WEBVTT - Tech News: Nature Goes Punk Rock 0:00:04.400 --> 0:00:07.800 Welcome to tech Stuff, a production from I Heart Radio. 0:00:12.039 --> 0:00:14.920 Hey there, and welcome to tech Stuff. I'm your host, 0:00:15.080 --> 0:00:18.240 Jonathan Strickland. I'm an executive producer with I Heart Radio 0:00:18.360 --> 0:00:21.120 and I love all things tech and it is time 0:00:21.200 --> 0:00:26.920 for the news for Thursday, February eleven, twenty one. Let's 0:00:26.960 --> 0:00:31.040 get into it. SI Tech Daily published an article describing 0:00:31.080 --> 0:00:35.600 how computer scientists were able to defeat deep fake detecting 0:00:35.640 --> 0:00:40.120 computer algorithms and systems. So deep fakes, for those who 0:00:40.240 --> 0:00:45.280 aren't really that familiar with the term, are fabricated digital videos. 0:00:45.360 --> 0:00:49.879 They show what appears to be someone, but it's not 0:00:49.920 --> 0:00:53.800 really them. It's been digitally created. And the typical way 0:00:53.960 --> 0:00:56.600 that these are made is by feeding a lot of 0:00:56.720 --> 0:01:01.480 videos of your target person. So it's easier if we 0:01:01.600 --> 0:01:06.600 use an example. Let's say Ronald McDonald. You've got videos 0:01:06.640 --> 0:01:10.559 of Ronald McDonald, and you just feed all those videos 0:01:10.560 --> 0:01:14.240 of Ronald McDonald, the one specific incarnation of Ronald, because 0:01:14.240 --> 0:01:17.720 obviously lots of different actors have played Ronald McDonald, But 0:01:17.959 --> 0:01:21.600 you feed commercial after commercial of a specific actor playing 0:01:21.680 --> 0:01:25.840 Ronald into your deep fake generator. The more videos you have, 0:01:25.959 --> 0:01:29.839 the better, because you are training the digital machine learning 0:01:29.880 --> 0:01:34.920 model on how to represent that person, and then the 0:01:35.160 --> 0:01:41.480 deep fake generator can superimpose your target, Ronald McDonald their 0:01:41.600 --> 0:01:45.280 face on top of another video. So you could take 0:01:45.319 --> 0:01:48.080 a video of a totally different person. Maybe it's someone 0:01:48.120 --> 0:01:50.840 who's doing a credible impression of the target. Maybe it's 0:01:50.880 --> 0:01:54.440 not even that they're just going through various motions, and 0:01:54.440 --> 0:01:58.120 then you use the deep fake generator to overlay the 0:01:58.160 --> 0:02:02.360 digital face of Ronald McDonald on top of the other 0:02:02.440 --> 0:02:06.120 actors face in the video, and it makes it seem 0:02:06.160 --> 0:02:10.040 as though the target, Ronald was doing whatever it was 0:02:10.360 --> 0:02:13.079 the actor was doing the whole time. And detectors tend 0:02:13.160 --> 0:02:15.640 to look for little signs and videos that would indicate 0:02:15.680 --> 0:02:19.280 a deep fake, particularly around the eyes. The eyes are tricky, 0:02:19.320 --> 0:02:22.680 but the scientists who worked on this found that by 0:02:22.760 --> 0:02:28.160 inserting something called an adversarial example, or actually a lot 0:02:28.200 --> 0:02:32.760 of them, into videos, they could fool these detectors. So 0:02:32.840 --> 0:02:36.600 generally the idea is to include tiny signs that force 0:02:36.760 --> 0:02:40.200 these detectors to make mistakes. Now, according to the scientists, 0:02:40.320 --> 0:02:44.440 it's even possible to introduce the sort of adversarial examples 0:02:44.480 --> 0:02:47.960 into a deep fake video without any real knowledge of 0:02:48.080 --> 0:02:51.920 how the detectors actually process the videos in order to 0:02:52.040 --> 0:02:55.680 detect deep fakes, but a little knowledge does go a 0:02:55.720 --> 0:02:59.519 long way. The team tested scenarios in which attackers might 0:02:59.560 --> 0:03:03.200 have complete access to the detector model. In other words, 0:03:03.200 --> 0:03:06.440 they know exactly what the detector is looking for, and 0:03:06.480 --> 0:03:11.120 with that and introducing these adversarial examples, they