Tech Brew Ride Home - Mon. 07/01 – Now Brussels Sprouts For Meta

Episode Date: July 1, 2024

Meta’s turn to be accused of breaching the DMA. Smarter AirPods as part of an AR/VR strategy. Better AI leaderboards. And is the Surface Laptop finally a true MacBook Air killer? Sponsors: Dragon Ball Legends Links: Meta's pay or consent model in crosshairs for breaching EU tech rules (Reuters) Kuo: Apple to begin mass production of AirPods with cameras by 2026 (9to5Mac) ‘Boring’ Bitcoin Sends Weekend Trading Volume to All-Time Lows (Bloomberg) Amazon’s Bargain Store Would Use Same Trade ‘Loophole’ as Temu, Shein (The Information) Chinese AI models storm Hugging Face's LLM chatbot benchmark leaderboard — Alibaba runs the board as major US competitors have worsened (Tom's Hardware) Surface Laptop review: Microsoft’s best MacBook Air competitor yet (The Verge) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the TechMean right home for Monday, July 1st, 2024. I'm Brian McCullough. Today, meta's turn to be accused of breaching the DMA. Smarter AirPods as part of an ARVR strategy, better AI leaderboards, and is the Surface laptop finally a true MacBook air killer?

Here's what you miss today in the world of tech. And we have another one. In its preliminary findings, the European Union says meta's pay or consent model breaches the DMA. But wait, the whole reason meta offered an ad-free version of its services for the very first time ever was because they thought that that's what the European regulators wanted, right? Quoting Reuters, meta-platforms was charged by EU antitrust regulators on Monday for failing to comply with landmark tech rules as they took aim at the U.S. companies newly introduced pay or consent advertising model, already the target of privacy regulators and activists ire. The tech giant launched the no-ad subscription service for Facebook and Instagram in Europe last November, saying users who consent to be tracked get a free

service, which is funded by advertising revenues, or they could pay for an ad-free service. The European Commission, which acts as the EU competition enforcers, said the binary choice breaches the Block's Digital Markets Act, or DMA, which seeks to rein in the power of big tech, as it sent its preliminary findings to meta. It said the binary choice forces users to consent to the combination of their personal data and fails to provide them a less personalized but equivalent version of meta-social networks. We want to empower citizens to be able to take control over their own data and choose a less personalized ads experience, EU antitrust chief Margath Vestager said in a statement.

Meta said its model complied with a ruling from Europe's top court. Subscription for no ads follows the direction of the highest court in Europe and complies with the DMA. We look forward to further constructive dialogue with the European Commission to bring this investigation to a close, a meta spokesperson said, end quote. Again, I'm going to make a prediction here that the way we're going to get to sort of the first generation of true AR, VR, VR, as the future is one piece at a time. Like we've got the smartphone in our pocket. We've got wearables like watches.

And we're starting to get thin smart glasses coming at us from various angles with various strategies. But don't forget the component of the wearable ecosystem that actually has the biggest install base and thus could actually have the biggest impact on an AR and VR future. The things you have in your ears every day. What if you made those smarter? Well, according to our friend Ming Chi Kuo, Apple plans to mass produce new AirPods with camera modules by 2026, which will include an IR camera similar to the iPhone face ID receiver, quoting 9 to 5 Mac. Apple's goal is reportedly for the camera-equipped AirPods to integrate with

Vision Pro and future Apple Vision headsets. The new AirPods is expected to be used with Vision Pro and future Apple headsets to enhance the user experience of spatial audio and strengthen the spatial computing ecosystem. For example, when a user is watching a video with Vision Pro and wearing this new AirPods. If users turn their heads to look in a specific direction, the sound source in that direction can be emphasized to enhance the spatial audio slash computing experience. Additionally, the report says that the IR camera on AirPods could detect environmental image changes, quote, potentially enabling in-air gesture control to enhance human device integration. In February, Bloomberg's Mark Gurman reported that AirPods with cameras are just one

of the wearable form factor ideas that Apple is investigating. The company is also exploring smart glasses similar to meta raybans, as well as a smart ring, as well as future versions of the Apple Watch, end quote. So this is what I'm getting at. The end goal of the Apple Vision would be, of course, thin glasses, but thin because you can utilize an entire ecosystem of other devices to do the computing you want. I've seen technology where you could also add sensors to earbuds that can do health things, like sense temperature and blood pressure, so you could take that away from the Apple Watch to do other things. I imagine you could put similar technology into smart glasses as well, since they're touching your skin around your brain and ears. I've even seen

tech that can detect user identity when you plug earbuds into your ears or also be used as ambient sensors for areas around the wearer. The point I'm getting at here is that it's not going to be one device. Again, imagine the phone in the pocket as the mothership. It doesn't have to come out as much of your pocket because you have all the various sensing and computing tasks distributed around the various smart wearables on your body. Here's another Bitcoin story that I found interesting. Is there a way we can measure how much those Bitcoin ETFs have actually impacted the Bitcoin trading market and ecosystem? Turns out there is. According to Kiko, The proportion of Bitcoin traded over weekends has fallen to 16% this year so far. That's down from

