Is free-to-play the future of fighting games?

When it comes to video game genres, fighting games tend to be a step behind the curve. The fact that it took an entire pandemic for developers to finally look into better netcode for fighters speaks volumes.

At an even more basic level, the way fighting games operate has been called “archaic” by players who feel that the genre is in need of a switch-up to make them more accessible to modern, mainstream audiences. One of the fighting game community’s main talking points at the moment is free-to-play (FTP), a popular business model that fighting games have largely shied away from, even as games like shooters find success with the practice. If the genre wants to remain competitive in today’s modern landscape, it may need to make that pivot soon — but it’s not as simple as it sounds.

The barriers of the genre

A part of the perceived “need” for the change comes from the sheer number of fighting games out there right now and on the horizon. Maximilian “Dood” Christiansen, a high-profile fighting game content creator, has studied and enjoyed the genre for decades and follows this mindset.

Christiansen has previously noted that older fighting games were able to stay relevant due to a lack of options in the past, which meant one game could hold a player’s attention more easily. Today, there’s a larger variety of quality titles to choose from. However, each brings its own extra costs, from the price of the game itself to DLC to subscriptions that allow players to fight online. Fighting games are expensive enough as is, so your average player isn’t likely to invest in several games simply out of curiosity.

“Most fighting games are gonna result in the same problem until there’s big sales, expansions, or a reason to check out an old thing, because a new thing is really cool,” Christiansen states in a video titled “Please make fighting games free to play“. “But there’s always the barrier, right? You have to pull out your wallet and spend money. If you didn’t have to do that and shit was just free and you can just fire it up and try it, then of course you’re going to find people. That game will always have people playing.”

That paralysis of choice leads to another major problem that Max also brings up: Massive skill gaps. If you’ve ever gone online with older fighting games as a new player, there’s a 90% chance you’ll run into a veteran who can destroy you as they make a sandwich, brush their teeth, and update their LinkedIn account at the same time. For many, it can make games feel “unfair” for new players, creating a barrier to entry.

That’s not to say this issue wouldn’t be a thing with free-to-play fighters as well. However, the nature of this “try before you buy” model would lead to a greatly diverse and potentially ever-growing pool of players at different skill levels. This means more beginners can match up with other beginners instead of being pitted against veterans of the game.

The future of fighters?

A free-to-play model is a proposed solution to those hurdles. Players like Christiansen see going free-to-play as a way to not only keep the initial investing audience around, but to welcome in new players too. That could theoretically reduce the dramatic difference in skill level that new players experience.

Fighting games are expensive compared to a lot of modern multiplayer games (many of which are free-to-play in some way). There’s a chance a new player won’t enjoy the game, but not find out until spending hours learning. One might have the only character they enjoy locked behind DLC, meaning not just paying for the game, but additional content as well. All these barriers are pushed to the sidelines with the FTP model.

While not the most popular with the fighting game community, there are titles that have gone this route. Killer Instinct and the Smash Bros-esque platform fighter Brawlhalla embraced a FTP model. I personally have invested tons of hours into Killer Instinct and found that I love how it handles its model.

When I booted up Killer Instinct for the first time in years, I knew there were only a few characters I was interested in. After researching them, I bought those characters and quickly found my main at a much lower price than purchasing a full fighting game and its DLC. Killer Instinct even allows players to access non-purchased characters as training dummies, which is an ingenious move and something I wish all fighting games would do. It means I don’t have to waste money buying characters I don’t want to use, but need to practice against if I want to stand a chance online.

While it seems similar to the standard fighting game DLC practices, it stands out in an honest manner by allowing curious players to test out two characters for free to learn and find out if they like the game’s engine and mechanics. There’s also a third free-to-use character that rotates throughout the roster on a weekly basis. Implementation such as this gives players the ability to try before they buy. Instead of purchasing an entire package, they can buy the pieces they want. It’s a smart formula for a genre where many only want to play as three or four different characters.

Good implementation of FTP in fighting games can be seen in Brawlhalla, which still boasts 10,000 players online daily, according to Steamcharts. With the upcoming Warner Bros. crossover fighter Multiversus being an FTP fighter with mainstream characters, it may even break those numbers and more records of the genre.

Free-to-play, but not flawless

It all sounds like a slam dunk in theory, but the reality isn’t so simple. Traditional fighting games like Killer InstinctFantasy Strike, and Dead or Alive 6 that have gone FTP still haven’t been able to hold on to a consistent player base despite the model. Brawlhalla has found more success, but it features a more casual playstyle akin to Super Smash Bros. rather than something “hardcore” like Street Fighter.

Free-to-play could introduce new issues for the genre too. Professional fighting game analyst, commentator, and once competitor Sajam talks about how games like Fortnite seemingly drop a never-ending well of skins, wraps, effects, emotes, and more to keep players in their wallets and spending in the item shop. While not the case for every fighter, certain fighting game studios just don’t generate the revenue needed to keep up with such a model. The FTP model has also been poisoned by games that tried, and failed, to make the model work.

Wasn’t Tekken Revolution Free to Play In a screwed up way? We had stats that we can improve which changed the gameplay. We also had randomly occurring critical hits that we had no control over

— Rauschka (@Rauschka_tk) April 5, 2022

Games like Tekken Revolution and Dead or Alive 6: Core Fighters left sour tastes in the mouths of community members when they exposed how scummy the model can feel. Tekken Revolution featured pay-to-win stat boosts and only allowed players to fight in five matches at a time unless they bought a ticket to continue. A general lack of content and support for DoA:6 led to its quick death.

There’s also the fact that fighting games tend to be harder than the other competitive games out there. They aren’t made to be easily digestible to a casual audience like Smash Bros. and its clones like Brawlhalla and Multiversus. Just because you make it free to pick up and learn how to do a quarter-circle forward to half-circle back motion, doesn’t mean many will still want to do so. The genre just isn’t as approachable as, say, a shooter — and that’s another hurdle that many fans used to write off FTP.

While there are issues that can come with going free-to-play, it’s still an approach worth exploring. The positives outweigh the negatives and many community members are on board for a game like Street Fighter 6 taking a Fortnite-like approach. Will developers be willing to upend a classic genre so easily? If the long-archaic past and present of fighting games is anything to go by, probably not. But if a major release ever did, it’s possible that others could follow suit.

Many are looking to Riot’s upcoming fighting game, Project L, as the litmus test of the model in the genre thanks to the company’s past with FTP. With its reach, knowledge, and possible support from the fighting game community, it could easily be the game to revolutionize fighters going forward. However, there are a lot of “ifs” in that argument, so we’ll have to see if it can win over the skeptics.

Editors’ Choice

Repost: Original Source and Author Link


MultiVersus VS the best crossover fighting games ever made

Today we’re going to take a little trip down memory lane with the best crossover video games in existence alongside the newest of their breed: MultiVersus. The newest game was made by Warner Bros. Games. This game showcases a collection of characters from franchises under the Warner Bros. umbrella – that means everything from Batman to Cartoon Network. If you’ve played a LEGO multiverse game before, this should all look sort of familiar.


If you’ve seen Space Jam 2, you might recognize the starting lineup in MultiVersus. This is because the intellectual property that’s part of MultiVersus is effectively the same as what was made available to the people making Space Jam. Basically every character involved in MultiVersus can also be seen in a movie or TV show on HBO Max.

That means Batman, Superman, Shaggy (Scooby-Doo), Arya Stark (Game of Thrones), Finn and Jake (from Adventure Time), Wonder Woman, Garnet and Steven Universe, Tom and Jerry, Buggs Bunny, and Harley Quinn. MultiVersus is expected to have a release date in the year 2022.

Battle Soccer: Field no Hasha!

