 Coming up on DTNS, Tom thinks he might have spotted the next Google, Tom's me. How much money do small U.S. carriers need to get rid of Huawei and ZTE equipment? And Patrick Norton's here to review the high points of the big hardware announcements for me. This is the Daily Tech News for Friday, September 4th, 2020 in Los Angeles. I'm Tom Merritt. Drawing the top tech stories from Cleveland, Ohio. I'm Len Peralta. I'm the show's producer, Roger Chang. And joining us, Patrick Norton, host of AVXL is back. Good to have you, man. Good to be here. Sarah Lane has the day off. Well deserved day off. But we're going to be talking about some of the EFA stuff. We got some AI stories. And we were just talking about, you know, everybody's deal with schools and however different areas of the country are dealing with that on good day internet. If you want that and some discussions of all kinds of other good stuff, go become a member patreon.com slash DTNS. Let's start with a few tech things you should know. Gamers in Africa may experience lower lag when playing online. If their ISPs are connected to the Angola Cables Network, Angola Cable used data collected from traffic routing during the lockdowns, when there was a lot of people playing games, to develop an IT transit system that they say reduces latency or lag by 20 to 50% for gamers in Africa and elsewhere in the Southern Hemisphere. Ping was reduced between 70 and 100 milliseconds, not two, but by 70 and 100 milliseconds during a demonstration event between gamers in South America and Africa. Russian search company Yandex will spin out its self-driving unit from MLUBV, which is a joint venture with Uber. After that spin out, Uber will own 19% with Yandex and its subsidiaries having the rest. Self-driving car unit currently has a testing fleet of 130 vehicles spread across Russia, Israel, and the United States. Speaking of Uber, Uber announced it will add an Uber Rent option to its mobile app in the UK. Users will be able to make a booking just before picking up the vehicle, although Uber is actually encouraging them to make a reservation at least 24 hours in advance. Uber has trialed similar services in Australia and France. Japan's antitrust regulator will take a closer look at Apple's App Store policies for any potential violations. Bloomberg sources say several Japanese game studios have expressed concern not so much about the 30% cut, but about the App Store including inconsistent enforcement guidelines, unpredictable content decision, and lapses in communication. That seems to be the thing, yeah. That seems to be the thing you definitely hear everyone agree on, even if they are not upset about the 30% cut. And a couple of ETH announcements from Huawei. The Honor Magic Book 14 and 15 laptops were announced. They'll have AMD's Ryzen 5 4500U CPUs inside. Magic Book 14 out September 21, starting at 749 euros and 99 cents. And the Magic Book 15 in October for 699 euros, 99 cents. The 16-inch MacBook Pro, sorry, that's what they want you to say. Magic Book Pro has AMD's Ryzen 5 4600H CPU. That one's coming September 7 for 899 euros, 99. And Huawei announced two Honor branded smartwatches. The ruggedized Watch GS Pro is said to meet US mill standard 810G procurement standards for withstanding extreme temps, humidity, rain, sand, and pressure. I doubt the US military will be buying any of them, but it meets the standard. It also offers a root back mode that'll either use GPS or GLONASS to help you retrace your steps if you're out hiking around. So it's kind of meant for outdoorsy stuff. Battery life is 100 hours in workout mode or 48 hours with GPS. That one's available September 7 for 249 euros, 99. And the more stylish typical Honor Watch ES has a 10-day battery life. Kind of a fitness tracker thing. Ship in September 7 for a mere 99 euros, 99 cents. All right, let's talk a little bit about a Facebook disclosure policy. Facebook has had a disclosure policy. Well, they've had a disclosure system, let's say, for a while, but they never codified the policy. Now they have got an official policy on third-party vulnerability disclosure. Under the new policy, Facebook says that when its researchers find critical bugs or security vulnerabilities in third-party software, this isn't dog-fooding. That's a whole different thing. This is we were using software in making Facebook and running Facebook, and we found a vulnerability. Facebook will give developers 21 days to respond and 90 days to fix the issues. Pretty standard. Company says it will make good faith efforts to report the bugs appropriately to third parties, but reserves the right to disclose sooner if a vulnerability is being actively exploited or even delay beyond the 90 days if everyone agrees there needs to be more time to fix it. Facebook says it will also disclose its own patched vulnerabilities, and in fact, in pursuant of that, the company published details