Never has a first-party game been more honest and self-aware about its circumstances than Segagaga. A coda to the Dreamcast's troubled life released a few short months after Sega announced its withdrawal as a platform maker, Segagaga morbidly revels in the dire situation of the eponymous(-ish) company and its workers with the sort of bleak gallows humor you'd expect to find in, say, World War II movies about "the front." Only, instead of soldiers inside grimy trenches quipping about bad rations and dwindling ammo supplies, it's unkempt, heavily overworked developers all but shackled to their cubicles inside of dingy vaults, trapped from seeing the light of day as protagonist Taro wards off roaming bands of influenza.

This is no allegory. Segagaga has nary a romanticized bone in its gaunt body to be found. "Everything you've heard about Japanese game development and more is true, and nowhere is it more true than here at Sega at the turn of the millennium," the game morosely reiterates with every mission doled out, every NPC conversation, and every combat encounter. A certain undercurrent of cynicism permeates throughout that can only come from those who have truly lived the life they're depicting and witnessed how the digital sausage is made, exaggerated or not. Even so, cynicism by itself is rarely constructive and Segagaga isn't a game that takes the easy way out by glibly giving into it. What it actually is, in no uncertain terms, is a game about loving somebody at their worst, somebody who hits the bottom again and again, only to break through the floor and fall even deeper with each passing day.

A quarter of a century later, it's a game that I myself deeply relate to in some ways, having entered my 12th year in Japanese-to-English game localization this past April. The only game industry I know is the Japanese one and never is it for the faint of heart. I have taken on weighty workloads with inadvisable deadlines made viable only with the energy and single-minded dedication of youth. I have had the long nights on the job, staying up until the break of dawn, workshopping and refining the most minute details in pursuit of a translation that sings. Things where the raw dollar-and-cents return I was poised to earn for such minuscule amounts of material couldn't possibly justify the investment I put into them. I have had—and sometimes lost—the battles with my superiors, my agencies, and my clients to do right by their games and their audience. I have had projects cancelled on me after months of working onto them, their divulgence forever verboten by NDAs. I have lost contracts and promises of work and money even after winning it all fair and square because I just happened to rub the wrong person in power the wrong way a little too much for a little too long. And I have stuck it out in this industry doing what I do despite it All, despite the politics and the pay and the lack of job security, because I genuinely believed in what I was doing like nothing else I have done in my life. For the love of the game that is Japanese video games.

Segagaga languished for decades as one of those stubborn holdouts whose English translation just never seemed like it could come together, even as white whales that eluded fan translators for decades were tamed in the end. I would know better than most: I was once asked early in my career if I would join the ranks of one such project that was always on-again, off-again. From what I gather, even Sega itself explored ways to bring out an official localization, belated as it would've been. Yet it remained the exceptionally rare kind of fan translation target where no amount of desire, reverse engineering insights, and translator prowess ever proved to be enough to coalesce into a final, playable patch. Even Tokimeki Memorial, a similarly troublesome game that many romhackers had written off despite its significance, fell a few years ago at last and you can now play it in a language you probably understand... kind of. (Look, there's still work to be done and it's a matter of public record that I have some strong opinions on the version those of you reading this can actually play as of this writing.)

This state of affairs ostensibly changed a few months ago when an English patch at last emerged almost 25 years to the day of its original launch. That is, with a lede that was buried by both the team behind it and especially by many enthusiast sites promoting its release: the translation wasn't produced from scratch by humans from start to finish. Rather, the script was, in the project's own words on its GitHub page, "developed using a combination of DeepL and ChatGPT 4o/4.5. That translation then went through a substantial, months-long human translator review." In professional circles, we call such a process MTPE, or "machine translation post-editing." During it, the first pass of a translation is done by a machine translator like, indeed, DeepL, ChatGPT, or Google. It's then a human translator's job to review that output by comparing the generated translation against the source text, make any corrections where necessary, and deliver a polished final product, or at minimum something more presentable than the machine could've managed by itself.

At least, that's the elevator pitch that translation agencies routinely sell to spendthrift clients. I've already litigated the practical realities of MTPE on the ground on this very blog at length, so I won't repeat myself here. (Though, at the risk of adding another 2500 words of homework, I encourage you to go back and read it first if you aren't familiar with MTPE and its myriad shortcomings.) As anyone familiar with that post and my work in general can likely surmise, suffice it to say, the abrupt conclusion of Segagaga's protracted localization journey in such a manner was deeply demoralizing. While preliminary translation patches made with more rudimentary machine translators have, for better or worse, existed for a long time, the precipitous increase in LLM usage combined with MTPE as a supposed guardrail tells me that these teams are either insufficiently considering the practical and ethical implications of their methodology, or disregarding them outright. For my money, the fact this is happening more frequently with high profile games such as Segagaga suggests to me that it may well often be a case of the latter.