28% in 2019, a trend likely accelerated by Bitcoin ETF's launch. Quoting Bloomberg, one of crypto's noteworthy traits is that unlike stocks, it trades during all hours of the day and even on Saturdays and Sundays. In the past, Bitcoin trading gained notoriety for its wild weekends, where the digital currency would experience wide price fluctuations. But that phenomenon seems to be cooling since Bitcoin's weekend trading volume has continued to dwindle from its high of 28% in 2019. The launch of Bitcoin ETFs is likely a big reason why. The decline of weekend trading is a, quote, trend that has been going on for years,

but has been exacerbated by ETFs, according to Kiko's senior analyst D'Sislava, Auburt. Unlike most crypto tokens that can be traded any time on exchanges such as, Binance, the Bitcoin ETS followed the schedule of the traditional stock exchange that they are traded on, which means no weekend trading. The proportion of Bitcoin traded on weekdays between 3 p.m. and 4 p.m. increased to 6.7% from 4.5% in the fourth quarter of 2020, Kiko said, that's the period known as the benchmark fixing window when owners of the ETFs determine the price of Bitcoin and then use it to calculate the ETF's net asset value. The institutional adoption of crypto through Bitcoin ETFs has also led to drastically lower price volatility, according to another

report from Kiko. When Bitcoin last reached record highs in November 2021, volatility surged to almost 106 percent. After Bitcoin reached an all-time high of $73,000 in March amid optimism about the ETFs, volatility was just 40%. The trend of lower volatility and the fact that it has remained under 50 percent since the beginning of 2023 indicates that Bitcoin is becoming a more mature asset, according to Kiko, end quote. Follow up to this story from last week. A source has told the information that Amazon does indeed plan to use that controversial trade rule used by Timu and Sheehan, which exempts packages worth $800 or less from tariffs.

It is going to use this for its rumored to be forthcoming discount section, which is going directly at Timu and Sheen. Quote, while the expected move will help Amazon offer prices that are more competitive with its ascendant rivals, it's also likely to thrust Amazon into the center of heated political debate around the trade rule. The rule exempts individual packages worth less than $800 from tariffs and has been critical to the rise of Timu and Sheehan. Using the trade rule would be a switch for Amazon from its traditional practice of encouraging merchants that sell on its site to send items in bulk to its U.S. warehouses, where it could

deliver them quickly to shoppers often in two days or less. Chinese merchants selling on the new bargain section of Amazon's website, however, will store goods and warehouses located in China, and Amazon will then send them directly to shoppers via air freight. That means individual packages to Amazon shoppers won't be subject to tariffs as long as they are valued at $800 or less. If Amazon goes ahead with these plans, it would join other U.S.-based e-commerce sellers that are increasingly finding ways to use cross-border shipments to avoid tariffs and offer lower prices to shoppers. Amazon's deliveries on its new bargain section items will take up to 11 days,

while fulfillment will cost merchants up to 45% less than its existing fulfillment service, according to a presentation given to Amazon sellers. She and Tamu are the highest profile online sellers sending goods to the U.S. duty-free, making up nearly 30% of packages shipped under the exemption known as de minimis. A U.S. Congressional Committee report said in 2023, out of all such, shipments nearly half came from China, according to the report. Several proposals before Congress would clamp down on de minimis, including a bill introduced by Republican Senator Marco Rubio and Democratic Senator Sherrod Brown that would bar its usage for imports from, quote, non-market economies like China's.

A bill from Republican Senator J.D. Vance and Democratic Senator Tammy Baldwin would also exempt Chinese imports from de minimis and lower the tariff-free limit for other countries. Both bills were introduced last year, and it's unclear whether they will pass. before the next Congress, end quote. Every time a new AI model gets released, it's wardens tout how it's fared on this benchmark or that one, but therein lies a problem. There are so many different metrics and no real consensus on which metrics actually matter. It's sort of an apples and oranges thing.