on six WhatsApp vulnerabilities. None of those showed evidence of being exploited, and all of those have been fixed. And WhatsApp now has a little dedicated area for this if they disclose any vulnerabilities in the future. But Patrick, I don't know about you. This strikes me as a pretty standard vulnerability disclosure plan. People seem to be generally pleased with it. I keep playing like, wow, this seems enormously normal. I wonder if they'll actually live up to the simple and fairly low bar they've set for themselves. Well, yeah. I mean, because it's Facebook, you think advertising, but this isn't the advertising side. This is the security researcher side of Facebook. No, I mean, literally, it all seems reasonable and smart and intelligent. And I'm so burnt out on Facebook saying things and then forgetting what they said in terms of policy. I guess the reason I brought up that it's a different wing is that Facebook is more than one person. So hopefully you're right. This group is like, no, no, we're security researchers. We follow the norms of the industry. And this is a good example. This is a good precedent for that, to say like, oh, no, you're security. Third-party vulnerability policy looks pretty good, pretty standard. And all of the folks that I've read reacting to it, they may have a quibble here or a quibble there in minor stuff. But Facebook has a bug bounty program that's fairly unremarkable or uncontroversial. So hopefully... I would like to apologize to the entire Facebook security community for any offense that might have taken... I'll just leave it there. Until later when they screw it up. No, they're not going to screw it up. We believe in you. All right, a report from the U.S. Federal Communications Commission estimates the cost of removing Huawei and ZTE equipment from small telecommunications networks in the U.S. will be $1.8 billion. In other words, to be able to figure out how to rip that all out and replace it, $1.8 billion. Now, about $1.6 billion of that would qualify for federal reimbursement if the U.S. Congress appropriates the funds, which they have not yet done. The report only covers carriers that receive universal service funds provided to subsidized coverage in underserved areas. So the real overall cost in the U.S. of ripping out Huawei and ZTE equipment is likely larger. In a related write-up, at least it seems related to me, Wired Steven Levy has a great interview with Qualcomm's founder, Irwin Jacobs, that sheds light on the rise of Huawei and ZTE. Jacobs notes that Qualcomm could have been like Huawei, where they were making chips and making handsets and making other equipment. But wanted to get carriers to adopt CDMA as a standard. And back then in the 90s, getting the carriers saw a handset maker as a competitor. So Qualcomm backed out of making handsets. They had only ever made them in, guess where, Asia. They thought other U.S. companies would kind of pick up the baton, but Motorola and others didn't. They left that business leaving Qualcomm's chipsets, which were still available for sale, because that was what they built their business on, as an easy way for Chinese companies like Huawei and ZTE to fill the gap. Huawei and ZTE also, Jacobs points out, benefited from the Chinese government funding their research and development, something that U.S. companies did not get from the U.S. government. Do you think, does the $1.6 billion, $1.8 billion, does that seem kind of low for un-knitting this, or is that just because it's so restricted? Yeah, it does until you think about like, oh, this is only USF, right? This is the FCC saying, okay, the people who would qualify for reimbursement qualify because we give them money to build out. So we'll give them money to replace, and it's only those folks who qualify. And you're also talking about small carriers. Folks who live in big cities, probably don't even run into these carriers, but they're usually in rural areas. Sometimes they're in other underserved areas. You know, like poor areas, stuff like that. But they're small. That's kind of the point. The ATTs and the Verizon's, they have their own capital budgets and they're not trying to, well, they may be trying, but I don't think they're gonna get any money for this. You're laughing because I said they would be trying. And I'm like, yeah, but that's not really the point here. This is... I'm trying not to oversaturate DTNS with Snark, so I was trying to pull myself back from the abyss there. But yeah, so I thought that was interesting, too. Just the idea that that's a lot of money to spend, to rip out equipment that may have an issue. There's a lot of suspicion. There's not a lot of hard evidence that there's an issue, but some people feel the suspicion is all you need. And that's a couple of billion dollars that has to be spent in order to get it out. It's been interesting to watch that whole, is there proof, is there not