As much as I genuinely urge such teams to rescind their work and take down any patches, even after the fact, I also recognize that the cat is unfortunately out of the bag and many more cats will probably be let out of just as many bags so long as access to LLM models remains affordable, something which itself may not remain true for long. Be that as it may, speaking as a former fan translator myself, LLMs and MTPE present further problems specific to fan translation that merit discussion in addition to the broader issues I've already explored at length in my previous post linked above. At the root of it all is an important question that I want to ask. Namely, what does translation do to the translated work and for it?

I don't mean what a translation, when done properly, accomplishes on a mechanical level. You know what my basic job description is. Nor am I here to debate the supposed merits of translation versus localization writ large. (It's a false dichotomy, even outside creative mediums.) What I'm really asking is, what happens to a work once it's been translated, especially for the first time? In entertainment media and video games in particular, where translations are often the most time and labor-intensive to realize, most works make the jump to another given language only once. Barring exceptional situations, in the vast majority of cases, the first translation into that language is the only one that work ever receives, officially sanctioned or otherwise. In turn, it becomes the de facto way that audience can and will ever engage with that piece, the very grammar and semantics available to that language informing how people observe even objective realities of that content.

In other words, when you're translating somebody else's creation and especially when that translation is taking place in your chosen language for the very first time, you're not only speaking on another creator's behalf. As I've said many times, translation is an act that inherently requires that you put words into people's mouths. Not only the fictional mouths of the characters you're borrowing, but also those of the real people behind them. To a large extent, you're defining nothing less than that work's very legacy and that of its creators in abstentia. That new audience's first and quite likely only impression of it will be profoundly dictated by the choices you make throughout that process, the ensuing conversations being ones in which those creators likely cannot participate.

That is a heavy, heavy responsibility. I say this not to discourage other translators from making bold, creative choices in their approach. On the contrary, the job often demands it and no amount of attempted deferment to the supposed objectivity of translation dictionaries (or, indeed, LLMs and machine translators) will lighten the weight of that burden. Rather, I say this to stress what is always at stake when you decide to sit down and attempt to represent the work and mindset of somebody else with your own words. It may well be the only shot that media ever gets at reaching people within your sphere, making it imperative to aim both well and true.

More than other fields of translation, games and other entertainment exhibit what I've come to term as "self-evident context." That is, as bespoke creations, the words and ideas contained within them aren't strictly confined to our own reality. They can and often do have their own underlying histories and etymology informing their usage within that world. Any respectable translation concerned with accuracy must take those differences into consideration, as well as how audiences in the source and target languages alike each perceive those word choices. Something in mecha fiction might in its native language, for instance, call its take on giant robots "dolls" because of their lack of autonomy. It's then up to the translator to determine if a direct translation of the word "doll" adequately maintains the intended metaphor, or if they need to adopt a superficially different term that can nevertheless evoke similar ideas.

This is what makes the use of machine translators of any sort insidiously dangerous to deploy on creative projects. The algorithms underpinning them by their very nature force them to bias potential translations based on likelihood, which they in turn ultimately determine by their average frequency in A:B comparisons, as well as potential proximity to other, related words. While they may sometimes happen to be incidentally fed fictional material as a matter of course, they have no means of discerning it as uniquely fiction in contrast with the rest of its archived material derived from the real world. All of it is simply lumped together as "language" in a murky stew comprised of infinite ingredients in varying proportions. This makes the systems mathematically incapable of weighing potential translation candidates rooted in fiction more heavily than anything else they can unearth when tasked with translating fiction. Doing so would essentially require them to be fed more fictional sources of material than real ones such that the probabilities and averages arrived at favor the former, which is inconceivable.

When considering that diligent translations also must bear in mind the holistic, big picture needs of the source's themes and narrative, not simply moment-to-moment, or even line-by-line semantic parity, machine translators and LLMs quickly turn from a time-saving shortcut to an outright liability. After all, if ChatGPT and other AI models can't be expected to genuinely remember past interactions with them to any reliable, meaningful degree, why should they be trusted to have the reading comprehension skills necessary to consume a script in its entirety and then make judgment calls about terminology, character speaking styles, and many other pressing variables? They simply can't, and if you can't trust them to maintain a working memory of what they're translating as they "translate" it, then it turns out that the better, quicker, and safer option is always, in fact, to turn to a knowledgeable, capable human who can from the outset.

