As a huge fighting game fan I’m always excited when a new game comes out. No other genre captures my attention in the same way. The feeling of discovery, the community and competitive spirit—a good release means potentially thousands of hours of playing and learning, not to mention all the sharing and talking with fellow players.
This week I’m excited about Guilty Gear Strive, releasing April 9 on PS4, PS5, and PC. There’s an open beta this week, and Ars got early access to try it out, including the newly improved netcode.
I’m happy to report that the beta is already doing the right things to make an online game feel as much like an offline one as possible.
The offline experience
In pre-pandemic days, which already feel like years ago, I would host a weekly fighting-game night. People gathered in my garage to run friendly-but-competitive sets, help each other learn games, and just talk about life. Fighting games to us aren’t just about competing—they’re about community and connecting with people.
For me, there’s just no purer gaming experience than playing against another human being. No CPU opponent or single-player narrative can bring that same feeling of give and take, testing your skills and reflexes but also your ability to adapt and dig into the psychology of your opponent.
If you’re the better player? It’s a chance to teach someone or test ideas against someone more forgiving. The moment when you see them avoid the setup they’d been walking into all night or challenge the move you’d been bullying them with can feel as good as you winning.
If the tables are turned, and you’re the weaker one, it’s a chance to learn, and that s much more satisfying when you finally take a match. One night, I ran a first-to-20 against a friend who’s a much stronger player than me, I lost the set 3-20. Those three games I took? Best feeling in the world, even after getting my butt handed to me.
The real joy though is when you find an opponent you’re roughly equal to. Trading games, constantly pushing yourself to find an advantage, adapting and watching them adapt back, that’s when fighting games shine in a way few other games can. Could be a good friend, or a total stranger, but it’s like having a conversation in another language.
A year ago, as COVID began to pop up in Los Angeles County, I had to cancel my night. All the other weeklies, the tournaments, friendly sets at your friend’s house—they all dried up. Offline fighting games became an endangered species around the world, and the only way to get our fix was to play online.
Fighting in an online world
Fighting games, by their nature, are reliant on reflexes and being able to quickly assess data. We count frames, we look for things we know are unsafe we can punish, we watch for the telltale movements that might indicate an overhead attack is coming so we can shift our blocking to high. If someone is dashing up in your face, can you react fast enough?
In short, it’s a crummy genre to cram through Internet tubes. Any lag can destroy the feeling of the game. Without offline play, reliant entirely on the Internet, what is a fighting game fan to do?
Luckily, there’s a bit of clever networking tech, called rollback netcode, that can alleviate most of the disadvantages of distance and physics (doesn’t matter how good your Internet connection is—there’s no beating the speed of light).
Unfortunately, not every fighting game has rollback netcode, particularly ones from Japanese developers who are the titans of the genre but have been slow to adopt the technology. They’ve relied more on delay-based netcode, which provides a much less consistent or enjoyable experience. Or, in the case of Street Fighter V, half-baked rollback that provides as many frustrating online experiences as it offers good ones.
We’ve already published what is probably the most comprehensive explanation of fighting game netcode and how rollback works. It’s a deep dive, with both technical explanations and a lot of short video clips to demonstrate the concepts—I highly recommend it for the curious. Here’s a quote from it that acts as a quick summary if you don’t want to dig in:
Since a game’s choice of netcode can never magically change the distance between a player and their opponent or prevent networks from dropping or delaying information, you may wonder how one netcode strategy could be drastically better than any other. The key lies in how the netcode handles uncertainty.
When there is no information from the remote player, delay-based netcode needs to pause and wait, as described in detail on the previous page. Rollback’s main strength is that it never waits for missing input from the opponent. Instead, rollback netcode continues to run the game normally. All inputs from the local player are processed immediately, as if it was offline. Then, when input from the remote player comes in a few frames later, rollback fixes its mistakes by correcting the past. It does this in such a clever way that the local player may not even notice a large percentage of network instability, and they can play through any remaining instances with confidence that their inputs are always handled consistently.
In short, good rollback feels much more like offline. Your timing is the same as offline, and with reasonable ping, it’s almost indistinguishable from sitting next to your opponent. Even matches played from California to New York, or across the ocean can be very playable, with minimal visual skipping.
Enter Guilty Gear
Back in late 2019 I interviewed Daisuke Ishiwatari, the creator of the Guilty Gear series, about the upcoming release of the latest version of his game. I didn’t know the pandemic was just around the corner, but I did want to ask him about rollback netcode. The previous Guilty Gear games didn’t have it, and their online experience suffered for it. Even in a world where offline play exists, it’s much more convenient to hop online at home to play whenever you want. Did the designers intend to do the work to add rollback to the game?
