I try to only have so many games installed at once, as part of my efforts to actually finish games and cut the list of ones I’d like to play/finish someday maybe down to a double digit number, instead of doing what I’ve done for years now, which is pick at games five or ten minutes at a time and inevitably end up going back to the same handful of familiar titles. It’s working, but the time estimates on How Long To Beat aren’t always particularly accurate. Sometimes it’s a disconnect between how I play versus how the average person plays. Sometimes what I view as “Main Story +” is closer to what the crowd thinks of as “100%.” And sometimes, as is the case of Going Under, the devs tell me that the game should be played without the handicaps and I take them at their word, which is apparently not what most people do. That, or everyone’s so much better at video games for me that the average completion time is half of mine. I could’ve called it quits on Going Under early, shuffled it off into my Regrets category on Steam, but I’m glad I didn’t. It has a good ending.
Not a mirror match, though, which is why its final boss was way harder than Transistor’s, the game I blasted through in one day. This time How Long To Beat was wrong in my favor. Seven hours? This game barely lasted five. I thought I’d play for two or three hours tonight, and by that point I was so close to the finish I decided I wanted to bring it home.
In Transistor, you play as Red, who wields the titular Transistor against an insane rogue AI network called the Process. The Process is a swarm of AI drones responsible for editing the city of Cloudbank, the setting of Transistor, in response to the whims of the citizens. Early on, when the city is normal and it’s just you the Process is after, the city’s internet terminals will have things like polls asking you what kind of sky you want tomorrow, and the winning vote gets to decide the color of the cosmos. A conspiracy called the Camerata created the Transistor to take control of the Process to try and reshape Cloudbank in their own image. Before they spring their trap, they attempt a few assassinations of key figures they expect to oppose their new regime, but in the process they lose the Transistor to Red. Apparently, whatever they’ve set in motion with the Process can’t be controlled without the Transistor, so you’d think they’d have found something else to do their murders with.
Admittedly, the Transistor is also a bigass sword that gives its wielder the ability to stop time with a function called Turn() and execute a number of functions like Get() which pulls targets closer and Breach() which can shoot through barriers and enemies to damage everything within range of the line, regardless of obstacles. You can execute these functions in realtime, but you want to use Turn() whenever it’s available because it lets you string together combos like using Void() 3 times to stack up the maximum debuff on an enemy, Crash() to stun them and make them even more vulnerable to a follow-up attack, and then the powerful Cull() attack function for maximum damage, all without the enemy being able to react. Your functions are all offline (barring a specific upgrade that can be equipped to only one of them at a time) while Turn() is recharging, but Turn() is so much more powerful than freeform function activation that you’re still better off running around dodging projectiles and hiding from enemy attacks during the recharge and then busting out some devastating Turn() combo.
The game opens with Red in an alley immediately after the Camerata attacked with the Transistor, with a stranger she doesn’t know impaled on the sword. She retrieves the Transistor, now imbued with the consciousness of the rando it impaled who then serves as the game’s narrator (these are the same guys who did Bastion and Hades, so excellently written and voice acted narration is part of the house style) and while you’re never able to use it to control the Process at all (that function of the Transistor is purely a MacGuffin), the sword itself lets you cleave through swarms of Process drones that make pretty short work of the entire rest of the city.
At the end, Royce Bracket, the final member of the Camerata, confronts you with his own Transistor, and for the first time you have to sit and watch helplessly while an enemy executes an entire Turn()’s worth of functions on you instead of the other way around. The problem with this is that if Royce were any good, you wouldn’t stand a chance. Royce uses the same functions as you, but he doesn’t use Void()/Crash()/Cull() to absolutely end you in his first Turn(). Since Transistor’s whole gameplay is about evading enemy attacks until your Turn() has recharged and then unloading a devastating attack combo, once Royce has used his Turn() (he takes his immediately as the battle begins, a reveal which is quite intimidating until you see how weak his actual combo is) he has nothing to do except run away from you while waiting to recharge. Once you start your own Turn(), Royce is as helpless as any other enemy, and you can completely wreck him with your own much more competent combo.
It’s kind of narratively satisfying, that this conspirator who tried to seize control of the Process only to completely lose control of it is just as hubristic when he dares to challenge you, someone who’s been using the Transistor for hours, with some combos that he just slapped together while looking at the menu for the first time just now. The way Red reacts at the end of the fight implies this is actually supposed to be a harrowing final confrontation, though, rather than a veteran absolutely annihilating a newbie.
Most games are built around a lopsided player powerset. Transistor gives the player the powerful Turn() for offense and nothing for defense except to run and hide and hopefully find cover. Against the Process drones, this works. You have to hide from enemy attacks between Turn()s, which you use to thin them out, leading to choices about which enemies you prioritize (do you focus your functions on damaging one hard-to-avoid enemy who you won’t even be able to completely kill in one Turn(), or do you spread your functions out to kill lots of easier enemies so that you can focus on dodging the hard one alone?) and what kind of functions to equip (do you use melee attacks like Cull() for maximum damage or artillery like Spark() that hits multiple enemies and lets you keep yoru distance or landmines like Load() that require lots of setup but do a lot of damage if they go off properly?). Process drones, on the other hand, spit out continuous, realtime attacks that the enemy’s one defense – running away – can avoid or at least delay, while having the numbers, in both hit points and unit quantity, to demand a lot of damage out of each Turn() to be effective.
Toss in a fight against another Transistor and this breaks down. There is one enemy, and while he has a decent chunk of HP (especially since he has several extra lives in a way that kinda sorta mimics the overload damage the player can take), it’s not enough to survive a properly executed Void()/Crash()/Cull() combo. Your own HP is more than sufficient to survive Royce’s combos, because if it wasn’t, there’d be nothing you could do about it. You can’t defend yourself during an enemy Turn() and none of your own functions are set up to build any kind of static defense that could slow down an enemy Transistor. The lopsided methods of attack and defense mean that a mirror match is trivial.
There’s some games that can break away from this. Fighting games spring immediately to mind, and while strategy games often have to give the enemy AI significant advantages to keep up with skilled human players, that’s just because making an AI that can defeat a high-end human player is really hard (against the very very top end of human players, even the strongest AIs remain unable to win most games). Racing games usually benefit from rubber banding to keep the race from becoming a foregone conclusion halfway through, but can also work played as a complete mirror match. Some games with mismatched attacks and defenses (and most games have mismatched attacks and defenses) still kind of work out. Hollow Knight, for example, has PvP mods that are reasonably good just by allowing players to play in the same Hallownest and hit each other, although a boss battle against an AI with the Knight’s moveset would be just plain worse than the game’s existing fights against Hornet (a warrior whose agility and resourcefulness make her thematically similar to the Knight, but whose specific moveset has almost nothing in common with his).
But the more a mirror match sounds exciting, the more it’s from a game like Transistor – one where a mirror match is unexpected. But it’s unexpected because it’s a bad idea. As a player, we sense that this mirror match wouldn’t actually be fun, that the course of the battle can only be one of two things: Either your combos are good enough to obliterate this poseur, or you walked into this sloppy and don’t stand a chance. The initial reveal was intimidating because I worried it might be the latter, and the fight itself was an anticlimax because it turned out to be the former.