People can cherry pick the metrics they want to give off whatever impression they want, to mix a bunch of fruit metaphors. So to that end, Hugging Face has unveiled the open LLLL LLM leaderboard V2 that test models across six benchmarks and surprise, Chinese models are dominating the top 10 with Alibaba's Quen taking the top spot. Quoting Tom's hardware. The new leaderboard seeks to be a more challenging uniform standard for testing open large language models or LLM performance across a variety of tasks. Alibaba's Quen models appear dominant in the leaderboard's inaugural rankings, taking three spots in the top 10. Hugging Face's second leaderboard

tests, language models across four tasks, knowledge testing, reasoning on extremely long contexts, complex math abilities, and instruction following. Six benchmarks are used to test these qualities with tests including solving thousand-word murder mysteries, explaining Ph.D. level questions in layman's terms, and most daunting of them all, high school math equations. A full breakdown of the benchmarks used can be found on Hugging Faces blog. The frontrunner of the new leaderboard is Quen. Alibaba's LLM, which takes first, third, and tenth place with its handful of variants. Also showing up our Lama 370B, meta's LLM, and a handful of smaller open source projects that manage to outperform the pack. Notably absent is any sign of chat GPT. HuggingFaces' leaderboard

does not test closed source models to ensure reproducibility of results. Tests to qualify on the leaderboard are run exclusively on Hugging Faces' own computers, which, according to CEO Clem DeLong Hughes, Twitter, are powered by 300 Nvidia H-100 GPUs. Because of Hugging Faces open source and collaborative nature, anyone is free to submit new models for testing and admission on the leaderboard with a new voting system, prioritizing popular new entries for testing. The leaderboard can be filtered to show only a highlighted array of significant models to avoid a confusing glut of small LLMs. As a pillar of the LLM space, Hugging Face has become a trusted source for LLM learning and community collaboration. After its first leaderboard was released last year as a means

to compare and reproduce testing results from several established LLMs, the board quickly took off in popularity. Getting high ranks on the board became the goal of many developers, small and large, and as models have become generally stronger, smarter, and optimized for the specific test of the first leaderboard, its results have become less and less meaningful. hence the creation of a second variant. Some LLMs, including newer variants of Meta's Lama, severely underperformed in the new leaderboard compared to their high marks in the first.

This came from a trend of overtraining LLMs only on the first leaderboard's benchmarks, leading to regressing in real-world performance. This regression of performance, thanks to hyper-specific and self-referential data, follows a trend of AI performance growing worse over time, proving once again as Google's AI answer have shown that LLM performance is only as good as its training data and that true artificial

intelligence is still many, many years away, end quote. Finally today, a gadget review. This is a bit belated, but the reviews of the Microsoft Surface Laptop 7th edition came out last week, and I wanted to at least mention them as I think, you know, the Surface laptop is probably one of the more credible competitors to the MacBook Airline. Going with The Verge as ever, and they say great native app performance, but emulated apps can be more battery intensive and AI features still feel gimmicky. Generally, though, they were effusive in their praise, giving it an 8 out of 10 score. Quote, this time around Microsoft has nailed it. Everything about the new Surface laptop feels way better. Microsoft has not only closed the MacBook Air Gap,

but also raised the bar for what you should expect from a Windows laptop. that starts at $999.99.99. I've fallen in love with the Surface laptop over the past week of using it. The battery life has blown me away, and the performance has mostly felt like using a regular thin and lightweight laptop. It no longer feels like Windows on Arm is being held back by underpowered hardware and major software issues. But to really make this new Surface laptop work its best, you still need Native Arm 64 apps. That situation is a lot better than it was a few years ago, with more of the top apps now native. But if you still need to run emulated ones, performance will vary depending on how complicated the application is and battery life will be impacted too.

I don't doubt that Premiere Pro will run a lot better when it finally comes to Windows on Arm later this year. But I think there will still be other app compatibility issues that some people will run into. There's also the legacy of Windows on Arm emulation not being great, so some app developers have just blocked their apps from even running emulated on RM64. Google Drive for desktop refuses to install with an error that says this Windows architecture is not supported. It's the same for any VPN apps, and if you need custom drivers for an old USB accessory, it's unlikely those drivers are available. Games also don't just work on the Snapdragon X Elite and X Plus, despite Qualcomm's assurances. The Surface laptop isn't a gaming laptop, so I wasn't expecting much here,

but I couldn't get Shadow of the Tomb Raider to stop crashing until I put it into full-screen exclusive mode and lowered the settings. I also tried Microsoft's new DLSS-S competitor, AutoSR, with Cyberpunk 2077, and all it did was force the resolution to 1024 by 768. I ran into similar issues with AutoSR and other games, which is probably why it's only officially limited to a small handful of titles. If you don't venture beyond the top Windows apps, you'll probably have a great experience like I did in terms of performance. in battery life. For anything more, you'll need to check to make sure your apps are compatible and run well. If this latest Windows on Arm Push is as successful as Apple's M1 Silicon,

that's an issue that should eventually disappear. Microsoft has created a great MacBook Air competitor here at $999. The base model 13-inch surface laptop is $100 less than Apple's M3 laptop, with Microsoft offering 16 gigabytes of memory instead of 8 gigabytes. If all you've ever wanted is a Windows-powered MacBook Air. That's pretty much what you're getting, end quote. I think my title for this episode today is maybe the most punny, the most dad joky of a title I've ever done in almost seven years of doing this show. Did I go too far?

Who can say? Talk to you tomorrow.