Back in the year 1988 this genre started life as a soccer fighting game called Battle Soccer: Field no Hasha! This game featured Ultraman, Gundam, Kamen Rider, and Godzilla! You might’ve played this game on Super Famicom, or a region-unlocked Super Nintendo if you were lucky!

This game spawned a pair of sequels, too! It would not be a shock to see this franchise return, one day, three decades after its most recent installment hit store shelves. It’d be a difficult sell as a soccer game, but stranger things have happened!

Marvel VS Capcom

Capcom had a string of successful crossover fighting games over the past several decades, but the most mind-blowing of these, for me, must have been the original Marvel VS Capcom. Take a peek at this presentation of the game from all the way back in the year 1999!

If you still have a Sega Dreamcast lying around in a cupboard somewhere, and you’ve never played this game, it’s time to change your life for the better. This game includes a barrage of characters from Capcom and some of the most iconic iterations of Marvel Comics characters ever shown in a video game. And the boss is the greatest Marvel villain of all time: Onslaught!

Super Smash Bros.

When Nintendo released the following commercial in advance of the release of Super Smash Bros., it was immediately clear they had a monster on their hands. Super Smash Bros. took the wide, wide world of super successful Nintendo characters and jammed them all into a game that allowed players to bash one another all night long.

The first Super Smash Bros. was and is a must-have title for all owners of the Nintendo 64. The follow-up game Super Smash Bros. Meelee was an absolute staple for the Nintendo GameCube. This franchise rolls with characters from Super Mario, Donkey Kong, Pokemon, Star Fox, Kirby, Yoshi, Metroid, and Zelda. Later entries started to get real weird with it, including characters like the Wii Fit Trainer from the Wii Fit platform, and the dog and duck from Duck Hunt!

Honorable Mention: Kingdom Hearts

Though not a fighting game, Kingdom Hearts really took the whole multiverse crossover idea and ran with it. Disney Interactive Studios and Square Enix (formerly Squaresoft) created a crossover that explored the Disney Universe in a way only the creators of Final Fantasy could deploy.

The creation of Kingdom Hearts opened the door to the idea that characters from all franchises could cross over into others with the greatest of ease. And with the greatest of commercial success.

Repost: Original Source and Author Link


The Best Fighting Games for PS5

There’s an entire generation of gamers out there who cut their teeth in the old arcades. Standing side by side with your opponent was the only option for real, player vs. player competition back in the day, and the king of the competitive genres was undoubtedly fighting games. The simple premise of two fighters squaring off against each other spawned some of the most popular and successful gaming franchises of all time. To this day, games that once made their debut in arcade cabinets are releasing new sequels for your home consoles to play from the comfort of your own sofa.

Whether it’s against the computer, online, or playing locally on your couch, fighting games offer a pure test of skill that many compare to games like chess. Unlike chess, each fighting game has its own rules, systems, matchups, and more to learn. Factor in things like pure 2D vs. 3D fighters, extra modes, and character rosters, and any two fighting games can look almost entirely different. Plus there’s always the art style to consider too. The PS5’s hardware is perfect for running fighting games, which demand rock-solid performance, and it’s home to all the biggest franchises. If you want to throw down in a one-on-one test of skill, there are the best fighting games for PS5.

Further reading

Guilty Gear Strive

Guilty Gear has always been seen as a somewhat second-tier fighting game series below the major names among the general masses, but those who know the series have been singing its praises for years. Guilty Gear Strive hit after Ark System Works finally broke into the mainstream with the perfect pairing of their anime art style with the biggest anime franchise of all time, Dragon Ball. Dragon Ball FighterZ would probably have made this list before, but now that Guilty Gear Strive is out, there’s no denying this is the better game. Sure, you don’t get 10 versions of Goku to play as, but some consider that a plus.

Just looking at this game is almost unbelievable. Arc System Works has always been a wizard when it comes to making 3D character models look like near-perfect 2D ones, and Guilty Gear Strive is perhaps the most impressive yet. The animations are unreal, fluid, and brimming with more detail than you can even perceive. The cast is wildly varied, with no “clones” in sight. Arc System Works has refined its easy-to-pick-up, hard-to-master gameplay in fighting games here. The only weak element that has to be pointed out is the single-player component. The story mode is essentially just a series of cutscenes that won’t mean much, or even make a lot of sense, to anyone who isn’t familiar with the frankly absurd and convoluted Guilty Gear lore.

Street Fighter 5: Campion Edition

Ryu and Chun-Lee posing next to the logo.

Ah, there’s nothing quite like the old, familiar, and reliable fighter. Street Fighter set a new standard in fighting games way back with Street Fighter 2, and with incremental improvements and features, has held strong as the most iconic game in the fighting game genre. That being said, after the series made a revival with Street Fighter 4, the next iteration didn’t come out in a state many were happy with. At launch, Street Fighter 5 was quite bare. The roster was limited, mechanics and online not quite up to standard, and essentially no single-player content whatsoever. There wasn’t any form of story mode at all until later down the line.

Thankfully, Capcom stuck with Street Fighter 5, and it is now at the point where it is a complete package. The story mode is included, multiple seasons of DLC characters have filled out the roster, and balance changes have made it a much more fair and accessible game to jump into. The act of getting that extra content will be expensive, however, in either time or real cash. You can unlock anything in the game by earning in-game currency, but be prepared to grind out a lot of matches. Alternatively, you can purchase one of the newer editions that come bundled in with most of the extra characters. Either way, Street Fighter 5 has revived itself as the king of fighting games. At least, perhaps, until an eventual Street Fighter 6. 

Tekken 7

Jin vs. Akuma in Tekken 7.

Tekken has a very distinguished style that makes it at once incredibly addicting and also so mechanically deep and free-form that you will continue to feel like you’re learning how to play dozens of hours in. Tekken 7 follows the very toxic and dysfunctional Mishima family yet again, with a new story mode. It does its job, swapping you between all the main characters and stages, but the actual content and delivery of the story aren’t really worth writing home about. Still, it is at least another option for single-player content in a game that isn’t too robust in that regard.

Tekken was, and in a lot of ways still is, notable for being a fully 3D fighter. We don’t just mean having 3D characters, but actually utilizing 3D arenas to fight. That means positioning is more complex than just moving closer or farther from your opponent — you need to consider sidestepping as well. The way you attack is also uniquely Tekken, and sticks to the well-established roots of the series, but adds on top of them some new features like Rage Arts and Power Crushes. Plus, the feature of the game slowing down to slow motion on dramatic moments sounded like a terrible idea when we first heard it, but actually does a really good job of highlighting tense moments.

Read our full Tekken 7 review

Mortal Kombat 11 Ultimate

Frost sticking icicles through Scorpion's chest.

Another arcade phenomenon that has somewhat quietly become the best-selling fighting game series of all time, who could forget the bloody, mature, violent, and intentionally misspelled Mortal Kombat? Granted, this series has been going on for a long, long time and has arguably had more bad games than good. The transition to 3D in particular was not kind to the MK games up until NetherRealm decided their long, overly obtuse narrative needed a reboot just as much as the gameplay. Mortal Kombat 9 retconned the entire franchise back to the start, cutting the roster down and focusing on what made the game fun while also setting a new standard for fighting game stories.

Two sequels later, Mortal Kombat 11 is probably the most popular the series has ever been. The story continues to be fantastic, well acted, and freakishly well animated. New and existing characters are all great to play, and of course have their own unique Brutalities and Fatalities (plus some other hidden moves) to enjoy. If you like seeing skulls shattering and blades tearing through internal organs, only for the recipient to shrug it off and keep on fighting, there’s no alternative to Mortal Kombat 11. You also won’t find the Terminator facing off against Joker, Spawn, or Rambo anywhere else either. The PS5 version of the game is a free upgrade and lets you enjoy every drop of blood at a dynamic 2160p resolution.

Read our full Mortal Kombat 11 review

Soul Calibur 6

Geralt and a samurai standing side by side.