could have 0:03:11.400 --> 0:03:17.200 a more than success rate in avoiding detection for videos 0:03:17.240 --> 0:03:21.160 that were uncompressed. They also tested scenarios in which hackers 0:03:21.240 --> 0:03:25.000 might only have very limited knowledge of the detector model, 0:03:25.280 --> 0:03:28.239 and even then with uncompressed video they saw an eight 0:03:28.440 --> 0:03:32.200 six percent success rate. Now, the approach used by the 0:03:32.240 --> 0:03:36.680 team preserves those adversarial examples even if a video does 0:03:36.760 --> 0:03:40.640 go through compression or resizing. Typically those are things that 0:03:40.680 --> 0:03:45.520 would remove adversarial examples from images, and under both scenarios 0:03:45.800 --> 0:03:48.800 the method worked at a slightly lower success rate if 0:03:48.880 --> 0:03:53.600 the video had been compressed, It was nearly successful if 0:03:53.640 --> 0:03:56.760 the team had complete knowledge of the detector model, and 0:03:57.560 --> 0:04:01.280 percent even if they didn't have that knowledge, so still 0:04:01.520 --> 0:04:07.120 incredibly reliable. They also chose not to publish the code 0:04:07.360 --> 0:04:10.880 that they were using for their adversarial examples because they said, 0:04:11.200 --> 0:04:14.600 if we did that, then bad guys could take this 0:04:14.880 --> 0:04:17.640 and make it their own and thus use it for 0:04:17.720 --> 0:04:21.599 real zs. So they recommended that teams that are designing 0:04:21.680 --> 0:04:26.320 deep fake detectors incorporate adversarial training when they build out 0:04:26.320 --> 0:04:29.120 their models. I think most teams already do this. Essentially, 0:04:29.560 --> 0:04:32.520 this creates a sort of seesaw effect. So you've got 0:04:32.560 --> 0:04:35.560 one computer system and its job is just to try 0:04:35.600 --> 0:04:39.240 and create the most convincing fakes possible, and you have 0:04:39.279 --> 0:04:43.160 a different system that is attempting to tell the difference 0:04:43.200 --> 0:04:47.160 between real videos and fake videos. So as the detectors 0:04:47.160 --> 0:04:51.080 get better, the faker system tries out new tricks to 0:04:51.320 --> 0:04:55.640 exploit potential weaknesses in the detector, and as the detectors 0:04:55.680 --> 0:04:59.960 fall behind, they begin to adjust their approach to get 0:05:00.000 --> 0:05:02.400 it are at picking out the fakes. And we've seen 0:05:02.400 --> 0:05:04.880 this sort of process in a lot of other areas 0:05:04.920 --> 0:05:08.400 of artificial intelligence. It really plays a big role in 0:05:08.480 --> 0:05:13.520 how AI evolves over time. Earlier this week, I talked 0:05:13.520 --> 0:05:17.000 about Tesla and bitcoin, and I did my normal angry 0:05:17.000 --> 0:05:20.280 old man rant about bitcoin being more of a commodity 0:05:20.400 --> 0:05:24.240 than it is a currency. Well, now I get to 0:05:24.279 --> 0:05:27.680 talk about another thing that irritates me about bitcoin, and 0:05:27.720 --> 0:05:31.520 that's how much juice bitcoin miners are using when they're 0:05:31.520 --> 0:05:35.120 going after those expensive digital coins. So let's get some 0:05:35.200 --> 0:05:39.599 quick explanations in here. You've got your digital currency bitcoin, 0:05:39.880 --> 0:05:43.520 and there are two main ways you can get bitcoin. 0:05:44.080 --> 0:05:46.280 There more than this, but these are the two big ones. 0:05:46.560 --> 0:05:49.880 You can purchase some so you're exchanging some other form 0:05:49.920 --> 0:05:53.320 of currency for some amount of bitcoin, or you can 0:05:53.360 --> 0:05:58.919 try and mine bitcoin. Mining essentially involves using a computer 0:05:59.040 --> 0:06:03.760 to guess the correct answer to a very hard mathematical problem, 0:06:03.800 --> 0:06:07.120 and the Bitcoin system, that is, the overall network that's 0:06:07.120 --> 0:06:12.159 in charge of tracking bitcoin transactions and dispensing mind bitcoins. 