proof, different agencies say they've seen proof, but it all remains very, very thin. We never got the follow-up on the Bloomberg Business Week story, did we? That's the only time anybody said they had found evidence, and everyone denied it, except for Bloomberg Business Week. So that was... I haven't forgotten, folks, still waiting to see if Business Week will ever clarify that story. They haven't yet, as far as I know. We talked with Andrew Main back in July about OpenAI's GPT-3 and what it can do to produce natural-sounding text. You remember Andrew had created a system for mimicking the voice of historical figures or anybody, including myself, that you could find text from. One problem with GPT-3, though, is while it can sound convincingly human, it isn't good at accuracy. Technology Review has a write-up about a company called DiffBot. Pay attention to this company. DiffBot, right now, takes text and creates a series of three-part factoids. That relate one thing to another. So, for example, if they looked at your bio, they would say, Patrick Norton is a writer. You know, subject, verb, object. Patrick Norton lives in St. Louis. Those factoids are all connected in a giant knowledge graph, and DiffBot's knowledge graph is huge. DiffBot, Google, and Microsoft are the only three U.S. companies that crawl the entire public web. They don't target it. They crawl everything. DiffBot uses some neurolinguistic processing, then, to extract factoids from the entire public web, and then adds them to its knowledge graph. It rebuilds its graph every four to five days, adding, they say, on average, 100 to 150 million new entities each month and use machine learning to create new connections and get rid of out-of-date ones. So, it's constantly trying to stay up-to-date. DiffBot's knowledge graph is available free to researchers, but it also has 400 paying customers, and those paying customers you may have heard of. Duck, duck, go, for instance. Snapchat, NASDAQ, all pay for access to this knowledge graph. In fact, Adidas and Nike use it particularly to search for counterfeit shoes, because you can go search for Adidas and Nike, but if you want to say, look, I just want to find the people that are actually selling shoes, DiffBot is good at that. It can say, okay, we'll return only the people who are actually trying to sell your shoes, and it even maybe give you a tip-off on which ones are legit and which ones are. DiffBot CEO, Mike Tong, says he plans to create a natural language interface to DiffBot. Right now you have to do some coding if you want to access it, but if you had that natural language interface, he wants to create a universal factoid question answering system where you could ask it any question, and it would not only answer accurately, but cite its sources. The reference librarian of the internet. And OpenAI's GPT-3 might play a part in creating that front end, because DiffBot isn't good at the natural language part, and OpenAI GPT-3 is. This struck me as, I'm not saying it is the next Google, but if there were to be the next Google, this is how I would kind of see it playing out, which is we're not going to be advertising-based, so we don't have that part. We make our money else-wise, and we can create a system that will just answer your question accurately without giving you a bunch of spam. I can think of nothing that would make me happier. I'm also kind of like, because you said there was, you know, the scale of the amount of, you know, search they're doing is so huge, and they've had $13 million in funding. They have 16 members according to CrunchBase, and they actually have a business model, which I think suggests an entire law on sort of the scale of venture capital investment versus practical nature of companies that I'm not qualified to make. Certainly, I'm sure. And these folks came out of Stanford, so, you know, very similar story of like smart folks at Stanford, figured out a new algorithm and then built a business on it, right? Sounds familiar, yeah. It's pretty impressive. You know, if DiffBot never turns into anything, don't, you know, I mean, you can talk me all you want. It's fine. I'm not guaranteeing anything here, but I definitely think this is something to keep an eye on. If you want to be able to later say, oh yeah, I heard about DiffBot way back in 2020. Tom Merritt told me about it on DTS. Please subscribe. Google Magenta is a Google AI project that makes AI-based tools to help the creative process. So you may have heard them. They can help you compose melodies or help you draw cats, because every AI thing has to start with cats somewhere. Now, Google Magenta has a tool to help you make lo-fi vibes to calm you down during stressful times. This was a project spearheaded by an intern, Vibrat Tio. He created a lo-fi player which lets you interact with objects in a web-based eight-bit room. So it's got eight-bit graphics, looks kind