Connect with us

Gaming

StarCraft II-playing AI AlphaStar takes out pros undefeated – TechCrunch

Published

on

Losing to the computer in StarCraft has been a tradition of mine since the first game came out in 1998. Of course, the built-in “AI” is trivial for serious players to beat, and for years researchers have attempted to replicate human strategy and skill in the latest version of the game. They’ve just made a huge leap with AlphaStar, which recently beat two leading pros 5-0.

The new system was created by DeepMind, and in many ways it’s very unlike what you might call a “traditional” StarCraft AI. The computer opponents you can select in the game are really pretty dumb — they have basic built-in strategies, and know in general how to attack and defend and how to progress down the tech tree. But they lack everything that makes a human player strong: adaptability, improvisation and imagination.

AlphaStar is different. It learned from watching humans play at first, but soon honed its skills by playing against facets of itself.

The first iterations watched replays of games to learn the basics of “micro” (i.e. controlling units effectively) and “macro” (i.e. game economy and long-term goals) strategy. With this knowledge it was able to beat the in-game computer opponents on their hardest setting 95 percent of the time. But as any pro will tell you, that’s child’s play. So the real work started here.

Hundreds of agents were spawned and pitted against each other.

Because StarCraft is such a complex game, it would be silly to think that there’s a single optimal strategy that works in all situations. So the machine learning agent was essentially split into hundreds of versions of itself, each given a slightly different task or strategy. One might attempt to achieve air superiority at all costs; another to focus on teching up; another to try various “cheese” attempts like worker rushes and the like. Some were even given strong agents as targets, caring about nothing else but beating an already successful strategy.

This family of agents fought and fought for hundreds of years of in-game time (undertaken in parallel, of course). Over time the various agents learned (and of course reported back) various stratagems, from simple things such as how to scatter units under an area-of-effect attack to complex multi-pronged offenses. Putting them all together produced the highly robust AlphaStar agent, with some 200 years of gameplay under its belt.

Most StarCraft II pros are well younger than 200, so that’s a bit of an unfair advantage. There’s also the fact that AlphaStar, in its original incarnation anyway, has two other major benefits.

First, it gets its information directly from the game engine, rather than having to observe the game screen — so it knows instantly that a unit is down to 20 HP without having to click on it. Second, it can (though it doesn’t always) perform far more “actions per minute” than a human, because it isn’t limited by fleshy hands and banks of buttons. APM is just one measure among many that determines the outcome of a match, but it can’t hurt to be able to command a guy 20 times in a second rather than two or three.

It’s worth noting here that AIs for micro control have existed for years, having demonstrated their prowess in the original StarCraft. It’s incredibly useful to be able to perfectly cycle out units in a firefight so none takes lethal damage, or to perfectly time movements so no attacker is idle, but the truth is good strategy beats good tactics pretty much every time. A good player can counter the perfect micro of an AI and take that valuable tool out of play.

AlphaStar was matched up against two pro players, MaNa and TLO of the highly competitive Team Liquid. It beat them both handily, and the pros seemed excited rather than depressed by the machine learning system’s skill. Here’s game 2 against MaNa:

In comments after the game series, MaNa said:

I was impressed to see AlphaStar pull off advanced moves and different strategies across almost every game, using a very human style of gameplay I wouldn’t have expected. I’ve realised how much my gameplay relies on forcing mistakes and being able to exploit human reactions, so this has put the game in a whole new light for me. We’re all excited to see what comes next.

And TLO, who actually is a Zerg main but gamely played Protoss for the experiment:

I was surprised by how strong the agent was. AlphaStar takes well-known strategies and turns them on their head. The agent demonstrated strategies I hadn’t thought of before, which means there may still be new ways of playing the game that we haven’t fully explored yet.

You can get the replays of the matches here.