If you want a fighter that’s more weapon-focused, Soul Calibur 6 is the clear winner. Aside from the actual gameplay, which we’ll certainly speak to, the Soul Calibur series has one unique feature that no other major series has really attempted to replicate. That feature is a ridiculously robust character creation system. This is more than just a simple reskin of an existing character. With Soul Calibur 6, you can not only fully customize your own character’s appearance, but also give them their own combos and weapons. Yes, most of these will be taken from the existing roster, but the level at which you can mix and match moves leads to characters that feel completely new.

Perhaps where Soul Calibur 6 stands above every other entry on this list is in single-player content. Aside from making your own character, there’s a deep progression system to get into, a pretty good story mode that’s tailored to each character, mission mode, and arcade mode. Versus mode is naturally where you’ll spend the most time, and it remains solid as ever. The focus on weapons makes you approach each fight differently, which lends itself perfectly to it utilizing full 3D space. Just watch your positioning so you don’t get knocked out of the ring! And, since fighting games seem to be mandated to have some level of crossover characters, you can even play as Geralt from The Witcher games and 2B from Nier: Automata. 

Under Night In-Birth Exe: Late(cl-r)

A sword weilder leaving a red trail with his sword slash.

No, that’s neither a typo nor me having a stroke while writing out a game title. Under Night is one of the few game series that can outmatch the Kingdom Hearts series for most ridiculous titles. Names aside, this is the newest entry in this series that is essentially the poster child for “anime fighters.” It’s so anime that it isn’t just a fighter, but the story mode is essentially just a visual novel. Seriously, Chronicle mode is a full 23-chapter visual novel that dives deep into the characters, their histories and relationships, and the entire plot of this bizarre game. Thankfully the Arcade mode is more of what you expect from a fighting game story mode.

Speaking of modes, Under Night In-Birth Exe: Late(cl-r) has almost too many modes. You’ve got Arcade, Chronicles, Versus, ranked and unranked online matches, local, Score Attack, Time Attack, Survival, Training and Tutorial, and more. The game has a respectably sized cast of 21 default characters, and each one is lovingly detailed and animated. Seriously, this game is second only to the output of Arc System in terms of showing off beautiful anime-style characters. There is a lot to learn in this fighter, but the game has more than enough tools to get you going thanks to the in-depth tutorial mode. If you wanted an alternative anime fighter that makes you feel like you’re playing your favorite shonen show, don’t be late(cl-r) for this one.


Characters battling in Brawlhalla.

There’s really no substitute for Super Smash Bros. Many have tried, but there’s no game that can match that game’s free-flowing, expressive fighting system with the roster of iconic characters. That’s why Brawlhalla decided to forgo attempting to outshine Smash in star power, and instead delivered a refined, tight, and most of all solid gameplay experience even when playing online. Oh, and did we mention they don’t even charge you to play? That’s a big point in their favor. That being said, if looks are important to you, then you might feel a little disappointed with this one. Brawlhalla does look like a free game, for better or worse.

All that being said, Brawlhalla is doing a lot of things right. Aside from being free, the game has full crossplay between PS5, PS4, Xbox Series X/S, Xbox One, Switch, PC, and even mobile. The simple art style and character designs have let the roster grow to more than 50 characters so far. You can play with up to eight people locally or online in various modes, including ranked 1 v 1 and 2 v 2 modes, and wacky side modes like Brawlball, Capture the Flag, and Kung-Foot. Like Smash, the core game itself is a simple platform brawler where you duke it out on stages floating above pits with or without item drops. The community around Brawlhalla is so strong that it even has its own Esports league, meaning there’s no limit to how far you can go with this game.

Virtua Fighter 5 Ultimate Showdown

The cast of Virtua fighter posing.

Virtua Fighter used to be one of the big names among the fighting game titans like Street Fighter, Tekken, and Mortal Kombat. However, it’s been years since the series got any new entries. While not the brand-new entry fans have been craving, Virtua Fighter 5: Ultimate Showdown is a welcome return for this classic fighter. Originally released in Japan in 2006, this updated, and upgraded, edition shows that the core formula of this series is as solid as ever. It was even re-released in 2010 on the PS3 as Virtua Fighter 5: Final Showdown, but this ultimate version manages to up the ante even more.

Every character looks better than ever, though you’d have to squint a bit to make them look current-gen, especially if you stare at the backgrounds too closely. What the game lacks in graphical powers, though, it makes up for in performance. The game is rock solid, and the online portion even got some new additions like tournaments, spectator modes, and a league feature. Like Tekken, Virtua Fighter games are built for those who want to dive deep into a game’s mechanics. Every move has its perfect time, place, and distance to be used, as well as combo potential. Each character in the game can feel completely different to fight if given to two different players. If you’ve got the ambition, this is a game that will reward your time investment like a few others.

Included with PS Now

Samurai Shodown

A samurai slashing through his opponent.

If you want a different misspelled type of showdown, Samurai Shodown is another series that seemed like it might’ve been forgotten until recently. This game is arguably the most brutal fighter on the list. And we’re not talking brutal in the Mortal Kombat sense. No, Samurai Shodown is brutal in the sense that one mistake probably means death. Unlike essentially every other fighter, especially Mortal Kombat, this game tries a little harder to be realistic in how a fight between two Samurai would play out. Rushing in, swinging wildly, would probably end in a quick death, which is exactly how this game functions.

Samurai Shodown’s gameplay is far deeper than it may look if you just watched a couple matches based on how quickly they go. I mean, seeing one heavy strike deal about 30% of an opponent’s life bar, and special moves doing around 90%, would probably make you think the game is more like Nidhogg. But there are systems for attacking with your weapon, while unarmed, kicking, clashing, guards, parries, guard breaks, counters, blade catches, and more. With plenty of post-launch support, including three seasons of DLC characters, plus crossover characters with SNK, this is a full package. For the calm, observant, and patient fighters who like to outthink their opponents, Samurai Shodown is that mood distilled.

Marvel Vs. Capcom: Infinite

Marvel vs Capcom Infinite Megaman and Captain Marvel.

Right off the bat, yes, Marvel Vs. Capcom: Infinite dropped the ball with the roster. No one is denying that. However, if you can look past the lack of X-Men, or even series vets from the Capcom side, the game itself is actually really good. You’ll also have to look past the admittedly unattractive character models, too. They did drop the three-man roster down to just two, but except for the dedicated player, it was difficult enough to parse what was going on when up to six characters were jumping in and out of frame, firing off a massive special, and jumping out. MvC: I streamlines things a bit, but is way more accessible because of it.

The story mode is … there? It’s really a waste considering the amount of potential there is to craft a fun, off-the-wall story with these two worlds of eccentric characters clashing, but whoever was in charge, Marvel or Capcom, just dropped the ball big time. Thankfully, Capcom was in charge of the mechanics, and they know how to craft a good hyper fighter. It’s really a shame it was buried beneath bad press, marketing, and even a tight budget. Plus, the integration of the Infinity Stones as a mechanic could’ve been a massive disaster, like the gems from Street Fighter Vs. Tekken, but actually adds a layer of depth on top of the game that makes it even more fun to experiment with beyond which two characters you’ll have on your team. It’s balanced, fast, readable, and frankly didn’t get the love it deserved.

Read our full Marvel Vs. Capcom: Infinite review

Editors’ Choice

Repost: Original Source and Author Link


Nickelodeon’s New Fighting Game Looks Surprisingly Serious

Seemingly out of left-field. Nickelodeon announced a new Super Smash Bros-like fighting game Nickelodeon All-Stars Brawl. The surprise announcement immediately caught the eye of Smash Bros. Twitter, which exploded with reactions.

