingly.
And finally, insight trumps in-depth reporting. I've learned more about some games reading savvy 500-word reviews than "comprehensive" 5,000-word prose vomitoriums. The latter appeals to those who've mistaken size for potency, a deficiency that shouldn't be accommodated. And just because you've finished a game (or you're the best player, competitively speaking) doesn't mean you understand why in conceptual terms. Playing games and reflecting on them are different endeavors entirely.
Follow me on Twitter @game_on.