Daisuke’s response didn’t engender a ton of confidence, but it did leave the door open for hope:
Where we are right now at ArcSys, in terms of rollback netcode, is we haven’t really arrived at the conclusion that we’ll need a super programmer as much, as the engineering team is kind of divided, actually. There are some that say this would be really good and others that say, you know, implementing this wouldn’t really work with the Guilty Gear system. And it makes sense for a game like Street Fighter, but how Guilty Gear is designed—this wouldn’t really fit the bill. So we’re actually right in the middle of investigating on the engineering team how that might look.
His quote from my interview was widely spread on Twitter as part of an outcry from fans for the developers to take netcode seriously. It became difficult to find any discussions of the game that didn’t include hopes about rollback. The question was, would the developers listen?
Turns out they did, and I was able to hop online and play some sets to get a hands -on feel for how both the game and netcode feel.
To begin with, Guilty Gear Strive is in general a gorgeous game. Arc System Works are masters of 3D model cel shading, creating characters that harken back to the original arcade-era sprites but that are able to spin in three dimensions and respond to dynamic stage lighting. There are times, however, when the camera movements can feel slightly excessive or odd, and the lighting, while impressive, can lead to some stages obscuring the characters more than others.
Neither issue is a dealbreaker, but they do show how tricky it can be to balance clarity for gameplay while adding dramatic flourishes. The previous release, Guilty Gear Xrd, didn’t use so many lighting tricks and kept the characters bright against the background at all times. I preferred that look, even if it was a little less fancy.
The other fly in the ointment, and something I’m particularly sensitive to as a graphic designer, is some very odd user-interface choices. Guilty Gear has always leaned into its heavy metal, almost steampunk aesthetic. This time around, the interface is oddly sparse and minimal, to the point where it can be hard to read some of the meters quickly at a glance. There is a multitude of fonts and styles that often clash, and the combo count and some of the counterhit notifications can get comically large, dwarfing everything else on the screen.
We’re still in a beta, but with the release date so soon I’m not confident there will be any significant interface changes before launch. They don’t detract too badly from the gameplay, but they are a bit distracting, and more importantly to my eyes at least, they’re an ugly mark on an otherwise beautiful game.
From a gameplay perspective, Strive has been simplified a bit from previous releases, which did have a reputation for being a bit difficult to pick up. We’re too early to say how long-time Guilty Gear fans will find the overall depth, but if you’re a newcomer, it’s definitely friendlier and more approachable, while retaining much of the basic feel of the original games. If you’re comfortable playing Street Fighter, you’ll be just fine with the level of complexity Strive offers in terms of inputs, execution, and general gameplay. Staples of the genre, like the quarter-circle fireball motion, are still present, so do expect spending some time getting comfortable if those aren’t already second nature to you.
Since this is a beta and not the full release, I’ll leave the general impressions there so I can focus on the last and most important thing. How good exactly is the rollback netcode? Is this a pandemic-friendly title you can play at a high level online?
Rollback to the future
The good news is the netcode is great. Because we have early access to the game the servers were nearly empty, which made it easy for me to find Culture Editor Sam Machkovech, since we were literally the only people in the West Coast lobby at the time. Los Angeles to Seattle was utterly smooth on both our ends. If you look at the screenshot I captured during our match above, you can see we had around 47ms of ping. With rollback, most people consider anything under 150ms playable, and if you don’t mind a little more visual skipping, you can go higher and still have a good experience.
I watched a match between fighting-game commentator and streamer Sajam, who is in Long Beach, California, and Footsies developer HiFight who lives in Japan. The ping stayed a steady 124ms, and the rollback frames never went higher than 3. The matches on Sajam’s end looked very smooth. We’ll see more reports as the beta opens to more people, but the early word seems consistent, Arc System Works did the netcode right.
This is the power of rollback. You can play people a few states away, but you can also play people across the ocean. Closer and lower ping is always better, but when half the world is playable, your pool of opponents is huge. That’s makes it easier to find people close to your skill level, and it means that even if your game of choice is popular locally, there’s probably always someone within playable range waiting for matches.
Now we just need more devs to see the power of good netcode. Street Fighter V‘s rollback is not great, but maybe Capcom will fix it for Street Fighter VI, which is currently under development. It’s especially key that developers in Japan recognize the benefits. Not only is Japan still the place where most fighting-game development happens, but American devs like NRS and Iron Galaxy already figured out great rollback for games like Mortal Kombat 11 or Killer Instinct. (This is a great interview with one of Killer Instinct‘s devs about netcode if you’re interested.)
In the meantime, I’m much more interested in learning Guilty Gear now. Hopefully by the end of 2021, we’ll be out of the worst of the pandemic, vaccines will be widely distributed, and I can start hosting a fighting game arcade night again. Until then, I’m stuck with online like everyone else, and the games with good netcode are the ones I want to be playing.