0:06:12.920 --> 0:06:16.960 The system determines how hard that mathematical problem should be. Now, 0:06:17.000 --> 0:06:21.120 ideally it should take about ten minutes before some computer 0:06:21.320 --> 0:06:24.240 on the network is able to solve the problem. So 0:06:24.279 --> 0:06:28.480 as more computational power joins this network, the system has 0:06:28.520 --> 0:06:33.600 to make the problems more challenging. Otherwise these screaming fast 0:06:33.640 --> 0:06:37.480 computer systems would solve the problems far too quickly. So 0:06:37.560 --> 0:06:40.279 the goal is to aim for that average amount of 0:06:40.320 --> 0:06:42.280 time that it takes to mind bitcoins, to keep it 0:06:42.320 --> 0:06:44.960 in around ten minutes. There's more to it than that, 0:06:45.080 --> 0:06:49.560 but that's a very basic explanation and since a single 0:06:49.600 --> 0:06:53.400 bitcoin is worth tens of thousands of dollars, nearly fifty 0:06:53.800 --> 0:06:57.560 dollars as I record this, and each successful mining of 0:06:57.880 --> 0:07:02.080 bitcoin pays out six point to five bit coins per go, 0:07:02.240 --> 0:07:05.400 so every ten minutes, another six and a quarter bitcoins 0:07:05.400 --> 0:07:08.359 get mind, that means there's a lot of money to 0:07:08.360 --> 0:07:12.120 be made, but it also means you need incredibly powerful 0:07:12.160 --> 0:07:14.760 computer systems if you're going to stand a chance of 0:07:14.840 --> 0:07:18.560 having the computer that guesses the correct answer to the 0:07:18.600 --> 0:07:22.040 mathematical problem. Otherwise someone else is going to beat you 0:07:22.080 --> 0:07:25.239 to the punch. And so people and groups have built 0:07:25.240 --> 0:07:29.560 out mining computers and really mining computer networks that are 0:07:29.640 --> 0:07:31.480 just trying to be the first to get these math 0:07:31.600 --> 0:07:35.560 problems correct. And these machines require a lot of electricity. 0:07:35.920 --> 0:07:39.480 Cambridge University researchers have estimated that the power consumption is 0:07:39.520 --> 0:07:42.560 in the neighborhood of one d twenty one point thirty 0:07:42.600 --> 0:07:47.320 six tara watt hours per year. That means all the 0:07:47.360 --> 0:07:50.760 bitcoin miners in the world collectively are consuming as much 0:07:50.800 --> 0:07:55.360 electricity as the entire nation of Argentina, and they're just 0:07:55.560 --> 0:07:59.880 behind Norway. Now. The only reason for this is because 0:08:00.120 --> 0:08:03.760 of those huge payouts, right, because if the amount you 0:08:03.880 --> 0:08:08.160 earned was a low amount, whether because you were only 0:08:08.200 --> 0:08:10.800 getting a fraction of a bitcoin every time you mind it, 0:08:11.040 --> 0:08:14.960 or the value of bitcoin had plummeted, or both, then 0:08:15.520 --> 0:08:18.600 it would mean that running your computer systems would actually 0:08:18.640 --> 0:08:21.640 be more expensive than the money you would get from 0:08:21.760 --> 0:08:25.640 mining bitcoins, and that would be a losing proposition, so 0:08:25.800 --> 0:08:28.600 you would quit right. You wouldn't want to keep spending 0:08:28.640 --> 0:08:31.720 money to make back a fraction of what you were spending. 0:08:32.360 --> 0:08:34.640 And maybe that will happen at some point. And if 0:08:34.679 --> 0:08:38.120 it does, and people drop out of the network because 0:08:38.280 --> 0:08:40.800 it costs too much money to continue mining, then the 0:08:40.840 --> 0:08:44.840 system will adjust that difficulty to those mathematical problems. Again, 0:08:45.080 --> 0:08:47.680 if it detects that it's taking more than ten minutes 0:08:47.720 --> 0:08:51.719 to solve the problems, it makes the problems easier. The 0:08:51.760 --> 0:08:54.840 idea of being that no matter what computer power is 0:08:54.840 --> 0:08:58.760 connected to the system, that ten minute to solution time 0:08:58.840 --> 0:09:03.200 remains fairly steady. So it's kind of like if you 0:09:03.400 --> 0:09:07.040 took the top performing students