But the concerns about LLMs and MTPEs in fan translation extend well past the immediate practicalities and responsibilities of the work itself. Any discussion of them must also grapple with the reality of how those initial translations are generated. When you as a fan translator take that source material and paste it into something like ChatGPT, you're not merely receiving a one-time output before the technology purges itself of your query. You're feeding the work of creators who toiled months, if not years, into a system that will ingest it, permanently retain it, and disperse it to other users globally without those creators' permission for as long as that system exists. That material, which was never yours and will never belong to you, is, by extension, not yours to effectively donate to the (aspiring) profit engines of Sam Altman and his ilk. While permission in general might be relative when it comes to fan translation, what is absolutely indisputable is who owns the work being built upon and it certainly isn't the fan translation teams engaging in this behavior.

Having almost certainly seen my work fed into generative AI by at least one renowned client by my count, I can say that I don't spend those long nights playing Japanese builds of games (sometimes at breakneck pace for prolonged periods), writing dialogue, and troubleshooting problems with my colleagues just to have those same clients turn around and ask a computer to produce a limp facsimile of my writing in a brazen investment grab. Is that client legally allowed to do so because they own the copyright to my translations? Yes, but it doesn't change the fact that I put in the time, effort, and quality that I do for the games themselves, for their developers, and for their players, not for a technology whose stated end goal is to make me redundant in the eyes of employers. And while I can't speak for the teams behind the games now receiving MTPE-based patches, I can't imagine they crunched and made sacrifices in their lives during development just to have their work served up on a platter to an objectifying, averaging, muddying algorithm, either, least of all those whose games have waited decades to make their international break. None of those individuals slept under their desks or were hospitalized so that a small group of people on the Internet many thousands of miles away could enjoy a hit of clout at their expense for but a fleeting moment.

For many game developers and other creatives in Japan, the extent of their impact on the world such as they see and feel it doesn't extend beyond the archipelago. Even acclaimed, globally celebrated individuals can go their entire careers and enter retirement feeling wholly anonymous, never even imagining an audience spanning oceans and continents. It may not be until a foreign fan reaches them, and especially in their own language, that they understand their work did travel abroad and make a difference well beyond their homeland. In my own experience, "white foreigner with a penchant for retro dating sims that time forgot" can make for a surprisingly quick and potent ice-breaker when introducing yourself in the right circles, it turns out.

As a translator who will never be contractually allowed to reveal the bulk of my work with the world, it's this kind of life spent in obscurity, one that has a way of minimizing your self-worth from a lack of recognition, that causes me to feel a deep, knowing sadness when I see yet another fan effort needlessly succumb to the siren calls of MTPE-tinted machine translations. Beyond the practicality of the whole exercise, beyond the ethics and morals of churning through material in that particular way, is a group of people whose efforts made and prices paid we'll never know.

Were they taken care of?
Were they taken advantage of?
Were they bled dry?
Did they watch life pass them by?
Were they happy to do it all anyway?
Or was it all they could do to get their pay?

We'll never know. All I do know is, when what remains of that is taken out of human hands, no matter for how long or how little, they and their legacies are condemned to their own vault, their chance of escape vanishingly little. Only a person can make a translation that respects and reveres the humanity imparted upon this earth. Either you do this for the love of the games themselves and the people behind them, or for the love of yourself. There is no line to be drawn down the middle.

Most games, but especially modern games, are made to be finished. I am expected to play from beginning to middle to end, and if I find myself struggling at some point along the way it’s generally safe to assume a myriad of helpful progress-smoothing settings are waiting to help me to readjust the experience to my liking. This is a good thing. I should be able to get as far in a game as I want to.

Want to. Am I carrying on because I want to, or because I feel like I should—because I can? Do I actually want to see what’s next, or am I just trying to get the whole thing over and done with so I can say I cleared it? Am I clawing my way through this stage because I like a challenge, or do I think I’m just too close to the end to quit?

Am I having fun?

That’s the only question that matters, really. This is entertainment. I’m supposed to be enjoying myself. And if I’m not? I’m allowed to stop. Yep, straight away. Drop it faster than a radioactive potato full of hornets. I’m not obliged to give a game a fair shake if I’m playing for my own sake, or to listen to anyone who says “No honestly, it really picks up once you reach the plot twist in chapter 10”. 

