Enlarge / Access to first-party games on launch day remains a major selling point for the Xbox Game Pass Ultimate tier. (credit: Microsoft)

The FTC says the across-the-board price increases that Microsoft recently announced for its Xbox Game Pass subscription service tiers represent "exactly the sort of consumer harm from the merger the FTC has alleged" when it sought to block Microsoft's merger with Activision. In a letter to the court posted as part of an ongoing appeal by the FTC in the case, the federal regulator alleges Microsoft's moves are a clear example of "product degradation" brought about by "a firm exercising market power post-merger."

The letter's primary focus is on the soon-to-be-discontinued $10.99/month Console Game Pass tier. That's being replaced with a $14.99/month Game Pass Standard tier (a 36 percent price increase) that no longer includes "day one" access to all of Microsoft's first-party titles. To maintain that key benefit, "Console" subscribers will have to spend 81 percent more for the $19.99 Game Pass Ultimate tier, which also includes a number of additional benefits over the current $10.99/month option.

Is this “based on the acquisition”?

The FTC notes that these changes "coincide with adding Call of Duty to Game Pass’s most expensive tier." Previously, Microsoft publicly promised that this Game Pass access to Activision's ultra-popular shooter would come "with no price increase for the service based on the acquisition."

Enlarge / Get ready to repeat this ~25-second slice of Mario over and over... and over... and over.

If you've ever seen a record-breaking video game speedrun or watched a Games Done Quick marathon, you may have entertained fantasies that you, too, could put up some decent times on your favorite old games. Sure, it would probably take a bit of practice, but what these speedrunners are doing doesn't look that difficult, does it? How hard can it be to press a few buttons with good timing for a few minutes?

After spending a few weeks with Nintendo World Championships: NES Edition, I no longer think that way. The game's bite-size chunks of classic Nintendo games highlight the level of precision needed for even a few minutes of speedrunning perfection, not to mention the tedium of practicing the same in-game motions dozens of times to build up the needed muscle memory. In the process, I gained a newfound respect for the skill displayed by the best speedrunners and found a fresh way to experience some classic NES games that I felt I knew backward and forward.

Gotta go fast

While Nintendo World Championships draws its name from a series of competitions dating back to 1990, it draws its inspiration much more directly from the more recent rise of the online speedrunning community. Thus, the game's main single-player mode is named "Speedrun," tasking players with putting up the fastest times in 150 mini-challenges spread across 13 different Nintendo-developed NES titles.

Enlarge / Artist's conception of Valve's micro-employees hard at work inside your Steam installation. (credit: Getty Images)

As a private and generally secretive company, Valve doesn't offer much outside visibility into its inner workings. So when years' worth of data on the company's employee and aggregate payroll numbers leaked recently, we were eager to take a deep dive to see what those numbers could tell us about the operation and evolution of a company that has a hand in the majority of PC gaming transactions.

The recent data comes from a poorly redacted document in Wolfire's antitrust lawsuit against Steam, as first noticed over the weekend by SteamDB's Pavel Djundik. While the key data in the document has now been properly hidden in the court docket, The Verge captured the raw numbers from a table labeled "Employee Headcount and Gross Pay Data, 2003-2021."

Breaking down that data by year and department with some simple graphs and statistics, seen below, gives us outsiders a rare partial glimpse into Valve's organization. All told, it's a bit hard to believe that this lynchpin of the PC gaming world has rested on the work of just a few hundred people for many years now.

Enlarge / Space Cadet Pinball has never been so portable... or so tiny! (credit: Turing Software / App Store)

One month after PC emulator UTM was rejected from the iOS App Store, the developers have announced that a new "UTM SE" version is now available for free on the App Store. But the app's performance is severely hampered by Apple's restrictions on so-called "just-in-time" (JIT) compilation, limiting the app's suitability for effectively emulating many PC games.

Built on the generic command-line emulation layer QEMU, the open-source UTM boasts support for "30+ processors," from x86 and PowerPC to RISC-V and ARM64. The App Store listing promises you can "run classic software and old-school games" through both a VGA graphics mode and text-based terminal.

Don't expect a seamless, RetroArch-style path to playing Space Cadet Pinball on your iPhone, though. The UTM developers link to pre-configured settings downloads for versions of Windows going back to XP, alongside guides for getting those OSes up and running on iOS. But users will need to bring their own legitimate Windows installation ISO and go through the cumbersome process of installing the OS as well as a version of SPICE tools to help coordinate access through iOS (downloading pre-built, UTM-compatible Linux builds seems more straightforward).

Enlarge / Man vs. machine in a sea of stones. (credit: Getty Images)

In the ancient Chinese game of Go, state-of-the-art artificial intelligence has generally been able to defeat the best human players since at least 2016. But in the last few years, researchers have discovered flaws in these top-level AI Go algorithms that give humans a fighting chance. By using unorthodox "cyclic" strategies—ones that even a beginning human player could detect and defeat—a crafty human can often exploit gaps in a top-level AI's strategy and fool the algorithm into a loss.

Researchers at MIT and FAR AI wanted to see if they could improve this "worst case" performance in otherwise "superhuman" AI Go algorithms, testing a trio of methods to harden the top-level KataGo algorithm's defenses against adversarial attacks. The results show that creating truly robust, unexploitable AIs may be difficult, even in areas as tightly controlled as board games.