of a math class, but 0:09:07.120 --> 0:09:10.199 you still wanted to have the same you know, average grade, 0:09:10.679 --> 0:09:14.440 then you make the tests a little easier. But the 0:09:14.520 --> 0:09:17.600 energy demands of bitcoin are one of the aspects of 0:09:17.640 --> 0:09:21.440 the cryptocurrency that I find really irritating. Not only is 0:09:21.480 --> 0:09:25.160 it a quote unquote currency with a value that fluctuates 0:09:25.320 --> 0:09:27.840 so much that you can't really use it as a 0:09:27.880 --> 0:09:32.360 currency with confidence, it's also encouraging people to consume way 0:09:32.400 --> 0:09:36.600 more electricity than they would otherwise, and that has further 0:09:36.800 --> 0:09:40.000 consequences depending upon where that electricity is coming from. It 0:09:40.080 --> 0:09:44.240 might mean that bitcoin itself is contributing to climate change 0:09:44.640 --> 0:09:47.960 because people are using so many computer systems that are 0:09:48.000 --> 0:09:50.600 drawing power off a power grid that could be powered 0:09:50.600 --> 0:09:55.679 by coal. And then there are other, admittedly more minor consequences, 0:09:55.679 --> 0:09:58.600 such as not being able to find a decent graphics 0:09:58.640 --> 0:10:02.240 processing unit these days is because cryptocurrency miners keep buying 0:10:02.280 --> 0:10:10.600 them all stupid bitcoin. Cybersecurity researcher Alex Berson recently gave 0:10:10.679 --> 0:10:13.800 us more evidence that supply chain attacks are something that 0:10:13.840 --> 0:10:18.880 companies really need to take seriously, which is a theme 0:10:18.960 --> 0:10:22.160 we've been seeing since the Solar Winds hack. So, in 0:10:22.200 --> 0:10:25.880 this context, a supply chain attack is when a hacker 0:10:25.920 --> 0:10:29.320 is able to inject malware into some sort of product 0:10:29.600 --> 0:10:33.240 from a trusted entity that distributes that product to its 0:10:33.280 --> 0:10:37.160 partners or customers. Or whatever. But and itself sounds a 0:10:37.160 --> 0:10:42.280 bit confusing, so let's use a grim Batman style analogy. 0:10:42.440 --> 0:10:45.720 Let's say the Joker has just about had enough of 0:10:45.720 --> 0:10:49.560 Commissioner Gordon, and he knows that Commissioner Gordon likes pizza, 0:10:50.000 --> 0:10:52.480 So the Joker has one of his goons deliver a 0:10:52.760 --> 0:10:58.160 poisoned pizza to Gordon's house. Now, Gordon might accept the pizza, 0:10:58.640 --> 0:11:01.559 he might decline it, he might accept it but not 0:11:01.880 --> 0:11:05.080 eat it. So in other words, this is a ploy 0:11:05.120 --> 0:11:08.440 that might work, but it might not. So let's say 0:11:08.559 --> 0:11:12.920 instead the Joker decides, ah, I'm going to infiltrate Gordon's 0:11:13.080 --> 0:11:16.600 favorite pizza parlor. And the Joker gets in there and 0:11:16.679 --> 0:11:20.559 poisons all the dough in that pizza parlor, and then 0:11:20.559 --> 0:11:23.080 he has one of his goons drop off some coupons 0:11:23.240 --> 0:11:27.080 for the pizza parlor over at Gordon's home. Now, the 0:11:27.160 --> 0:11:30.000 Joker is banking on the fact that Gordon, he knows 0:11:30.040 --> 0:11:33.120 this pizza parlor, he likes the pizza there, and he 0:11:33.280 --> 0:11:36.440 just got some coupons for it, and he's gonna probably 0:11:36.600 --> 0:11:39.880 order pizza from that pizza place, and he will welcome 0:11:39.960 --> 0:11:43.560 the poisoned pizza into his home. Eagerly. That's kind of 0:11:43.600 --> 0:11:46.800 what happens with a supply chain attack because the malware 0:11:47.320 --> 0:11:51.160 is hidden inside a product that's from a trusted partner, 0:11:51.360 --> 0:11:54.520 so the victim is more likely to incorporate that malware 0:11:54.600 --> 0:11:58.240 into their own systems. Vers San was able to create 0:11:58.320 --> 0:12:02.480 an attack through squaw adding, which is grabbing a valid 0:12:02.600 --> 0:12:06.840 internal package name on code that happened to be distributed 0:12:06.840 --> 0:12:10.199 through geth hub. Essentially, he was leaning on an open 0:12:10.280 --> 0:12:14.319 source approach to code distribution, and a lot of companies 0:12:14.760 --> 0:12:18.480 use that for various products and it worked really well. 0:12:18.960 --> 0:12:22.240 His quote unquote malware which really didn't do much but 0:12:22.320 --> 0:12:25.960 returned some very basic information. Because he didn't want to 0:12:26.520 --> 0:12:30.000 actually cause any problems, he didn't want to compromise these systems. 