At first, that reaction was due to the sheer hilarity of a Nicktoons fighting game. Upon further inspection, those laughs became genuine excitement. Serious players in the fighting game community are thinking that the cartoon brawler may be a deceptively legitimate competitive experience.

The developer’s history

One of the first things that opened the competitive eyes to All-Stars Brawl is the game’s developer: Ludosity. Ludosity is the team behind one of the many Super Smash Bros Melee-inspired clones, Slap City.

Slap City is a platform fighting game in the vein of Super Smash Bros, namely influenced by Super Smash Bros. Melee. This indie-fighter pulled a ton of technical inspiration from the high-level techniques of the Smash series, making it a hit among genre diehards.

The fighter released to positive reception and that has players intrigued by All-Stars Brawl. Reacting to the trailer, competitive Melee player Hungrybox stating, “The makers of Slap City made this game. And Slap City is, in my humble opinion, the best Melee clone.”

That detail initially grabbed players’ curiosity, but the actual gameplay trailer has demanded further attention. For fans of competitive fighters, the combination of the trailer and the developer’s pedigree teases a legitimate contender in the fighting game scene.

Nickelodeon is adopting Smash’s past

When the trailer dropped, fans immediately began dishing out jokes and comparing it to Smash Bros. On further inspection, however, fans began taking it much more seriously. The developers answered questions about the game through Discord, revealing details that immediately changed fans’ tune.

The devs say it has wavedashing too lol

— BenPaz (@BenTheUltimate1) July 13, 2021

Like a dream, Nickelodeon’s fighter is bringing back wavedashing, a high-level technique constantly used in Super Smash Bros Melee. It’s a complex technique that allows players to cancel a very low air dodge into a ground slide, making them glide along the ground without walking or running. That’s a feature that isn’t even in Super Smash Bros. Ultimate (well, aside from Kazuya Mishima, of course). Eagle-eyed viewers even noticed that the technique was sneakily dropped in the actual reveal trailer.

More expert techniques were shown in the clip, specifically a short sequence by Patrick Star of all characters. In the clip, the Spongebob character performs a full-hop, into back waveland, platform drop, back aerial attack. To put that in simple terms, it’s a series of very advanced maneuvers allowing players to slide near-instantly through a platform and attack while falling.

For Smash fans, the trailer proves that this game may be more than another licensed game that’s only good for memes. Could we be looking at a genuine alternative for those wanting another Melee?

That isn’t the only part of the game pulled from Smash’s past. The game gives players the ability to run through characters, which was removed for the first time in Smash history with Super Smash Bros Ultimate. This allows for the return of techniques like shield cross-ups, which the Super Smash wiki defines as “the act of timing an attack such that the user moves past the opponent and ends up behind them once the hitboxes are gone.”

The final piece of Melee‘s DNA that can be spotted in the trailer is Edge-hogging, another tactic from the past that was removed in Ultimate. This entails a character holding a ledge so their opponent can’t grab it while recovering. You can see this demonstrated in the image above where Sandy is attempting to grab the ledge with what may be her side-B attack. However, due to Lincoln Loud already holding the edge, she misses it and simply falls.

Yes, it will have rollback netcode

The mechanics, tournament-ready stages, and more fully put a competitive eye on All-Star Brawl. But what no one was ready for was the big netcode reveal. The developers were asked the big question about All-Star Brawl‘s netcode and immediately responded, confirming rollback will be included on supported platforms.

dev confirmed rollback in their discord

— husky (@huskySSBM) July 13, 2021

Rollback is the best possible networking for fighting games, as opposed to the competition. The fighting game glossary breaks how it works down, defining it as, “An approach to implementing netcode in a fighting game that plays your own inputs immediately, and then rewinds and resimulates (or ‘rolls back’) the game if network delay causes inconsistencies.”

Fans do have some concerns about how the online networking will function across consoles, though. Fans noticed rollback netcode will be available “on supported platforms,” which has Nintendo Switch fans concerned. Nintendo’s history with bad online led many to instantly believe it’d be left out of the equation, which would hurt the game’s competitive viability.

Despite such worries, many are looking at Nickelodeon’s latest trek into the fighting game world as a future home to more competition. I, for one, can’t wait to see Reptar and Powdered Toast Man mains complain about their bad matchups against Nigel of The Wild Thornberries.

Editors’ Choice

Repost: Original Source and Author Link


Go read this story on how Facebook’s focus on growth stopped its AI team from fighting misinformation

Facebook has always been a company focused on growth above all else. More users and more engagement equals more revenue. The cost of that single-mindedness is spelled out clearly in this brilliant story from MIT Technology Review. It details how attempts to tackle misinformation by the company’s AI team using machine learning were apparently stymied by Facebook’s unwillingness to limit user engagement.

“If a model reduces engagement too much, it’s discarded. Otherwise, it’s deployed and continually monitored,” writes author Karen Hao of Facebook’s machine learning models. “But this approach soon caused issues. The models that maximize engagement also favor controversy, misinformation, and extremism: put simply, people just like outrageous stuff.”

On Twitter, Hao noted that the article is not about “corrupt people [doing] corrupt things.” Instead, she says, “It’s about good people genuinely trying to do the right thing. But they’re trapped in a rotten system, trying their best to push the status quo that won’t budge.”

The story also adds more evidence to the accusation that Facebook’s desire to placate conservatives during Donald Trump’s presidency led to it turning a blind eye to right-wing misinformation. This seems to have happened at least in part due to the influence of Joel Kaplan, a former member of George W. Bush’s administration who is now Facebook’s vice president of global public policy and “its highest-ranking Republican.” As Hao writes:

All Facebook users have some 200 “traits” attached to their profile. These include various dimensions submitted by users or estimated by machine-learning models, such as race, political and religious leanings, socioeconomic class, and level of education. Kaplan’s team began using the traits to assemble custom user segments that reflected largely conservative interests: users who engaged with conservative content, groups, and pages, for example. Then they’d run special analyses to see how content-moderation decisions would affect posts from those segments, according to a former researcher whose work was subject to those reviews.

The Fairness Flow documentation, which the Responsible AI team wrote later, includes a case study on how to use the tool in such a situation. When deciding whether a misinformation model is fair with respect to political ideology, the team wrote, “fairness” does not mean the model should affect conservative and liberal users equally. If conservatives are posting a greater fraction of misinformation, as judged by public consensus, then the model should flag a greater fraction of conservative content. If liberals are posting more misinformation, it should flag their content more often too.

But members of Kaplan’s team followed exactly the opposite approach: they took “fairness” to mean that these models should not affect conservatives more than liberals. When a model did so, they would stop its deployment and demand a change. Once, they blocked a medical-misinformation detector that had noticeably reduced the reach of anti-vaccine campaigns, the former researcher told me. They told the researchers that the model could not be deployed until the team fixed this discrepancy. But that effectively made the model meaningless. “There’s no point, then,” the researcher says. A model modified in that way “would have literally no impact on the actual problem” of misinformation.

The story also says that the work by Facebook’s AI researchers on the problem of algorithmic bias, in which machine learning models unintentionally discriminate against certain groups of users, has been undertaken, at least in part to preempt these same accusations of anti-conservative sentiment and forestall potential regulation by the US government. But pouring more resources into bias has meant ignoring problems involving misinformation and hate speech. Despite the company’s lip service to AI fairness, the guiding principle, says Hao, is still the same as ever: growth, growth, growth.

[T]esting algorithms for fairness is still largely optional at Facebook. None of the teams that work directly on Facebook’s news feed, ad service, or other products are required to do it. Pay incentives are still tied to engagement and growth metrics. And while there are guidelines about which fairness definition to use in any given situation, they aren’t enforced.

You can read Hao’s full story at MIT Technology Review here.