Although there are plenty of “good” reasons for dropping a game—poor pacing, irritating checkpoints, some mandatory boss battle built out of broken hitboxes and spite—I’m also allowed to put a game down for no reason beyond simply having had my fill of it. I’ve spent some time with it, and that time was fun. That is enough. In much the same way that one slice of cake is a pleasant treat and two is an invitation to lie on the sofa with a belly full of regret, moving on when I’m at my limit can be a positive, mindful act. Maybe I’ll go back later—easier than ever thanks to generous storage on modern devices and the general adoption of cloud saves—when I’m in the mood for more of this metaphorical cake, maybe I’ll leave it unfinished forever, my head filled with nothing but fond memories.

But the ending. 

What about it? I grew up playing European computer games that used to pride themselves on taking months to clear. Months. At best. On purpose. This extended and often insurmountable slog was seen as a good, desirable trait for a game to have, length and difficulty prized above all else. Some games didn’t have an ending at all, or contained literally hundreds of stages, which pretty much amounts to the same thing. Some were of a far more reasonable length and then ended with a simple text message or an angry rant aimed at the very people who had bought and cleared the game. Was this disinterested and/or belligerent attitude towards players superior to the less demanding design philosophies that came after? Oh goodness, no. But I am grateful for the lesson they unintentionally taught me: Playing a game purely for the sake of playing it has to be enough. Finishing it is some separate and unrelated thing, a nice but unnecessary—and unexpected—bonus. 

Even without that “training” under my belt, why would I make the effort to reach an ending I don’t really want to see? What’s to be gained from spending even more time with something I already stopped enjoying, just to see a conclusion I’m not that interested in? Who is benefiting from this unwanted toil? “I stopped enjoying this a week ago but I soldiered on anyway” isn’t something I can brag about. Gritting my teeth through unpleasant gaming experiences doesn’t confer any benefits in the way pushing through exhaustion in a long run would. Nobody else cares whether or not I roll the credits on some lengthy RPG, tough action game, or popular metroidvania I didn’t really like. 

Having to see the ending for the sake of playing “in the right way” or chasing some muddled notion of getting “value for money” turns games into a chore at best, a task to be cleared before I can move on to something I’m looking forward to, and at worst into content to be consumed. In these sorry examples it doesn’t matter how I feel, so long as I’ve obediently devoured exactly as much product as my personal guilt or the hobby at large deems to be the correct amount.

I don’t want to view games that way. I don’t want to look back and wonder how many hours I wasted conforming to nobody in particular’s idea of experiencing games “properly” when I could have been doing something I enjoyed instead. I don’t want to have a backlog of games—a series of orders to work through and clear—I want to own a library of them, something to dip into and enjoy it as I see fit. Some days that’ll take the form of repeat clears of an old favourite. Others will be the beginning of an enthralling epic I’ll willingly lose a month to. And every now and then I’ll play one for half an hour (or less), thank the game for a fun experience, and move on.

I’m going to take this train of thought even further: it’s OK to not start games, too. I’m under no obligation to play a game that doesn’t look appealing, even when I’d like to give its spinoff or sequel a go. Entire multi-game arcs can be thrown in the bin, because I can pick a series up at any point it catches my eye. I am not an idiot; I’m sure I can catch up on the most salient plot points or previously established mechanics as I go, and use contextual clues and plain old reading to infer the rest. I can practice combos, memorise spell names, and wade through skill trees if I need to. The story will not collapse if a character alludes to something that happened 20hrs ago in a completely different game, and if it does then that was never my narrative fumble to fix.

I must remember to play with purpose, and with my own best interests in mind. If I stop trying to finish (or start) games because I “should”, I’ll only end up playing—and enjoying—even more of them.

(My Switch says it's been 10+ hours, but I feel like it's been 40+. This could just be the time dilation of practice.)

Like I've said, I’ve been really stuck on this year’s release of Ridge Racer in the Arcade Archives series. It’s probably my most-played Switch 2 game at this point, and as of this writing I am among the top 50 players on the time trial leader board. One car, one track. I’ve been perfecting it for months. When a game sticks on me, it really sticks.

Why Ridge Racer? (aren’t you a Sega man?)

I actually do have a lot of hours in Virtua Racing on Switch and even the old Daytona USA port on 360/PS3, I just never got as far up the leaderboard as I did in this game. My old Joycons started drifting too, making my upgrade to a Switch 2 a perfect time to get back to classic arcade racing.

Why not the PS1 version? That’s the same, right?

It’s actually not! At the time the PR line was that PS1 games were exactly like having the arcade at home, but that’s just not true of what was arguably the flagship title of the PS1 launch. The arcade version was on the System 22 arcade board (it was the later System 11 that was PS1-based) and the PS1 version doesn’t look nearly as good. It also adds faster playable cars, so time attacking it is completely different. PS1 Ridge Racer has been re-released over and over again, but the Arcade Archives port is the first time the original arcade release has been ported to consoles.

