Genre: Third-Person Shooter
Publisher: Atari
Developer: cavia, inc
Release Date: February 27, 2007

Ever rewrite a review twice? That's what I had to do for Bullet Witch, a game that I initially hated like the plague and then grew to love immensely, but I still found myself unable to overlook the undeniable, massive — though not quite insurmountable — flaws and its innately limiting aesthetic. Bullet Witch is a game that mixes fan service, awesome, and the infamous question, "If this game had more time, how much higher would review scores be?" I am also going to go right up to the top and say it here: This game is not worth full price. Rent it, or buy it used for under $30, but pay $50? Absolutely not.

Cavia hit on a potentially awesome — if fairly specifically niche — concept, figured out enough of the basics to make it fun, and then promptly rushed the game out the door, thereby creating a truly infuriating result. It's so infuriating that part of the conditions for the U.S. release by Atari was a second round of bug fixes. The problem is that even the second round was apparently too rushed and only fixed critical, gameplay-breaking bugs while leaving some atrocious graphical and other bugs in place. As if this weren't bad enough, the U.S. release of Bullet Witch has the sorry timing to be perfectly comparable to Gears of War, Crackdown, and the beta of Halo 3 on the same console. In a word: ouch. The question then becomes, does this game stand on its own legs at all? The most accurate answer I have been able to devise is that it stands on one leg and uses the game's trademark "gunrod" as a crutch.

To best demonstrate this tendency, take a screenshot of the game, and then put it in motion. As a still shot, it looks to be decently in the middle in the 360's range, with some neat character design and multiple looks for each non-boss enemy. In motion, however, the main character's dress will promptly go through her leg in a clipping error, and objects will fall through other objects with great frequency. The oddity seems to be that many objects react to physics, while others don't. For reference, Half-Life 2 did this too — the Gravity Gun didn't work on many fences. The difference is that Valve remembered to code the physics objects to have some reactions, even if slightly unrealistic, for non-physics objects, while Cavia didn't. Needless to say, this looks ridiculous, but somehow, it never quite ruined abstracted gameplay.

The sound effects are a hats-off moment in terms of actual explosions and gunshots. The vocals are a highly mixed bag; the bad guys sound suitably twisted and the generic humans sound okay, but the recurring NPCs sound atrocious. The music is unmemorable, with the exception of when you cast a large spell and an idiosyncratically good orchestral rhythm bumps up the tension.

Actual gameplay elements are similarly mixed, although the basic controls are very good once you get used to them. In particular, I found accusations of the magic system being unintuitive to be rather unfounded, though if the game had any more spells, it would definitely break down fast.

The catch is the L trigger. Not only is it not documented in the game manual as being a way to jump (thus the mistaken statement in our preview that there is in fact no way to jump), but that jump is also not very controllable. I cannot count how many times I jumped off cliffs to my death while trying to avoid a few bullets, but once you've made the adjustment, the feature becomes usable.

Level designs start off as generic, one-tactic straight-aways, but things get sort of interesting when actual tactical options start showing up in the second level — should you use the car for cover, or throw it to wreck their cover? When you reach the latter half of the game, level design jumps from being good to excellent, with a few occasional annoying spots.

Difficulty level is also hit-and-miss. Figuring out decent strategies will bring areas that are head-bangingly difficult and render them too easy. To be frank, the A.I. can be pretty brilliant and dodges quite a lot when shot at, which requires you to have actual aiming skill, since there is no lock-on; this could be either a welcome change or a sore spot for gamers. On the other hand, the A.I. can seem equally incapable, such as when it forgets that your melee attack one-shots the generics, runs at a wall trying to get to you, or shoots at the wall when it thinks you're in range. Exploiting these weaknesses isn't possible too often, but it can occur just often enough to really highlight Cavia's shortcomings on the A.I. front. Strangely, though, the A.I. errors, above-mentioned clipping quirks, and control oddities combined to be surprisingly fun, especially if you want to think just a little about shooting things in a campy style.

Bullet Witch also suffers from lacking enemy types. In the first two levels, you will encounter four enemy types, one of which is a boss. Admittedly, the first type, known as the "Geist," is a bunch of insane demonic soldiers who vary a fair amount in appearance and significantly in armament. The third level introduces three more types that are promptly forgotten for the remainder of the game. The fourth and fifth levels add none, and the final stage only adds a few more.

Those six levels are the reason this game isn't worth full price; they're perhaps each half an hour long on a moderately paced runthrough if you don't retry too often. If you like the game, you'll get several replays out of it, but a title that lasts maybe five hours on normal difficulty level if you are good at choosing tactics is intolerably short. (The exception to this statement are the four-dollar Burger King games, but even those titles last longer!) This is made worse with the existence of few secrets, no multiplayer mode — this is a commendable decision in that it allowed focus on the single-player campaign, until you throw in the Gears of War comparison — and downloadable content that didn't work. It seems like there are in-game prerequisites to getting the unlockable content, but at that point, I have to ask, Why is it downloadable?!

Bullet Witch is the most love-hate game I have played in a long time. I love the fact that there are just enough twists in concept and play to make it a creative twist on the third-person shooter genre, and there is a certain level of joy that shines through regardless of the glitches. I absolutely hate that this is the most obviously rushed title I have played in a similarly long time, where too much focus was put in areas that should have received attention after other tasks — such as debugging — had received a lot more attention.

Hats off to Cavia for twisting the concept of a third-person shooter and making a game that is enjoyable in spite of its flaws. The hats come back on for all of the errors coming so close to ruining the core play, a pathetically short play length, and Atari's potentially worst-possible timing for the title. Had this game been released several months earlier, its score would probably be a fair bit higher, but in the wake of Gears of War and Crackdown, it can only be recommended for rent or for really cheap used purchase. Even then, it's only recommended to people who like the concept, in spite me wanting to give it more love in hopes that either Cavia will try again and take the time to make the new result better, or we'll get more variants like this in the third-person shooter genre.

Score: 5.8/10