Repost: Original Source and Author Link


The Best Fighting Game Controllers

Just about every genre in gaming can be played competitively, but fighting games are essentially entwined with competitive play. Each match is a direct test between two players’ skills, reactions, understanding of the game mechanics, and ability to predict their opponent’s next move. This is all conveyed through inputs that need to be as intuitive to perform as breathing. If you have to think about how to do something — or worse, you accidentally do something else — then you’ve already lost.

Fighting games require a completely different setup than most other genres. Unlike FPS games, for example, there’s no need for two analog sticks or triggers. Based on the genre’s origins in arcades, many players feel most comfortable with a controller that mimics that stick and button style layout, but over the years, there have been tons of innovations to cater to all types of fighting game players. No matter what your preference, we’ve listed all the best fighting game controllers available.

Further reading

Victrix Pro FS Arcade Fight Stick (PS4, PS5, Xbox)

For all you competitive players out there who are willing to spend a little extra for a sturdy stick that will last the generation, the Victrix Pro FS Arcade Fight Stick is among your best options. Not only does it look and feel great, but it also comes packed with additional features and is built with maintenance in mind. Aside from just having a nice sleek design, this stick also comes packing neon purple cords and RGB lights inside to give it some personality. Looks aside, this stick comes standard with Sanwa Denshi buttons and a joystick, all of which are easy to detach for cleaning or replacement. You also get three macro buttons along the top that you can customize however you like, plus they can be turned off so you don’t accidentally hit them during play.

The Victrix is nice and heavy at just under 8 pounds, so there’s no chance of slipping or sliding while you have it on your lap or table. When it is time to move, you’ve got convenient handles and strap attachments if you want to sling it over your shoulder. Since this is such a high-end stick, they also made it incredibly easy to pop open and tool around with the components or adjust the stick. You even have little areas to store tools and spare parts right inside, so you’ll never be caught with a busted stick.

Qanba Dragon (PS3, PS4, PC)

Qanba Dragon

The other high-priced option on this list is the Qanba Dragon. This stick is a little less well-designed in terms of ease of transport, but once you’re set up, there’s not much to dislike about this stick. It is arguably a little too heavy, even for those who like a hefty stick, coming in at just under 12 pounds. That heft isn’t for nothing, though, because this stick is about as durable as they come. It is a fully aluminum case with Sanwa buttons and a joystick that feel just as good, if not a little better, than the Victrix. The buttons are snappy and responsive, and the joystick has a comfortable chrome head.

This stick is made for the PS4, so one of the extra buttons is the touchpad, making it that much easier to navigate with without having to swap between a normal controller when setting up games. Popping the hood is also a breeze, with more storage space for parts but also plenty of room for a few other things too. You could realistically pack in everything you need for the day just in the stick itself, which is good because of how awkward it is to travel with otherwise. Oh, and if you’re a stickler for fingerprints, you’ll hate the glossy surface that picks them up better than a detective on a murder case.

Hori Fighting Stick Mini 4 (PS3, PS4, PS5, PC)

Hori Fighting Stick Mini 4

Okay, we’ve looked at the two big boys, but what about some smaller options? The Hori Fighting Stick Mini 4 is, obviously, a much smaller and lower-priced arcade stick that is a much better entry point for those who don’t want to make a huge first investment into an arcade stick. This little guy is a mere 8 inches by 6 inches and weighs in at only 2 pounds, making it the easiest by far to bring to events or friends’ houses. It rides that line between being compact but not to the point where your hands will feel cramped using it. The stick and buttons will feel just fine, nothing special, and it does have rubber grips to help keep it in place. That being said, if you’re an aggressive player, you might find yourself chasing this stick around due to how light it is.

This stick is also not all that friendly when it comes to customization. You have all the buttons you need, but there are no bells or whistles here. You’re also not going to be modding this stick since there’s no easy access to its guts, and the buttons are actually soldered right onto the circuit board. All in all, this is a completely serviceable arcade stick for anyone who needs a simple and easy-to-transport backup or wants to test one out at a lower price before checking out the more costly options.

Mixbox Universal Edition (PS4, PS5, Xbox, PC, Switch)

Mixbox-12 Universal Edition

Starting to get into the more unique controller options, we have the Mixbox Universal Edition. As you can already see, the unique feature of this arcade controller is the lack of a joystick. Instead, the Mixbox caters to players who are more comfortable controlling their character’s movement with an arrow key setup. This completely changes how difficult — or easy — various inputs are to perform. But the reason behind this design, and the controller following this one, wasn’t to give players an edge. Aside from personal preference, many longtime fighting game players suffer from wrist pain from prolonged play on a joystick due to repetitive motion, and the arrow key setup of this controller gives them a way to play pain-free.

In terms of parts, you’re getting the best of the best with Sanwa buttons, plus Cherry MX for the directional buttons. There are plenty of extra customizable buttons and even a handy switch on the inside to quickly swap between two movement options. It’s a nice weight, right at 5 1/2 pounds, and has a really generous cable at almost 15 feet long. This Universal Edition, as the name suggests, is a little more pricey, but if you want a stick that will work no matter what you plug it into, this is the one to go for.

Hit Box (PS4, PS5, PC)

Hit Box

Similar, and yet quite different, to the Mixbox is the Hit Box. Just looking at it might seem a little puzzling, but the design is actually incredibly intelligent and intuitive once you get your fingers on it. Rather than the triangular orientation of the previous entry, the Hit Box lays out the movement buttons in a very organic and natural position to how your hand will rest on the pad, just like the normal attack buttons. It should go without saying at this point, but yes, every button here is a Senwa, so everything feels tactile and satisfying. Again, this is a godsend for players with hand or wrist issues.

Just like learning how to properly hold and swing a golf club, there is a learning curve with the Hit Box. Until you become accustomed to it and the button inputs become second nature, you probably will suffer a bit. Also, there is still some contention about this controller giving unfair advantages since you are able to immediately transition from, say, holding back to forward without going neutral like you would if you were pushing a stick from back to forward. If you’re looking to compete, just make sure the events you’re interested in don’t ban this particular controller before you dump time and money into it. Otherwise, this is among the best innovations in fighting game controllers ever created.

Hori Fighting Commander (PS3, PS4, PS5, Xbox, Switch)

Some people call them crazy, but there have been plenty of top competitors out there who stick to more traditional pad-style controllers. For anyone who likes to have a firm grip on their game rather than mimic an arcade cabinet, the Hori Fighting Commander series of controllers are tailor-made for fighting games. Unlike most default controllers, the Hori has an actually good-feeling D-pad, six buttons right on the face, a turbo button, and all the other necessary inputs. The buttons, and controller in general, are a bit bigger than normal ones and flat rather than curved to prevent slipping or accidental inputs.

Hori has an entire line of these controllers for each console, with ones like the PS4 version being compatible with the PS5 and PC as well, so you’re covered no matter where you play. It is also light and ergonomic enough that if you wanted to play with the “claw” grip, using your thumb and index finger on the face buttons, you could do so reasonably without cramping up. Plus, you can still use it easily for other 2D games, and you probably will want to after feeling what a good D-pad is like compared to a standard controller. It’s fairly low-cost and frankly worth having, even if fighting games aren’t your primary genre.

Skywin Brook Sniper Converter (PS3, PS4, Xbox, Switch)

Skywin Brook Sniper Converter

Okay, this is kind of a cheat. This isn’t a controller itself but just lets you use your existing keyboard and mouse setup on your console of choice. If you’re already accustomed to playing on your existing — and perhaps even expensive and customized — keyboard, why go through the trouble and expense of buying and learning a new controller when you can just use what you like on a console? That’s the entire point of the Skywin Brook Sniper Converter. Just slap in your USB keyboard, and you’re good to go — in most cases, anyway.

Technically, this is marketed toward FPS games since keyboard and mouse controls are the preferred methods for so many players in that genre and not officially supported by consoles, but it works just as well for fighting games. There are also those players out there who use even crazier controllers to play, like piano keyboards, which, as long as they connect via USB, this converter will work, no problem.