AlphaStar is inarguably a strong player, but there are some important caveats here. First, when they handicapped the agent by making it play like a human, in that it had to move the camera around, could only click on visible units, had a human-like delay on perception and so on, it was far less strong and in fact was beaten by MaNa. But that version, which perhaps may become the benchmark rather than its untethered cousin, is still under development, so for that and other reasons it was never going to be as strong.

AlphaStar only plays Protoss, and the most successful versions of itself used very micro-heavy units.

Most importantly, though, AlphaStar is still an extreme specialist. It only plays Protoss versus Protoss — probably has no idea what a Zerg looks like — with a single opponent, on a single map. As anyone who has played the game can tell you, the map and the races produce all kinds of variations, which massively complicate gameplay and strategy. In essence, AlphaStar is playing only a tiny fraction of the game — though admittedly many players also specialize like this.

That said, the groundwork of designing a self-training agent is the hard part — the actual training is a matter of time and computing power. If it’s 1v1v1 on Bloodbath maybe it’s stalker/zealot time, while if it’s 2v2 on a big map with lots of elevation, out come the air units. (Is it obvious I’m not up on my SC2 strats?)

The project continues and AlphaStar will grow stronger, naturally, but the team at DeepMind thinks that some of the basics of the system, for instance how it efficiently visualizes the rest of the game as a result of every move it makes, could be applied in many other areas where AIs must repeatedly make decisions that affect a complex and long-term series of outcomes.

Source link

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published.

Gaming

Nintendoes what Valve don’t: Game barred from Steam will launch on Switch

Published

on

Enlarge / Nothing weird going on here. No siree.

Japanese publisher Spike Chunsoft announced that the first official English translation of visual novel Chaos;Head Noah won’t be coming to Steam as planned “due to Steam’s guideline-required changes to the game’s content.” But while the game is apparently too risqué for Steam, the family-friendly folks at Nintendo apparently have no problem with a Switch version that Spike Chunsoft says will still launch in the US on October 7 as scheduled.

“Spike Chunsoft, Inc. believes these [Steam guideline-required] changes would not allow the game to be released to its standards,” the publisher said in its announcement. “The company is looking into delivering the title through alternative storefronts, and when details are decided will make another formal announcement. Until then your patience and understanding is appreciated.”

Nintendo says this scene is appropriate for its store page, so we figure you readers can handle it.
Enlarge / Nintendo says this scene is appropriate for its store page, so we figure you readers can handle it.

Chaos;Head Noah was initially listed for Steam pre-sale in April, but that page was taken down in August, according to tracking site SteamDB. At the time, that led to some concerns about the eventual fate of the Steam version, which Spike Chunsoft finally confirmed today.

Valve’s apparent push for content restrictions comes even though the extremely similar thematic sequel Chaos;Child has been available in English on Steam since 2019 (following its initial 2014 release in Japan on the Xbox One). The English PS4 version of Chaos;Child received an M for Mature rating from the ESRB, which described game scenes of strangling, torture, and “exposed brains” alongside sexual content like “two female characters moaning off screen while discussing each other’s breasts.”

How bad is it?

Chaos;Head Noah is an enhanced port of Chaos;Head, the game that launched the cult-classic Science Adventure series of visual novels (which also includes Steins;Gate and its sequels). The game follows a series of murders and suicides in Tokyo’s Shibuya neighborhood and allows players to change the story progression by indulging in various positive or negative “delusions.” Some of those delusions can reportedly get extremely gory and/or suggest (but not directly show) imminent sexual violence.

“I don’t think it gets much worse than anything already in Steam’s library,” PQube Games Head of Localization Andrew Hodgson (who worked on the English translation of Steins;Gate) told Ars Technica of the “titillating and violent content” in the game. “It’s far from adult, even if it can be quite gruesome in certain scenes.”

Just your average, everyday game on a Nintendo console.
Enlarge / Just your average, everyday game on a Nintendo console.