0:12:30.320 --> 0:12:32.240 He was doing this a sort of a type of 0:12:32.280 --> 0:12:36.560 penetration testing to see if the companies regarded against this 0:12:36.679 --> 0:12:40.160 kind of attack. He found that it was effective in 0:12:40.559 --> 0:12:45.040 at least thirty five organizations, and these included really big ones. 0:12:45.320 --> 0:12:48.440 Apple was one of them. For example. He called it 0:12:48.600 --> 0:12:53.000 a quote dependency confusion bug end quote, and indicated that 0:12:53.240 --> 0:12:56.600 if a malicious hacker had taken that approach, they might 0:12:56.600 --> 0:13:00.000 have been able to do stuff like install backdoor access 0:13:00.040 --> 0:13:03.680 points into what would otherwise appear to be a secure system. 0:13:03.720 --> 0:13:07.079 To get into full detail of what he did would 0:13:07.120 --> 0:13:10.880 require an understanding that goes beyond what I have. But 0:13:11.120 --> 0:13:13.319 it does show that this kind of attack can be 0:13:13.360 --> 0:13:17.320 devastatingly effective if used by malicious hackers, and it could 0:13:17.360 --> 0:13:20.600 mean that in the future we'll see companies quarantine a 0:13:20.720 --> 0:13:25.000 system in order to test out patched software before incorporating 0:13:25.040 --> 0:13:29.160 that patch into the full system overall, just in case 0:13:29.400 --> 0:13:33.520 someone has compromised a trusted partner. I guess you just 0:13:33.559 --> 0:13:36.320 can't trust anyone. That's going to be a theme that 0:13:36.360 --> 0:13:39.720 comes back later. And in our last little bit for 0:13:39.760 --> 0:13:44.560 this segment, Twitter's chief financial officer Ned Siegel said in 0:13:44.640 --> 0:13:48.120 an interview that the band that Twitter has placed on 0:13:48.280 --> 0:13:53.480 former President Donald Trump's account is in fact permanent. He's 0:13:53.520 --> 0:13:57.560 really most sincerely banned. I guess you could say Trump 0:13:57.559 --> 0:14:00.160 received the band toward the end of his presidency after 0:14:00.240 --> 0:14:03.320 the riots at the US Capitol on January six, and 0:14:03.360 --> 0:14:06.000 that ban is going to be in place even if 0:14:06.040 --> 0:14:10.800 Trump is re elected to president sometime in the future. Yikes, 0:14:10.840 --> 0:14:14.920 So Siegel says to CNBC, the way our policies work 0:14:15.080 --> 0:14:19.120 when you're removed from the platform. You're removed from the platform, 0:14:19.280 --> 0:14:22.960 whether you're a commentator, you're a CFO, or you are 0:14:23.040 --> 0:14:27.480 a former or current public official. Remember, our policies are 0:14:27.480 --> 0:14:30.680 designed to make sure that people are not inciting violence, 0:14:30.920 --> 0:14:33.840 and if anybody does that, we have to remove them 0:14:33.880 --> 0:14:37.000 from the service. And our policies don't allow people to 0:14:37.040 --> 0:14:41.280 come back. End quote. Earlier this week, Twitter held and 0:14:41.400 --> 0:14:45.280 earnings called to share information with shareholders about how the 0:14:45.320 --> 0:14:47.840 company is doing. And I'm sure there was some concern 0:14:47.920 --> 0:14:51.240 that banning Trump was going to have a negative impact 0:14:51.440 --> 0:14:55.560 on the business side of the service. According to c NBC, 0:14:55.720 --> 0:14:59.880 Twitter beat analyst projections for earnings and for revenue into 0:15:00.000 --> 0:15:03.120 twenty twenty, but it did fall behind when it came 0:15:03.160 --> 0:15:07.480 to growing the user base. Twitter CEO Jack Dorsey also 0:15:07.520 --> 0:15:12.040 revealed that of Twitter's users are actually outside the United States. 