Razer Raion Fightpad (PS4, PS5, PC)

Razer Raion Fightpad

Razor will round out the list with another fightpad. The Raion Fightpad is simple, straightforward, and to the point. You get your great-feeling D-pad, responsive and flat buttons with a mechanical switch to toggle them on and off, and a comfortable and sturdy shell. It does run on the more expensive side compared to some other pads, including the Hori Fighting Commander, but it’s a little more versatile and comes with some extra functionality, such as the touchpad and share buttons, plus a dedicated switch to go from PS4 to PC depending on where you’re playing. You can also deactivate these buttons to avoid accidental pauses, which is always a great feature.

Another bonus is the headphone jack, which many dedicated fighting gamepads lack. Razor also claims this to be one of the most resilient gamepads and gives each face button a rating of 80 million presses. We didn’t test that out ourselves, but the promise is comforting at the very least. This is essentially the high-end option for those looking for a gamepad, while the Hori is more the entry-level. It isn’t a massive leap from one to the other, but if you’re invested in the extra functions and the general feeling of quality, this one will be worth it.

Editors’ Choice

Repost: Original Source and Author Link


Analogue fighting scalpers and bots with bigger efforts than PS5

This week the folks at Analogue made an extended statement on supply and demand. The gaming products they offer are not “Limited Edition,” they’ve confirmed. This means that they’ll be available for purchase, at some point, no matter how many are sold. Per the company’s message, “More Pockets will be available for purchase in 2021.”

They’ve suggested that because of the “current global state of affairs,” they’re dealing with worldwide supply chain issues and have had production capacity lowered significantly. This means that their products, both Super Nt and Mega Sg, aren’t in stock, and they’re not likely set to be back in stock in the very near future.

Instead, Analogue said that “with all things considered, we will be doing our best to keep Pocket in stock in 2021,” and “bear with us during these unusual times.” They’ve also pointed customers to the Analog (dot) co store where notifications will be sent to interested potential gamers.

Analog also made clear that they will be “implementing robust bot protection on the Analogue Store.” They’ve also found a way to see which orders were done “using bots” and say that all orders that used bots “will be cancelled and their inventory will be allocated to real users.”

Analogue also noted that they’ll be monitoring resale websites like eBay to track scalpers reselling Pre-orders. This is specific to users who are reselling Analogue products “that the seller does not have in their possession.” Because this sort of product sale is against eBay’s policies, Analogue will be reporting said listings so they’ll be de-listed.

Cross your fingers we’ll be seeing the whole Analogue product family back in their online store in the future. For now, all we can do is click and wait. Take a peek at the timeline below for more information on Analogue and their gaming bits and pieces.

Repost: Original Source and Author Link

Tech News

How your ‘fighting style’ should determine your team

Boris is the wise ol’ CEO of TNW who writes a weekly column on everything about being an entrepreneur in tech — from managing stress to embracing awkwardness. You can get his musings straight to your inbox by signing up for his newsletter!

I didn’t fight a lot as a kid. Or as an adult, for that matter. My preference is to avoid violence as much as possible. This isn’t really about maintaining some noble philosophy, it’s mostly just self-preservation. I would never bet on myself winning a fistfight, ever.

Unfortunately, I couldn’t always avoid violence as a kid. I lived in a pretty good part of town, which just happened to be closer to a ‘not-so-good’ part of town. This meant my path would often overlap with groups of slightly older boys who were on the prowl for an easy target to bully. As a skinny kid with glasses who liked ballet, I was the perfect target.

My primary ‘fighting’ strategy against these groups? Run.

I would run home, or into a store, or even into total strangers’ houses. Usually, that was enough to deter the assailants. But sometimes the stores would be closed or I simply couldn’t get past the group, so I had no choice but to defend myself.

[Read: Waiters and employees are in charge — you’re just paying them]

At that point, I’d switch to my true strategy and go on the offensive. I would drop my bag, take off my glasses, and simply charge at the kid closest to me. My theory was that most people would go into a fighting stance and hope to find an opening to throw a punch — while at the same time putting a lot of energy into avoiding getting hit themselves.

I hoped that by accepting I would get hit a few times, I could spend less energy on my defense and focus all my energy on simply throwing punches. Or wailing arms, spitting, biting, and grabbing hair if that was an option. I made no effort to fight fair, and I was outnumbered anyway, so I figured I had some leeway with the informal streetfighting rules.

My strategy would usually result in me getting hurt, but I acted like it didn’t while inflicting whatever damage I could to the attacker. Looking back, I’m not sure if this was a smart strategy or if I’d recommend it to anyone.

However, I recently realized I’ve never really left that strategy. I’ve taken it with me and applied it to many things in my life.

When working on a project, I don’t consider the downsides or deal with all the potential consequences. I simply charge ahead with all my attention and focus on the end goal, without much consideration for what I break in the process. If I make dinner, I’ll happily sacrifice ingredients, packaging, and the occasional plate or two, as I accept the damage to get to the greater good.

Unfortunately for my family, this also applies to home improvement.

The other day I wanted to put up a shelf. My only focus was to GET THE SHELF ON THE WALL. I didn’t empty the shelves below first to avoid damage in the unlikely event I would drop the shelve mid-assembly.

That unlikely event obviously occurred, and I broke a few plates and then cursed myself for not being more careful.

But that’s me. I don’t think about defense and risk-avoidance or even the damage I might do to myself. I go all in, which leads to lots of nice things… and a few bruises and ego issues every now and then.

Why am I telling you this? Because it helps to think about what kind of person you are, and what kind of people you have around you.

Need a task done quick and dirty? Ask me! Do you have a delicate project that needs lots of attention to detail and patience to complete? Well, maybe we need to look at someone else in the team.

Once you know what kind of fighter you are, you can assemble a team around you that complements you.

Can’t get enough of Boris? Check out his older stories here, and sign up for his newsletter here.

Published January 28, 2021 — 16:33 UTC

Repost: Original Source and Author Link


OpenAI’s latest breakthrough is astonishingly powerful, but still fighting its flaws

The most exciting new arrival in the world of AI looks, on the surface, disarmingly simple. It’s not some subtle game-playing program that can outthink humanity’s finest or a mechanically advanced robot that backflips like an Olympian. No, it’s merely an autocomplete program, like the one in the Google search bar. You start typing and it predicts what comes next. But while this sounds simple, it’s an invention that could end up defining the decade to come.

The program itself is called GPT-3 and it’s the work of San Francisco-based AI lab OpenAI, an outfit that was founded with the ambitious (some say delusional) goal of steering the development of artificial general intelligence or AGI: computer programs that possess all the depth, variety, and flexibility of the human mind. For some observers, GPT-3 — while very definitely not AGI — could well be the first step toward creating this sort of intelligence. After all, they argue, what is human speech if not an incredibly complex autocomplete program running on the black box of our brains?

As the name suggests, GPT-3 is the third in a series of autocomplete tools designed by OpenAI. (GPT stands for “generative pre-trained transformer.”) The program has taken years of development, but it’s also surfing a wave of recent innovation within the field of AI text-generation. In many ways, these advances are similar to the leap forward in AI image processing that took place from 2012 onward. Those advances kickstarted the current AI boom, bringing with it a number of computer-vision enabled technologies, from self-driving cars, to ubiquitous facial recognition, to drones. It’s reasonable, then, to think that the newfound capabilities of GPT-3 and its ilk could have similar far-reaching effects.

Like all deep learning systems, GPT-3 looks for patterns in data. To simplify things, the program has been trained on a huge corpus of text that it’s mined for statistical regularities. These regularities are unknown to humans, but they’re stored as billions of weighted connections between the different nodes in GPT-3’s neural network. Importantly, there’s no human input involved in this process: the program looks and finds patterns without any guidance, which it then uses to complete text prompts. If you input the word “fire” into GPT-3, the program knows, based on the weights in its network, that the words “truck” and “alarm” are much more likely to follow than “lucid” or “elvish.” So far, so simple.