of retro, and you mix your own soundtrack. Now, Tio used human-made baselines, drum beats, et cetera. He also wrote four melody options. That was the bass. But then machine learning kicks in and riffs off those melodies. So every time you log in, you get a different one, right? It's taking the melody that was human-made and changing it somehow. And then the fun of it is you can just sort of click around in there and add in a baseline, change the beats per minute. And there's a couple of cool things in here where you, like, click on the TV in the room and the machine can interpolate between, say, the chill and the dense melody. You can click on a guitar on the wall and you can add that. You can add a baseline, but click it on the bass, click on the piano, add a piano. There's also a YouTube channel where you can type commands in chat to change the music along with the rest of the room. It's a fun demo toy, but it's an interesting example of being able to take the sort of machine learning stuff that's out there and create something with humans involved. It's also incredibly compelling. You like the one I got here? I'm going to add a drum. Let's see what happens. Thank you for holding. Oh! Your call is very important to us. I don't know what you think about this. I mean, it's just a fun tool for machine learning or is there anything more to it? I think anything, in one, I'm a sucker for anything that allows me to tap four times and create an entirely different melody or beat or track. I actually want to go to the YouTube channel and start experimenting with the commands you can type in. Yeah, there's a limit on what it can do, so you can't really mess with it too much. I think that's why it was safe to put into a YouTube channel like that. But yeah, if you're into a little lo-fi music, you can click on the cat and change the background sound from rain to a beach or change the cat to a dog and then you get cityscape sounds. No, no, it's kind of fun. Go check it out magenta.github.io slash lo-fi-player or we'll have it in the show notes at dailytechnewshow.com. Hey, folks, if you want to... I'm just going to use this. If you want to get all the tech headlines each day at about five minutes, be sure to subscribe to dailytechheadlines.com. All right, we've touched on a lot of hardware announcements out of IFA in Berlin this week. And now that we've had a few days to kind of digest them all, Patrick, help us break down some of the important takeaways. I feel like the biggest stories this week were sort of IFA-adjacent rather or IFA-timed rather than actually at IFA. Absolutely, yeah. I think the biggest one for me was of course the 11th Gen Tiger Lake CPUs. Well, I'll be honest with you, it's a toss-up between Nvidia's announcement of the 3000 series GPUs and the 11th Gen Tiger Lake CPUs. But I'll start with Intel because this announcement makes me feel like Intel isn't quite on its back foot and gasping for breath as much as it's felt with the processor announcements through this year. AMD was the first to PCI Express 4 on the desktop, Intel beat them to laptops. Early benchmarks for these processors are kind of hopeful. We have not seen any real-world benchmarks at this point. It updated 10 nanometer Superfin transistor technology which is basically saying more better process is bringing higher frequencies which is helping with performance. You know, there's like 50 or so laptops from the Nova Dell, Acer, MSI, several others that are going to be hitting between now and the holidays. I really want to see what the battery life looks like because they're talking about like a 20% improvement or sort of a 20, not a 20% improvement in battery life but that there is some significant, well, progress, I'll say in power consumption in various parts. And if I try to get any deeper than that, I'll get it wrong. But the EVO certification I think is really interesting. It's demanding nine plus hours of real-world battery life, fast charging as in 30 minutes of charging to get four hours of use out of the laptop, if you've ever wanted to charge a laptop in between stops on flights on a plane or if you're constantly running around that's a big deal. Wi-Fi 6, which I don't, Wi-Fi 6 I think is less compelling than we thought it would be. It's more future-proofing at this point than anything really, yeah. Yeah, the real-world improvements I don't think are really showing up a lot for home or small business networks. It's early yet. NVIDIA, again, technically wasn't an EFA but everyone I know that's a gamer with any kind of money is freaking out over the 3070, 3080 and 3090. Yeah, I think you were bold picking Intel in front of NVIDIA because Intel is more of a like, oh, okay, that wasn't a bad announcement. Everybody's freaking out about the 3090, I know. Yeah, and the 3090 is a ridiculous, ridiculous card. It is essentially a, it's essentially a Titan, right? They're just calling