(There is always emulation, but something seems very off to me about the MAME Ridge Racer videos I’ve watched.)

Anyway, that first arcade release of Ridge Racer has the kind of arcade purity I’m looking for. Like I said, one car, one track. Learn to drive the course, execute, repeat. Time Trial, the game’s final challenge, makes things even more straightforward by eliminating all rival cars but one. Even when you know exactly what you have to do, it’s a feat of skill and concentration to knock out those perfect three laps in a row.

The Gamesoft Robo Fun Club is a solo labor of love that depends entirely on paid subscriptions from readers to exist. If you enjoyed this piece and you'd like to see more, I'd love to see you become a regular or paid subscriber. Paid subscribers get exclusive posts and keep me fed. Thanks for reading the pitch and hope to see you subscribe!

The rules

I practice the Time Trial course, effectively the game’s highest difficulty. Ridge Racer has four difficulty modes, each of which raises your car’s maximum speed and one of which exposes a new and more difficult section of the course. Time Trial and Advanced are the same course, but Time Trial clears the road of all but one rival and makes your car faster, which in turn makes key parts of the course more difficult.

In short, hug the inside of the track, give yourself room to turn into corners, and take advantage of the game's generous drifting mechanic to take curves fast and smooth without drifting too hard and losing speed unnecessarily. But if you didn't catch that, a turn-by-turn guide follows.

I make no promises that the way I run this course is the absolute fastest and best plan, cause it's not. But I am also top 50 on the leaderboards, so I think I’m probably on to something.

The course

As the announcer counts down and the race queen struts by, you might be looking for your view change button to pull the camera out of the driver’s seat and behind your car. Well, there is none: in this first version of Ridge Racer you drive from a car’s eye view or nothing. If you’re not used to driving this way in games, it’ll take some time to imagine the space around you and learn exactly how much space there is between your camera and the wall. Sometimes you’ll be wrong. Sometimes I cut a turn too close and hit a wall I didn’t think was there.

Unlike most other arcade racing games, this game wants you to hit the accelerator the moment the race announcer says “GO!”, and no sooner. Hit it early and your engine will stall while your rival speeds off, and the announcer will have a laugh at you about it. Get it right and he’ll say that’s “just what I wanted to see~!” As a rule, if the announcer praises you after a turn, that means you nailed it. “You must be one genius of a driver, you’ve gotta teach me~!”

Once you’re off you hit a wide city straightaway: note the huge screen that replays your run from a TV-camera distance and displays your lap time. It’s the kind of detail that blew minds back in the early 90s, a touch that indicates this game is the actual future.

Anyway, the first and most obvious racing principle is to stick to the inside of the track, which is naturally the shortest distance. When you’re going for a time trial record, that means sticking as close as you can to the inside wall without actually hitting it. I like to imagine it with a scene from Initial D, where Takumi hugs the edge of the road so tight that one of his car's tires is actually spinning in the gutter between the road and the mountain’s edge. (You can actually do this in the ID racing games.) I try to think of that tire in my head, “locking” it up against the wall without touching.

In a few seconds you’ll enter the yellow-lit tunnel, another technical “wow” moment in a game that is half tech demo. Ridge Racer does not waste time with its course design, and gets the player right down to sights they could never see in 16-bit 2D. As you continue to hug the wall here, you’ll make an easy turn right into a blocked-off section and plow through the warning signs that were meant to keep you off it. At the end of the tunnel you’ll want to shift left to get ready for the upcoming corner.

One of the ways Ridge Racer manages difficulty is to limit the player’s maximum speed. Novice mode is a gentle Sunday drive, and there you would stroll effortlessly through this turn at full throttle. But at the speed you’re driving in Time Trial, it’s going to require a little finesse.

The basics of cornering, to be brief: start on the outside to give yourself room to turn, then slow down and turn wide into the inside of the curve until you’ve reached its apex, the exact middle point, and then punch the accelerator to leave the turn running.

In a realistic game we might slow down before we take the turn, but in Ridge Racer it goes like this. On this turn we’re going to let go of the gas for a moment and turn simultaneously. Once we’ve slowed down enough— that’s something you’re going to have to feel out yourself— punch the gas and go. We’re looking for a smooth, controlled turn with a minimum of speed lost. If the tires squeal, you’ve overdone it. We need to drop just enough speed that we can run through the next turn without decelerating.