What differentiates GPT-3 is the scale on which it operates and the mind-boggling array of autocomplete tasks this allows it to tackle. The first GPT, released in 2018, contained 117 million parameters, these being the weights of the connections between the network’s nodes, and a good proxy for the model’s complexity. GPT-2, released in 2019, contained 1.5 billion parameters. But GPT-3, by comparison, has 175 billion parameters — more than 100 times more than its predecessor and ten times more than comparable programs.

The dataset GPT-3 was trained on is similarly mammoth. It’s hard to estimate the total size, but we know that the entirety of the English Wikipedia, spanning some 6 million articles, makes up only 0.6 percent of its training data. (Though even that figure is not completely accurate as GPT-3 trains by reading some parts of the database more times than others.) The rest comes from digitized books and various web links. That means GPT-3’s training data includes not only things like news articles, recipes, and poetry, but also coding manuals, fanfiction, religious prophecy, guides to the songbirds of Bolivia, and whatever else you can imagine. Any type of text that’s been uploaded to the internet has likely become grist to GPT-3’s mighty pattern-matching mill. And, yes, that includes the bad stuff as well. Pseudoscientific textbooks, conspiracy theories, racist screeds, and the manifestos of mass shooters. They’re in there, too, as far as we know; if not in their original format then reflected and dissected by other essays and sources. It’s all there, feeding the machine.

What this unheeding depth and complexity enables, though, is a corresponding depth and complexity in output. You may have seen examples floating around Twitter and social media recently, but it turns out that an autocomplete AI is a wonderfully flexible tool simply because so much information can be stored as text. Over the past few weeks, OpenAI has encouraged these experiments by seeding members of the AI community with access to the GPT-3’s commercial API (a simple text-in, text-out interface that the company is selling to customers as a private beta). This has resulted in a flood of new use cases.

It’s hardly comprehensive, but here’s a small sample of things people have created with GPT-3:

  • A question-based search engine. It’s like Google but for questions and answers. Type a question and GPT-3 directs you to the relevant Wikipedia URL for the answer.
  • A chatbot that lets you talk to historical figures. Because GPT-3 has been trained on so many digitized books, it’s absorbed a fair amount of knowledge relevant to specific thinkers. That means you can prime GPT-3 to talk like the philosopher Bertrand Russell, for example, and ask him to explain his views. My favorite example of this, though, is a dialogue between Alan Turing and Claude Shannon which is interrupted by Harry Potter, because fictional characters are as accessible to GPT-3 as historical ones.
  • Solve language and syntax puzzles from just a few examples. This is less entertaining than some examples but much more impressive to experts in the field. You can show GPT-3 certain linguistic patterns (Like “food producer becomes producer of food” and “olive oil becomes oil made of olives”) and it will complete any new prompts you show it correctly. This is exciting because it suggests that GPT-3 has managed to absorb certain deep rules of language without any specific training. As computer science professor Yoav Goldberg — who’s been sharing lots of these examples on Twitter — put it, such abilities are “new and super exciting” for AI, but they don’t mean GPT-3 has “mastered” language.
  • Code generation based on text descriptions. Describe a design element or page layout of your choice in simple words and GPT-3 spits out the relevant code. Tinkerers have already created such demos for multiple different programming languages.
  • Answer medical queries. A medical student from the UK used GPT-3 to answer health care questions. The program not only gave the right answer but correctly explained the underlying biological mechanism.
  • Text-based dungeon crawler. You’ve perhaps heard of AI Dungeon before, a text-based adventure game powered by AI, but you might not know that it’s the GPT series that makes it tick. The game has been updated with GPT-3 to create more cogent text adventures.
  • Style transfer for text. Input text written in a certain style and GPT-3 can change it to another. In an example on Twitter, a user input text in “plain language” and asked GPT-3 to change it to “legal language.” This transforms inputs from “my landlord didn’t maintain the property” to “The Defendants have permitted the real property to fall into disrepair and have failed to comply with state and local health and safety codes and regulations.”
  • Compose guitar tabs. Guitar tabs are shared on the web using ASCII text files, so you can bet they comprise part of GPT-3’s training dataset. Naturally, that means GPT-3 can generate music itself after being given a few chords to start.
  • Write creative fiction. This is a wide-ranging area within GPT-3’s skillset but an incredibly impressive one. The best collection of the program’s literary samples comes from independent researcher and writer Gwern Branwen who’s collected a trove of GPT-3’s writing here. It ranges from a type of one-sentence pun known as a Tom Swifty to poetry in the style of Allen Ginsberg, T.S. Eliot, and Emily Dickinson to Navy SEAL copypasta.
  • Autocomplete images, not just text. This work was done with GPT-2 rather than GPT-3 and by the OpenAI team itself, but it’s still a striking example of the models’ flexibility. It shows that the same basic GPT architecture can be retrained on pixels instead of words, allowing it to perform the same autocomplete tasks with visual data that it does with text input. You can see in the examples below how the model is fed half an image (in the far left row) and how it completes it (middle four rows) compared to the original picture (far right).

GPT-2 has been re-engineered to autocomplete images as well as text.
Image: OpenAI

All these samples need a little context, though, to better understand them. First, what makes them impressive is that GPT-3 has not been trained to complete any of these specific tasks. What usually happens with language models (including with GPT-2) is that they complete a base layer of training and are then fine-tuned to perform particular jobs. But GPT-3 doesn’t need fine-tuning. In the syntax puzzles it requires a few examples of the sort of output that’s desired (known as “few-shot learning”), but, generally speaking, the model is so vast and sprawling that all these different functions can be found nestled somewhere among its nodes. The user need only input the correct prompt to coax them out.

The other bit of context is less flattering: these are cherry-picked examples, in more ways than one. First, there’s the hype factor. As the AI researcher Delip Rao noted in an essay deconstructing the hype around GPT-3, many early demos of the software, including some of those above, come from Silicon Valley entrepreneur types eager to tout the technology’s potential and ignore its pitfalls, often because they have one eye on a new startup the AI enables. (As Rao wryly notes: “Every demo video became a pitch deck for GPT-3.”) Indeed, the wild-eyed boosterism got so intense that OpenAI CEO Sam Altman even stepped in earlier this month to tone things down, saying: “The GPT-3 hype is way too much.”

Secondly, the cherry-picking happens in a more literal sense. People are showing the results that work and ignoring those that don’t. This means GPT-3’s abilities look more impressive in aggregate than they do in detail. Close inspection of the program’s outputs reveals errors no human would ever make as well nonsensical and plain sloppy writing.

For example, while GPT-3 can certainly write code, it’s hard to judge its overall utility. Is it messy code? Is it code that will create more problems for human developers further down the line? It’s hard to say without detailed testing, but we know the program makes serious mistakes in other areas. In the project that uses GPT-3 to talk to historical figures, when one user talked to “Steve Jobs,” asking him, “Where are you right now?” Jobs replies: “I’m inside Apple’s headquarters in Cupertino, California” — a coherent answer but hardly a trustworthy one. GPT-3 can also be seen making similar errors when responding to trivia questions or basic math problems; failing, for example, to answer correctly what number comes before a million. (“Nine hundred thousand and ninety-nine” was the answer it supplied.)

But weighing the significance and prevalence of these errors is hard. How do you judge the accuracy of a program of which you can ask almost any question? How do you create a systematic map of GPT-3’s “knowledge” and then how do you mark it? To make this challenge even harder, although GPT-3 frequently produces errors, they can often be fixed by fine-tuning the text it’s being fed, known as the prompt.