The original Chaos;Head was originally released for Japanese PCs in 2008 before the enhanced Noah hit the Xbox 360 in 2009. That console port (and a later Vita re-release) received CERO Z content ratings in Japan, which “assumes that the game should not be sold or distributed to those younger than 18 years old” and is roughly equivalent to an ESRB “AO for Adults Only” rating in the US. CERO’s “content icon” system for that game only included a warning about “crime,” however, and not violence or sexual content.

Subsequent Japanese ports of Chaos;Head Noah for the PS3, PSP, Android, and iOS were heavily edited to remove some of the more extreme images and descriptions of violence. In turn, those ports received a lower CERO D rating (roughly equivalent to the ESRB’s “M for Mature” rating) in Japan. A source in the visual novel translation community (who asked to remain anonymous) confirmed that both the Switch and proposed Steam English-language versions of the game were based on this edited-down script.

A Japanese Chaos;Head port for the Nintendo Switch, released earlier this year, received the higher CERO Z rating (and “crime” content icon) despite using the edited version of the game that previously received a CERO D rating. The English translation will launch on Switch in the US next month, with an “M for Mature” rating and content descriptors that warn of “Blood and Gore, Sexual Themes, Language, [and] Intense Violence.”

Continue Reading

Gaming

YouTube age-restriction quagmire exposed by 78-minute Mega Man documentary

Published

on

Aurich Lawson / Capcom

A YouTube creator has gone on the offensive after facing an increasingly common problem on the platform: moderation and enforcement that leaves creators confused by the logic and short on their videos’ revenue potential.

The trouble centers on a longtime YouTube video host whose content is popular among the retro-gaming devotees at Ars Technica’s staff. The creator, who goes by the online handle “Summoning Salt,” chronicles the history of various classic games’ speedrunning world records. His hour-plus analyses demonstrate how different players approach older games and exploit various bugs. The games in question are typically cartoony 2D fare instead of violent or M-rated titles.

Summoning Salt asks why his YouTube video was age-restricted.

On Friday, Summoning Salt took to social media to claim that his latest 78-minute documentary about 1989’s Mega Man 2, which went live in mid-September, has been “age-restricted” by YouTube’s moderation system. Bizarrely, the video had been age-restricted roughly one week ago, only for YouTube to relent to the creator’s appeal and claim that the restriction had been placed in error.

Thus, Summoning Salt was surprised to learn on Friday that the video had been re-age-restricted—which he claims severely limits a creator’s ability to monetize content on YouTube. An age restriction flag works against content creators in two ways: it limits the advertisement pool that might run in pre-roll and mid-view breaks, and it essentially slams the door on YouTube’s recommendation algorithm, which might otherwise tease Summoning Salt’s content to new viewers.

Remember, this is Mega Man 2 we’re talking about

Summoning Salt’s (age-restricted) analysis of Mega Man 2 world records.

YouTube’s initial notice did not clarify what moderation flag Summoning Salt’s latest video—a video that documents the 18-year history of people playing and exploiting the NES game Mega Man 2, embedded above—had triggered. His appeal eventually teased an answer from YouTube’s moderation team: “explicit language in certain parts.” As Summoning Salt explained, the video includes a three-second outburst of six F-words, taken directly from a Twitch streamer’s microphone during a passionate gameplay moment.

Summoning Salt, a speedrunning-fluent creator, took his analysis tools to the microsecond level and looked for other unrestricted YouTube content in the gaming category to see whether his video’s curses-per-capita percentage (0.16 percent) had been exceeded. He immediately found an unrestricted example from another popular retro-minded channel, Angry Video Game Nerd, which had nearly double the swears in a video one-twelfth as dense in the script. (It’s unclear how many of AVGN’s videos, famously full of curse words, are flagged with age restrictions.)