it a 3090. Some people have said the 3070 or the 3080s and 3070. I don't want to kind of get into a lot of the product positioning. I just, as I said to a friend of mine yesterday, I'm like, whatever you buy, there will be several more models sometime next year that will be delivering more performance for the same prices now or the same performance now for less money because NVIDIA especially always seems to fit every single price point in the stack. And they develop more and more cards. It's a little maddening actually when you start looking at sort of all the different options you have available. But there are so many CUDA cores available, especially in the 3090, like in excess of 10,000, which is an insane number. And everybody that reviews GPUs was kind of like, oh, they did that thing where they make the new GPUs sound amazing with really weird benchmark chart decisions. But fundamentally, these should be vastly, they should deliver vast improvements in performance, especially on RTX, we hope. Again, they haven't given anyone any cards to test. So they're kind of doing a thing that NVIDIA does, which is announcing the cards, getting as many people to buy them as possible. And then we'll see what the performance is. I think they worked really hard after, I think RTX was kind of underwhelming in real use. So they were trying to make some really serious efforts at the 3000 series. They've kind of sat on this for a long time. And they want to make sure the 3000 series... This feels like when ray tracing is finally worth having in there, right? Like they hadn't do it earlier to play the groundwork and this feels like the time it'll maybe pay off. Yeah. I mean, as you put it in one of the notes in the script, watch the GeForce RTX 3080 get medieval on an RTX 2080 TI and Doom Eternal. That was PC World's headline, not mine. I'm so sorry. My apologies, PC World. You know, this is also for the first time in years, actually, which I'd bought a larger power supply for my main desktop box one, because power supplies are still kind of erratic in terms of the ability to buy a decent high quality power supply right now. But the 3080 is looking for 320 Watts. The 3090 is like 350 Watts. The 3070 is a mere 220 Watts. You know, by the way, rumors are already hitting. We're actually Tweak Town, I should say, is reporting that supplies will be incredibly thin on these cards until next year. Just because people want them, is that why? I'll say go to Tweak Town to learn more. I think it's fair to drive the traffic their direction. But that's also kind of a classic situation with almost any GPU launch in recent memory. There's also been some interesting things where the 2000 series cards have mostly vaporized from the channel. And there's also some sort of coin mining activity that is particularly compelling to use GPUs for that is starting to drive GPU demands again. Man, I thought they'd all move to A6. A6 are no longer as, according to a friend who is deep in the mining lifestyle, he had actually got rid of, he had actually disassembled all of his systems with their umpteen GPUs per box. Because A6, he couldn't pee with A6 and now the algorithm seems to have switched in a way that makes the GPUs much more competitive for mining or profitable at this point. Qualcomm had a bunch of announcements that I think were kind of interesting to look at because kind of extending on that story earlier where they were like, well, we were doing this but we made this decision and now Qualcomm's kind of like, okay, we need lots and lots of revenue because it's that time of the year again, everybody wants revenue. The next Snapdragon 4 is essentially going to bring 5G to inexpensive phones to the mass market. The Snapdragon 8CX Gen 2.5G, which is ARM for Windows laptops and they're claiming better than 10th core i5 performance for like half the watts. So if a Core i5 is pulling 14 watts, they're claiming they deliver comparable performance at seven watts. The other part they talked about was their adaptive noise cancellation for true wireless earbuds. A lot of ear buds, a lot of wireless earbuds or true wireless earbuds are using Qualcomm chips as part of the device. My beloved Cambridge Audio Melomania or one of the first I knew of that actually used it. The performance is excellent and this is bringing some really slick Apple AirPod prograde active noise cancellation. So with the idea is that they will, as your earbuds shift around in your ears or maybe contact changes, they will actually adapt the noise cancellation but they will probably do it for a lot less money. And I think it's interesting to watch all of these developments on Qualcomm and Qualcomm kind of moving away from just being like, we make phones because of course there are so many competitors. Apple of course has their own in-house processing but to see them kind of really trying to create a Windows ARM processor that can