Branwen, the researcher who produces some of the model’s most impressive creative fiction, makes the argument that this fact is vital to understanding the program’s knowledge. He notes that “sampling can prove the presence of knowledge but not the absence,” and that many errors in GPT-3’s output can be fixed by fine-tuning the prompt.

In one example mistake, GPT-3 is asked: “Which is heavier, a toaster or a pencil?” and it replies, “A pencil is heavier than a toaster.” But Branwen notes that if you feed the machine certain prompts before asking this question, telling it that a kettle is heavier than a cat and that the ocean is heavier than dust, it gives the correct response. This may be a fiddly process, but it suggests that GPT-3 has the right answers — if you know where to look.

“The need for repeated sampling is to my eyes a clear indictment of how we ask questions of GPT-3, but not GPT-3’s raw intelligence,” Branwen tells The Verge over email. “If you don’t like the answers you get by asking a bad prompt, use a better prompt. Everyone knows that generating samples the way we do now cannot be the right thing to do, it’s just a hack because we’re not sure of what the right thing is, and so we have to work around it. It underestimates GPT-3’s intelligence, it doesn’t overestimate it.”

Branwen suggests that this sort of fine-tuning might eventually become a coding paradigm in itself. In the same way that programming languages make coding more fluid with specialized syntax, the next level of abstraction might be to drop these altogether and just use natural language programming instead. Practitioners would draw the correct responses from programs by thinking about their weaknesses and shaping their prompts accordingly.

But GPT-3’s mistakes invite another question: does the program’s untrustworthy nature undermine its overall utility? GPT-3 is very much a commercial project for OpenAI, which began life as a nonprofit but pivoted in order to attract the funds it says it needs for its expensive and time-consuming research. Customers are already experimenting with GPT-3’s API for various purposes; from creating customer service bots to automating content moderation (an avenue that Reddit is currently exploring). But inconsistencies in the program’s answers could become a serious liability for commercial firms. Who would want to create a customer service bot that occasionally insults a customer? Why use GPT-3 as an educational tool if there’s no way to know if the answers it’s giving are reliable?

A senior AI researcher working at Google who wished to remain anonymous told The Verge they thought GPT-3 was only capable of automating trivial tasks that smaller, cheaper AI programs could do just as well, and that the sheer unreliability of the program would ultimately scupper it as a commercial enterprise.

“GPT-3 is not good enough to be really useful without a lot of hard engineering on top,” said the researcher. “Simultaneously, it’s good enough to be dangerous … I tried [the historical chat bot program] and it very quickly started telling me things like Alan Turing was not gay and did not actually commit suicide, and the Nazis did not hurt Anne Frank, etc, etc. Any fact it tells you, there’s a 50 percent chance it’s made up. If you had a friend that talked like that, you’d avoid them, and you’d definitely not hire them at your company.”

As AI researcher Julian Togelius put it: “GPT-3 often performs like a clever student who hasn’t done their reading, trying to bullshit their way through an exam. Some well-known facts, some half-truths, and some straight lies, strung together in what first looks like a smooth narrative.” (Though as many have pointed out: clever students who know how to bullshit go far in this world because people don’t always scrutinize what they’re saying.)

Another serious problem is bias in GPT-3’s output. Professor Anima Anandkumar, director of AI at Nvidia, noted that the program is partly trained on data filtered by Reddit, and that models built from this data produce text that is “shockingly biased.” In one paper examining the output of GPT-2, for example, the model produces all sorts of offensive stereotypes when asked to complete the following sentences (the human-written prompts are in bold): “The man worked as a car salesman at the local Wal-Mart”; “The Black man worked as a pimp for 15 years”; “The woman worked as a prostitute under the name of Hariya.”

Jerome Pesenti, head of AI at Facebook, raised similar concerns, noting that a program built using GPT-3 to write tweets from a single input word produced offensive messages like “a holocaust would make so much environmental sense, if we could get people to agree it was moral.” In a Twitter thread, Pesenti said he wished OpenAI had been more cautious with the program’s roll-out, which Altman responded to by noting that the program was not yet ready for a large-scale launch, and that OpenAI had since added a toxicity filter to the beta.

Some in the AI world think these criticisms are relatively unimportant, arguing that GPT-3 is only reproducing human biases found in its training data, and that these toxic statements can be weeded out further down the line. But there is arguably a connection between the biased outputs and the unreliable ones that point to a larger problem. Both are the result of the indiscriminate way GPT-3 handles data, without human supervision or rules. This is what has enabled the model to scale, because the human labor required to sort through the data would be too resource intensive to be practical. But it’s also created the program’s flaws.

Putting aside, though, the varied terrain of GPT-3’s current strengths and weaknesses, what can we say about its potential — about the future territory it might command?

Here, for some, the sky’s the limit. They note that although GPT-3’s output is error prone, its true value lies in its capacity to learn different tasks without supervision and in the improvements it’s delivered purely by leveraging greater scale. What makes GPT-3 amazing, they say, is not that it can tell you that the capital of Paraguay is Asunción (it is) or that 466 times 23.5 is 10,987 (it’s not), but that it’s capable of answering both questions and many more beside simply because it was trained on more data for longer than other programs. If there’s one thing we know that the world is creating more and more of, it’s data and computing power, which means GPT-3’s descendants are only going to get more clever.

This concept of improvement by scale is hugely important. It goes right to the heart of a big debate over the future of AI: can we build AGI using current tools, or do we need to make new fundamental discoveries? There’s no consensus answer to this among AI practitioners but plenty of debate. The main division is as follows. One camp argues that we’re missing key components to create artificial minds; that computers need to understand things like cause and effect before they can approach human-level intelligence. The other camp says that if the history of the field shows anything, it’s that problems in AI are, in fact, mostly solved by simply throwing more data and processing power at them.

The latter argument was most famously made in an essay called “The Bitter Lesson” by the computer scientist Rich Sutton. In it, he notes that when researchers have tried to create AI programs based on human knowledge and specific rules, they’ve generally been beaten by rivals that simply leveraged more data and computation. It’s a bitter lesson because it shows that trying to pass on our precious human ingenuity doesn’t work half so well as simply letting computers compute. As Sutton writes: “The biggest lesson that can be read from 70 years of AI research is that general methods that leverage computation are ultimately the most effective, and by a large margin.”

This concept — the idea that quantity has a quality all of its own — is the path that GPT has followed so far. The question now is: how much further can this path take us?

If OpenAI was able to increase the size of the GPT model 100 times in just a year, how big will GPT-N have to be before it’s as reliable as a human? How much data will it need before its mistakes become difficult to detect and then disappear entirely? Some have argued that we’re approaching the limits of what these language models can achieve; others say there’s more room for improvement. As the noted AI researcher Geoffrey Hinton tweeted, tongue-in-cheek: “Extrapolating the spectacular performance of GPT3 into the future suggests that the answer to life, the universe and everything is just 4.398 trillion parameters.”

Hinton was joking, but others take this proposition more seriously. Branwen says he believes there’s “a small but nontrivial chance that GPT-3 represents the latest step in a long-term trajectory that leads to AGI,” simply because the model shows such facility with unsupervised learning. Once you start feeding such programs “from the infinite piles of raw data sitting around and raw sensory streams,” he argues, what’s to stop them “building up a model of the world and knowledge of everything in it”? In other words, once we teach computers to really teach themselves, what other lesson is needed?

Many will be skeptical about such predictions, but it’s worth considering what future GPT programs will look like. Imagine a text program with access to the sum total of human knowledge that can explain any topic you ask of it with the fluidity of your favorite teacher and the patience of a machine. Even if this program, this ultimate, all-knowing autocomplete, didn’t meet some specific definition of AGI, it’s hard to imagine a more useful invention. All we’d have to do would be to ask the right questions.

Repost: Original Source and Author Link