Most of this course should be run at close to top speed; you really don’t need to slow down much. I play automatic transmission in this game, because manual doesn’t have a higher top speed and there is rarely a need to go under top gear anyway.

Exiting this turn you’re going to want to push to the right lane to set up for the long left turn over the bridge, coming up immediately. Position yourself correctly and you shouldn’t need to slow down at all during this turn: get it a little bit wrong and you have a run-ending collision with the wall coming. If I feel myself getting too close, I let go of the gas for a micro-second.

Again, hug the right wall briefly as you drive uphill. When you hit the peak of the hill and a sign becomes visible, move left to the middle of the road. Turn hard to the right and you should get through this curve cleanly. You don’t have clean visibility here, so you just have to trust your placement and timing.

The next turn looks like you're supposed to drift really wide and hit the outside of the road, but that's too slow. Get in the left/middle of the road and turn hard, only momentarily letting up on the gas. You should be able to get through this one fast.

This leads immediately into an S-curve. The fastest way through an S-curve is a straight line, so position yourself to pass straight through the middle. If you’re actually turning all the way through these curves rather than shifting gently, you’re doing it wrong and losing time.

Exit towards the left to take on the first truly tough corner on the course: the only one the announcer will actually warn you about, because if you don't slow down here you will crash.

This turn calls for a slightly longer drift than the one you took on the first turn. A little bit after the sign, let go of the gas and start turning in. When you’re close to the center of the curve (and yes, the mountain wall), punch the accelerator and you’ll drift nicely out of the turn. You can either drift this too hard, drop a lot of speed, and hear your wheels spin out, or you can pass through smoothly with a just-right drift. To give you an idea, I go from top speed to about 210 km/h when I take this turn perfectly. It’ll take a lot of practice and a firm idea of exactly how long to stay off the pedal to get consistent with this turn.

On the next straightaway, you have a speed bump to contend with. Don’t hug the inside (left) wall here or you’ll be in trouble when you land. Give yourself a little distance from the wall and take the jump, keeping the car straight for the duration. You don’t want to hit the ground skidding.

Once you land, there’s a long, high-speed left turn through a tunnel that leads into the next section of the course. The Novice and Intermediate courses have this section cut off and simply loop after the tunnel, but Advanced and TT unlock a tough (and noticeably, completely generic with no decorations or scenery) section that really tests your driving technique. By the same inside-track principle, you want to cut as tight a path as possible through the long left: remember, the shortest path is best.

Once you have entered the advanced section of the track you’re going to make a quick left— keep to the left lane as well— before a deceptive 90-degree right turn. You’re going to zip through this one with a mini-drift. Turn and let go of the gas for only the briefest moment before hitting the gas again, putting you through the sharp turn with barely any speed lost. Two turns afterwards can be taken at full speed easily.

Next up is a big hairpin turn, the longest and sharpest on the course and the only spot in the whole game where you actually have to slow down significantly. The same principle applies here as to the other big turns, but it’s a much longer drift. You should hit the gas at around 190 km/h. You’re going to skid here, so don’t worry about it and do your best to steady the car for the coming series of curves.

You enter a tight double S-curve here, easily the hardest part of the course and something that takes a lot of practice to come to terms with. You need to go through at full throttle, but positioning yourself incorrectly will slam you into a wall. When you come out of the first turn, it’s possible to position yourself so that the coming S-curve is basically just a straight line, like the one you slipped through earlier in the course. But you have to pull right after that to avoid hitting the wall.

It’s tough! I recommend using the Arcade Archives port’s save states to practice this part rather than playing full laps over and over again, like I did before I noticed the port had save states.

The next two turns are easy and to be taken at full throttle: it’s what follows them that is some trouble. In TT you’re going to hit a small sharp turn, then another speed bump, immediately before a big turn. You want to position yourself before you hit the bump, land with your wheels straight, and then take that long turn.

And that is it for the course: all that’s left to do is to take that long straightaway to the finish line, hugging the right wall as you do so. Now do it three times in a row, as the tension mounts with each repetition.

My best lap is about 1’09”, and I’m struggling to pull it off three times in a row in a single run. (This would put me in the top ten players: the best player has a completely insane time like 3'25".) If you’re not shooting for world records or anything, I think 1’12” or so is about the point at which you’ve mastered the course, and the rest is just shaving off time.

This turned out a bit like when I wrote up Rondo of Blood, huh? Just like then, I didn't expect the final product to come out nearly this long. The hardest part was definitely screen-shotting from an arcade game that only has cockpit view on the Switch. As such, I dunno how great those shots of the turns came out, but they're about all I can do. Sorry!