0:15:12.280 --> 0:15:15.280 Twitter is also still dealing with being in the crosshairs 0:15:15.360 --> 0:15:19.440 of some world events, such as the ongoing farmer protests 0:15:19.560 --> 0:15:23.640 in India that I reported on previously. Well, we have 0:15:23.800 --> 0:15:26.400 some more news stories to cover, but before we get 0:15:26.440 --> 0:15:36.400 to any of those, let's take a quick break. On 0:15:36.440 --> 0:15:39.640 Tuesday's episode, I talked about how Apple's head of hardware 0:15:39.680 --> 0:15:43.440 had moved over to the company's Virtual and augmented reality department, 0:15:44.000 --> 0:15:46.840 which I guess we can just call mixed reality. That's 0:15:46.920 --> 0:15:50.360 kind of the the umbrella term for those technologies. Well, 0:15:50.400 --> 0:15:52.960 today we can follow up a little bit more on that. 0:15:53.360 --> 0:15:57.440 Mac Rumors published an article including designs by artist Antonio 0:15:57.520 --> 0:16:00.960 di Rosa, who has worked with Apple on several occasions 0:16:01.000 --> 0:16:05.280 to create artists renderings of different design concepts. Now, these 0:16:05.280 --> 0:16:07.920 renderings are meant more as a way to consider, you know, 0:16:08.000 --> 0:16:12.000 different design options. They may or may not resemble a 0:16:12.080 --> 0:16:15.440 final product from Apple. In fact, there's no guarantee that 0:16:15.480 --> 0:16:18.680 there will be a final product in some cases, although 0:16:18.920 --> 0:16:22.040 it seems pretty certain that this is going forward and 0:16:22.240 --> 0:16:26.480 de Rosa's pieces, these mixed reality headsets are given the 0:16:26.560 --> 0:16:30.760 name Apple View, but there is no indication that Apple 0:16:30.800 --> 0:16:34.000 has actually settled on any name as of yet. I'm 0:16:34.000 --> 0:16:37.120 still holding out for Eye Eyes or something like that. 0:16:37.760 --> 0:16:40.040 So what was it looked like? Well, it kind of 0:16:40.040 --> 0:16:42.360 looks like someone's got an iPhone strapped to their head. 0:16:42.520 --> 0:16:44.800 I mean, it just imagined that you've got a headband 0:16:45.400 --> 0:16:48.600 and there's a visor that fits over the eyes. It 0:16:48.680 --> 0:16:53.120 is kind of like a horizontal somewhat oval shaped iPhone, 0:16:53.160 --> 0:16:55.720 and it's curved so that it can fit over the 0:16:55.760 --> 0:16:58.720 eyes of the user. Oh and you better believe there's 0:16:58.720 --> 0:17:01.760 an Apple logos act dab in the middle of that thing. 0:17:02.440 --> 0:17:06.119 According to the website the information, the Apple design is 0:17:06.160 --> 0:17:09.880 said to incorporate more than twelve cameras in it, so 0:17:10.400 --> 0:17:14.800 this gadget could potentially be both for virtual and augmented reality. 0:17:15.080 --> 0:17:17.720 Those cameras could be used to provide a live video 0:17:17.800 --> 0:17:20.960 feed of the users surroundings, and the device could lay 0:17:21.000 --> 0:17:23.760 digital information on top of that, or it might also 0:17:23.800 --> 0:17:26.320 be used so that the headset can just have a 0:17:26.359 --> 0:17:30.040 sense of a user's environment and their head position and 0:17:30.080 --> 0:17:33.919 orientation if they are using it as a virtual reality device, 0:17:35.040 --> 0:17:37.320 thus giving queues to users so that you know, they 0:17:37.359 --> 0:17:40.120 don't do stuff like run into a wall or something. 