Three failed strategies

In the pre-print paper "Can Go AIs be adversarially robust?", the researchers aim to create a Go AI that is truly "robust" against any and all attacks. That means an algorithm that can't be fooled into "game-losing blunders that a human would not commit" but also one that would require any competing AI algorithm to spend significant computing resources to defeat it. Ideally, a robust algorithm should also be able to overcome potential exploits by using additional computing resources when confronted with unfamiliar situations.

Enlarge / Whatever you do, don't ask the AI how many horizontal lines are in this image. (credit: Getty Images)

In the last couple of years, we've seen amazing advancements in AI systems when it comes to recognizing and analyzing the contents of complicated images. But a new paper highlights how many state-of-the-art "vision learning Models" (VLMs) often fail at simple, low-level visual analysis tasks that are trivially easy for a human.

In the provocatively titled pre-print paper "Vision language models are blind" (which has a PDF version that includes a dark sunglasses emoji in the title), researchers from Auburn University and the University of Alberta create eight simple visual acuity tests with objectively correct answers. These range from identifying how often two colored lines intersect to identifying which letter in a long word has been circled to counting how many nested shapes exist in an image (representative examples and results can be viewed on the research team's webpage).

• 

  If you can solve these kinds of puzzles, you may have better visual reasoning than state-of-the-art AIs. [credit: Rahmanzadehgervi, Bolton, Taesiri, Nguyen. ]

Crucially, these tests are generated by custom code and don't rely on pre-existing images or tests that could be found on the public Internet, thereby "minimiz[ing] the chance that VLMs can solve by memorization," according to the researchers. The tests also "require minimal to zero world knowledge" beyond basic 2D shapes, making it difficult for the answer to be inferred from "textual question and choices alone" (which has been identified as an issue for some other visual AI benchmarks).

Enlarge / Artist's conception of Microsoft executives after today's Game Pass pricing announcements. (credit: Getty Images)

For years now, Microsoft's Xbox Game Pass has set itself apart by offering subscribers launch-day access to new first-party titles in addition to a large legacy library of older games. That important "day one" perk is now set to go away for all but the highest tier of Game Pass' console subscribers, even as Microsoft asks for more money for Game Pass across the board.

Let's start with the price increases for existing Game Pass tiers, which are relatively straightforward:

• "Game Pass Ultimate" is going from $16.99 to $19.99 per month.
• "Game Pass for PC" is going from $9.99 to $11.99 per month.
• "Game Pass Core" (previously known as Xbox Live Gold) is going from $59.99 to $74.99 for annual subscriptions (and remains at $9.99 for monthly subscriptions).

Things get a bit more complicated for the $10.99/month "Xbox Game Pass for Console" tier. Microsoft announced that it will no longer accept new subscriptions for that tier after today, though current subscribers will be able to keep it (for now) if they auto-renew their subscriptions.

After years of work, the third-party modders at Team Folon announced over the weekend that their long-awaited (and recently delayed) "DLC-sized" unofficial fan mod, Fallout: London, is now in "QA testing" and awaiting the "final green light" from release partner GOG. But that release will apparently need to make use of a "downgrader" to work around a recent Bethesda "next-gen" update that upended all of the team's modding efforts.

In December, Team Folon announced that its work on the Fallout: London mod would culminate in a planned April 23 release, lining up with England's widely celebrated St. George's Day. Then, just weeks before the planned launch, Bethesda announced that its own planned "next-gen" update for the 2015 game would be coming out on April 25, two days after Fallout: London's launch target.

"That has, for lack of a better term, screwed us over," Team Folon Project Manager Dean Carter said in an interview with the BBC shortly after Bethesda's upgrade announcement. In a separate video, Carter said that the next-gen update "requires many of our internal systems to be updated," since the content they created in the Fallout 4 Script Extender likely wouldn't work under the updated code. That turned out to be a valid concern, as the next-gen update did end up breaking many Fallout 4 mods and saves after its release.

Enlarge / If your right hand starts typing "delve," you may, in fact, be an LLM. (credit: Getty Images)

Thus far, even AI companies have had trouble coming up with tools that can reliably detect when a piece of writing was generated using a large language model. Now, a group of researchers has established a novel method for estimating LLM usage across a large set of scientific writing by measuring which "excess words" started showing up much more frequently during the LLM era (i.e., 2023 and 2024). The results "suggest that at least 10% of 2024 abstracts were processed with LLMs," according to the researchers.

In a pre-print paper posted earlier this month, four researchers from Germany's University of Tubingen and Northwestern University said they were inspired by studies that measured the impact of the COVID-19 pandemic by looking at excess deaths compared to the recent past. By taking a similar look at "excess word usage" after LLM writing tools became widely available in late 2022, the researchers found that "the appearance of LLMs led to an abrupt increase in the frequency of certain style words" that was "unprecedented in both quality and quantity."

Delving in

To measure these vocabulary changes, the researchers analyzed 14 million paper abstracts published on PubMed between 2010 and 2024, tracking the relative frequency of each word as it appeared across each year. They then compared the expected frequency of those words (based on the pre-2023 trendline) to the actual frequency of those words in abstracts from 2023 and 2024, when LLMs were in widespread use.