It was my twelfth birthday. I used some money I got from some relatives to buy Kingdom Hearts for the Playstation 2, the “Platinum” release with its of-the-era silver and black box (the UK equivalent of America’s “Greatest Hits” releases). It had been out for a few years, and I had almost bought it on a few occasions before this, but there were also cheaper games that also looked fun. This time I did decide to pick it up, and honestly twelve is the perfect age to get into that game.

Squaresoft’s hybrid of action role-playing game and 3D platformer seemed so cool, as my only encounters with role-playing games before this were traditional turn-based ones. Seeing all these real-time action-packed fights blew my mind, I almost didn’t believe my friend when they called it a role-playing game. The unique menu-based interface did help bridge that mental gap.

I had never spent any time with Final Fantasy at this point outside of watching the odd flash animation parody on Newgrounds, but this was a time when I had firmly gotten into anime. Kingdom Hearts was full of characters that appealed to a kid who had at the time had just finished watching all of Dragon Ball. The aerial acrobatics of the combat also contributed to that appeal. The fact that it was a Disney crossover wasn’t given much thought. On occasion I still forget that it is a Disney game, with how much of a presence its original characters have.

I wasn’t very good at videogames back then, so I cheated using something called an “Xploder” to make myself completely invincible. I still found it challenging to navigate the world. Somehow I would keep getting lost, and anxieties about that still play in my head every time I revisit the game. Despite this invincibility I would also still lose to a few of the game’s harder bosses (likely down to the cheats I was using resulting in some sort of integer overflow glitch, causing me to enter combat with 1 health point). I still had a great time going through the game, getting attached to all the characters, feeling a bit sad at the ending being a bit of a downer, but I still wanted to give the game a “proper try”.

I started the game on its easiest difficulty, without any cheating, and found it to be very difficult. I hit a brick wall in the game’s penultimate area, stuck on a fight against the dragon from Sleeping Beauty. I put the game to rest, and a little over a year later my Xploder stopped working, on the day I received Kingdom Hearts II.

Fast-forward to a little over 20 years later it’s Christmas 2025. I’ve finished Kingdom Hearts without cheating 3 times at this point, each time on its “normal” difficulty level. It felt fairly easy on its original Playstation 2 version, but Kingdom Hearts Final Mix, the version that’s the basis for every modern re-release, is balanced to be a little more difficult. Each time I went to that version I would find the later battles a bit of a challenge. Feeling brave, I began my next run of Kingdom Heart Final Mix on its hardest difficulty. This was the easiest time I ever had with the game, but also the most fun.

Rather than just run through the main events of the game straight away, I decided I would try to do everything I could. In a game that I frequently revisited, I would finally be well and truly finished with it. Doing so let me engage with the game in ways I hadn’t really considered before. It turns out all of the things I had skipped over actually made the game more interesting.

The most time-consuming optional activity in Kingdom Hearts is its crafting system. It doesn’t have a huge amount of items, but in order to get all the materials for it a player needs to do a fair amount of grinding. I never thought of Kingdom Hearts as a particularly grindy game, but seeking these items out unveiled that the game could turn into one. This particular grind involves gimmicky fights that can’t be won simply by beating up the enemy, which is why I often skipped them before, you couldn’t just mash the attack button.  I have also played enough games where the grind in optional activities is what motivated me to focus on the main game, thankfully this wasn’t one of those.

It turns out those gimmicky fights made the game more exciting when engaged with properly. They’re weird little minigames that involve doing unusual things like playing a shell game, deflecting projectiles, or trying to do the biggest possible combos. I would’ve thought that having to repeat these things would make me sick of them, but the opposite happened, I appreciated them more. Challenging them again meant that I couldn’t just finish them by rushing through, I had to understand them in order to optimise my time spent grinding. Each of these fights had me using more of the game’s magic and abilities in ways that deepened my understanding of how they worked, and I carried through that into how I played the rest of the game.

It’s that knowledge that ended up making the later parts of the game significantly easier. Late game challenges became short fights, some only taking me around a minute to defeat them, including the dragon mentioned earlier. The reward for completing the crafting system was also the game’s most powerful weapon, and all that grinding boosted my character’s level enough that even the game’s optional superbosses weren’t too difficult. It felt good to finally finish parts of a game that had seemed impossible so long ago, alongside the comforts of childhood memories it all brings forth. The game had become significantly easier, but in a way that I had earned through effort and understanding. I didn’t beat those superbosses through brute force from my extra strength, I learned what abilities would be useful in offensive and defensive contexts and applied that.