really pound on Intel is an interesting move for them to make. Very curious to see real-world performance. Yeah, finally let's finish with thoughts on just the broader question of IFA having some in-person but really it was minimal. And this was the first of the huge tech conferences that was consumed majority online. It's been interesting to look at because it feels like IFA will probably be much closer. In 2021 will be much closer to a traditional IFA. They basically cut down the number of invitees. They sort of cut everything down to the bone. And at the event this year, of course we already know that CES has been canceled and it was interesting because it was this very weird moment where I got an email from IFA because I thought there might be a possibility I could attend IFA in Meet Space to use one of my favorite phrases. And IFA was basically like, you're no longer invited. You're in America, you guys have issues and keep your filthy... We're keeping it to 5,000 people tops anyway, so sorry. Thank you for your interest. Stay in America. But it's been interesting to watch because at this point it seems like they are going to be going back to their full-scale operation in 2021. We wait with bated breath where at the same time this year, CES decided to go all virtual. And it's been interesting because a lot of what people seem to be hearing is that a lot of organizations are starting to sort of like, okay, if we're going to pull away from these events, then we can start doing our own virtual events. How do we do this? Is there traction we can get? I feel like there's in a year we're either going to be kind of right back on to the sort of traditional trade shows that we're used to or that trade shows are going to be limping along and becoming a part of the past. Yeah, and just becoming big streamed press conferences. Well, folks, you got thoughts on that. Join in the conversation in our Discord, which you can join by linking to a Patreon account at patreon.com slash DTNS. Shout out to the patrons at our master and grand master levels, including Carmine Bailey, John and Becky Johnston, and Chris Benito. Len Peralta, you always find a good angle to draw on our topics. What have you found for us today? Well, you had mentioned a word. You said a word to a minimalist about tech events. And that's my take on this. Tech events like the tech themselves are getting smaller and minimal, at least this year they are. And this is my take on that, even going with a very minimalist look of just, this isn't supposed to be Patrick, but he's someone sitting at a desk saying, these launch events are getting to be even smaller than the chips. So yeah, so this was just something very, very quick and easy and small. And I believe this one is called Everything is Getting Smaller, and it's available right now at my Patreon, patreon.com forward slash Len, or at my online store at lenneperaltastore.com. Patrick, does this capture the way you're feeling about these events? I'll be honest with you, for me, so much of the importance of the events is all of the things you get to do that are not specifically getting the commercial from the vendors, all the communication, all of the sort of behind the scenes conversations that go on. So I figure probably more sobbing in my case. Add some tears to that drawing. Well, if you want to see Patrick when he's not crying, where should you go? AVXL.com is a very, very good place to go. That's a podcast I host with Robert Herron. We talk about home theater and audio technology. And we're going to start talking about the joys of putting in the Atmos height speakers. Let me tell you something. You haven't realized you're running out of cable until you start installing four speakers in the ceiling. So much wire. Ceiling speakers are watching you. Go check it out. Atmos is really kind of obscene, though, in a cool way. That's AVXL, the A-V-E-X-C-E-L, like the spreadsheet, AVXL.com. Also, Paolo used the perk that we have for some patrons where they can ask us to record something for them to have us record his ways directions. So he got me, Roger, and Sarah to all record, like, in 100 kilometers, in 200 kilometers. Well, probably not kilometers. I think they were just meters. But you can now use them, too. Not only did he want to use them for himself, but he made them into ways links. So you could have Sarah telling you where to go in your car, or Roger, or me. You can find those links at patreon.com. Slash DTNS. Thank you, Paolo, for making those and letting us share them with the rest of the crowd. That was pretty awesome. That's fun. Yeah. Our email address is feedback at dailytechnewshow.com. We're live Monday through Friday, 4.30 p.m. Eastern, 20.30 UTC. You can find out more at dailytechnewshow.com. Live, don't forget, holiday Monday here in the United States Labor Day, but we'll be back Tuesday with a Frenchman, Patrick Beja. Talk to you then.