Ultimately, Summoning Salt points to YouTube’s unclear recommendations to content creators for content like curse words. According to YouTube’s own rules, the line between “moderate profanity” (allowed in YouTube’s unrestricted videos) and “strong profanity” comes down to not only specific word choice but also frequency, and YouTube merely suggests that the line is crossed when reaching a threshold of “used in every sentence,” or having certain swear words appear in prominent moments like the first 30 seconds of a video or as text in a thumbnail.

Summoning Salt noted that the moderation team initially responded with a “full review” in roughly 40 minutes, less than the length of the whole video. Such a swift review process implied that an auto-moderation system used voice analysis to chronicle the number of swear words, and Summoning Salt told Ars via email that YouTube has tools in place to auto-mute what it detects as offending content—but that YouTube doesn’t apply them in the case of age-restriction disputes. This leaves creators out of the revenue circuit once YouTube raises such a flag. He also told Ars that his videos have only been restricted in the past by YouTube due to copyright flags over included music, which he has zero issue with.

Continue Reading

Gaming

Savor the sinister delights of del Toro’s Cabinet of Curiosities trailer

Published

on

Guillermo del Toro’s Cabinet of Curiosities is a new anthology series coming this month to Netflix.

So-called cabinets of curiosities—aka wunderkammers (“wonder-rooms”)—were hugely popular in the 17th century. They were largely random collections of strange-yet-fascinating stuff, including natural history specimens, archaeological artifacts, religious or historical relics, the odd work of art, and any other quirky item that caught the cabinet creator’s fancy.  The concept also inspired auteur director Guillermo del Toro when putting together a new anthology horror series for Netflix: Guillermo del Toro’s Cabinet of Curiosities. The streaming platform just dropped the official trailer for the series, and it looks like just the right kind of fright fare to bring some stylishly spooky frissons to the Halloween season.

As we’ve reported previously, the series was first announced in 2018 and features eight episodes written and directed by filmmakers handpicked by del Toro. The list of directors includes Jennifer Kent, who directed 2014’s phenomenal The Babadook; her episode, “The Murmuring,” is based on an original story by del Toro and features Babadook star Essie Davis (aka Miss Fisher). “Dreams in the Witch House,” based on an H.P. Lovecraft short story, is directed by Catherine Hardwicke (Lords of Dogtown, Twilight).

“Graveyard Rats” is directed by Vincenzo Natali (In the Tall Grass, Splice), while Guillermo Navarro (Narcos) directed “Lot 36,” also based on an original story by del Toro. Keith Thomas (Firestarter) directed “Pickman’s Model,” another episode based on a Lovecraft short story; David Prior (The Empty Man) directed “The Autopsy”; Panos Cosmatos (Mandy) directed “The Viewing”; and Ana Lily Amirpour—who directed the exquisite A Girl Walks Home Alone at Night—directed “The Outside.”

The star-studded cast includes Rupert Grint, Ben Barnes, Crispin Glover, Peter Weller, Kate Micucci, Nia Vardalos, David Hewlett, Demetrius Grosse, Sebastian Roche, F. Murray Abraham, Hannah Galway, Steve Agee, and Michael Therriault, among others.

Along with several simultaneously gorgeous and horrific images, we got a “first look” teaser in August, featuring del Toro talking about the project, especially its striking visual effects. “With Cabinet of Curiosities, what I’m trying to say is, ‘Look, the world is beautiful and horrible at exactly the same time,'” he said. That’s certainly the vibe we’re getting from the full trailer.

“Picture your mind as a cabinet where you lock up your darkest thoughts and deepest fears,” del Toro says in the opening voiceover. “What would happen if you opened that cabinet for the world to see? We are about to find out.” What follows is a cornucopia of scenes from each of the eight stories, all of which have their own distinctive look that nonetheless fits the overall “beautiful and horrible” aesthetic del Toro is aiming for.

Guillermo del Toro’s Cabinet of Curiosities will be released on Netflix across a four-day event. Two episodes will be released on October 25, with two more episodes coming out each day through October 28.

Netflix

Listing image by Netflix

Continue Reading

Trending