Playing this as someone with a much more intense knowledge of Final Fantasy hasn’t affected how I feel about the game, it still holds up to me in the same way it did when I was twelve. It would be interesting to see if the same were to be true of its follow ups.

I didn’t watch Ripley at first. The black and white was a barrier (I know, but shut up, I was not alone). I have no trouble with black and white photography, so why black and white film? Perhaps because film feels more real to me, meaning it should be more real, meaning color. Either way, I hesitated to watch Ripley. At worst, I thought the black and white would be pretentious, at best ... well, I couldn’t think of that scenario. So I kept putting it off and putting it off as praise was heaped on this latest of a million adaptations of Patricia Highsmith’s thriller, The Talented Mr. Ripley. Then close friends started to recommend it, then I felt like a jerk (then I felt pretentious for avoiding it). So I put it on, and … wow.

By now most of us know this story. In 1950s New York, tenement-housed petty fraudster named Tom Ripley (a gecko-esque Andrew Scott) is hired by the wealthy father of a trust-fund kid, Dickie Greenleaf (a charmingly befuddled Johnny Flynn), to bring Dickie back from Italy, where he is frittering his life away painting terrible paintings while his girlfriend, Marge (Dakota Fanning in full Mona Lisa mode), writes a terrible book. Basically, instead of doing what he was paid to do, Tom falls for his target’s lifestyle, a lifestyle that has up until now been elusive to him. It’s this out-of-placeness that the black and white of this series, created by Steve Zaillian, fully exposes. Tom is often in Ripley’s monochromatic frames alone, disconnected from other human beings, often overwhelmed by his surroundings—dwarfed by buildings and their interiors, constricted by hallways and stairs—a man lost at sea, never comfortable, a guy who can get lost in a Caravaggio but can’t quite settle into the lushness of his new environs.

CANON FIRE is made possible by the generous contributions of readers like you. Support more writing like this on Patreon. Thank you!

With a bubblegum pop aesthetic and low stakes story about rescuing your cat, Super Crush KO takes the essentials of stylish action games like Devil May Cry and Bayonetta and crushes them down to a 2D plane. There’s a familiar air combo focused rhythm; punches flow into launchers and directional specials, allowing you to close the distance and interrupt enemy attacks, with a rapid fire blaster for continuing juggles and keeping a combo alive as you approach.

Relaxed from the technical movement and execution requirements of its inspirations, Super Crush KO instead focuses on flowing through stages uninterrupted. Simply finishing a stage is a relaxed activity, but a mobile game style structure  pushes you to replay stages and aim for perfect combos, higher ranks and side objectives. 

The casual attitude makes it enjoyable to play when you want the same feeling character action games give you, without the intensity and sensory overload they often come with. So much of the genre’s fun comes from the absurd amount of tools you’re given for player expression that simplifying it might seem counter to the appeal, but Super Crush KO manages to keep that same satisfaction intact. 

With fewer tools each of them becomes more effective, allowing you to traverse space very quickly, with the speed increasing as the game introduces environmental obstacles that launch you at high speed vertically and horizontally. Dodging in Super Crush KO is almost platform fighter like, giving you invincibility and acting as a good way to reposition yourself, with only a short cooldown to limit you. Your normal combo is unremarkable, but a suite of directional specials gives you strong options to close the distance, interrupt incoming attacks, and launch enemies.

The aerial “Air Pop” in particular allows you to dash in all eight directions, homing in on enemies and chaining up to three in a row to extend the duration of an air combo near indefinitely. An energy meter limits the uses of special attacks, but defeating enemies fills it quickly, creating a loop where you can chain specials together to quickly defeat enemies, throw some regular combos and bullet juggles in for filler, and inflict enough damage to put you right back to full energy.

Super Crush KO lays on the feedback, with hitstop and screenshake for everything from big hits to dodging through bullets, and a ranking and score log tallying up every action so that even small actions feel satisfying. If I have any complaints it’s that the bosses don’t take advantage of the flexibility in the air combo system, making them more static than the regular stages, and that Air Pop’s short startup, multidirectional movement and homing properties kind of make it too useful, relegating other specials to more niche situations. 

As someone deep into the beat-em-up genre, Super Crush KO is an easy, casual game to come back to, with plenty of side objectives to attempt mastery. It’s got the same compelling loop that time trial platformers like Dustforce have, with less focus on the technical mastery of the inputs, but the ways you string together moves. The welcoming aesthetic, pop colors and cute character art all enhance that low stakes mood. If you’ve got any interest in the art of crafting an air combo, give this one a shot.