0:17:41.040 --> 0:17:43.880 The general rumor right now is that it is going 0:17:43.920 --> 0:17:47.080 to be geared more as a VR device than an 0:17:47.080 --> 0:17:49.320 a R device. Maybe there will be some a R 0:17:49.480 --> 0:17:52.679 use cases, but they'll be more limited. That actually shocks 0:17:52.720 --> 0:17:55.640 me considering the number of cameras incorporated into this thing, 0:17:56.040 --> 0:17:59.199 though presumably those cameras will provide the headset with a 0:17:59.240 --> 0:18:01.800 lot of a bill these to track a user's head 0:18:01.800 --> 0:18:06.640 motions anyway. It also is supposed to have to eight 0:18:06.760 --> 0:18:09.600 K displays in there, one for each eye, and all 0:18:09.640 --> 0:18:12.600 of this helps explain why most of the websites I've 0:18:12.640 --> 0:18:15.920 seen talking about this thing are guessing that's going to 0:18:16.040 --> 0:18:19.520 cost around three thousand dollars when it does become available 0:18:19.520 --> 0:18:22.439 for purchase. It's supposed to be announced before the end 0:18:22.480 --> 0:18:25.600 of this year, and JP Morgan predicts that will go 0:18:25.640 --> 0:18:28.879 on sale within the first quarter of twenty twenty two, 0:18:29.119 --> 0:18:32.640 so just a year. That's kind of exciting. And speaking 0:18:32.760 --> 0:18:35.280 of cameras, well, we all know that we live in 0:18:35.320 --> 0:18:38.120 a world peppered with cameras. Lots of us have one 0:18:38.280 --> 0:18:40.840 within reach at any given moment thanks to our phones, 0:18:41.280 --> 0:18:45.080 and sure those cameras catch plenty of ridiculous and trivial stuff, 0:18:45.280 --> 0:18:47.600 but it also means that tons of people have powerful 0:18:47.640 --> 0:18:51.320 tools to document important events at a moment's notice, and 0:18:51.320 --> 0:18:55.399 that includes capturing video of police activity, something that some 0:18:55.520 --> 0:18:58.720 police have really tried to discourage in the past, but 0:18:58.760 --> 0:19:01.199 at least here in the United States, it is completely 0:19:01.280 --> 0:19:05.360 legal to record video of a police officer performing official 0:19:05.480 --> 0:19:09.400 duties in public, even if those police officers don't like it. 0:19:10.000 --> 0:19:13.040 Bling Bling reports that now there's a growing trend of 0:19:13.080 --> 0:19:17.440 police officers trying to leverage copyright takedown notices in an 0:19:17.440 --> 0:19:21.560 effort to discourage people from posting and sharing videos of them. 0:19:21.920 --> 0:19:25.080 The idea is pretty simple. So police officer sees that 0:19:25.200 --> 0:19:27.960 someone is trying to capture them on video, So the 0:19:27.960 --> 0:19:31.600 police officer whips out his or her own cell phone 0:19:32.119 --> 0:19:35.600 and blasts some copyrighted music. So the idea here is 0:19:35.640 --> 0:19:39.600 that the person capturing the video is getting that music 0:19:39.720 --> 0:19:42.639 on the microphone as well. If they upload the video 0:19:42.680 --> 0:19:46.240 to a social platform, the owner of the copyrighted music 0:19:46.280 --> 0:19:49.600 will issue a takedown notice for the unlicensed use of 0:19:49.640 --> 0:19:53.639 the music, particularly if it's all an automated system. Pretty 0:19:53.720 --> 0:19:58.000 shady stuff, right, And for activists this is really hard 0:19:58.040 --> 0:20:01.800 because if they get enough complaints about videos that they're posting, 0:20:02.240 --> 0:20:05.760 they might receive a ban from their platforms. And many 0:20:05.800 --> 0:20:09.480 of these activists are trying to perform a public service. 0:20:10.280 --> 0:20:12.760 In the wake of the Black Lives Matter movement, we 0:20:12.840 --> 0:20:17.080 have seen how important it is to hold police accountable, 0:20:17.200 --> 0:20:19.800 to make sure that those who are supposed to be 0:20:19.880 --> 0:20:24.199 protecting and serving and enforcing the law are themselves also 0:20:24.600 --> 0:20:29.280 subject to the law. The Drive has an interesting article 0:20:29.400 --> 0:20:32.760 about how farmers are turning to hackers in order to 0:20:32.800 --> 0:20:35.840 be able to repair their own equipment. All right, So 0:20:35.880 --> 0:20:37.760 this gets out a few different things that I've talked 0:20:37.800 --> 0:20:42.000 about on previous episodes of tech Stuff. There's this concept 0:20:42.080 --> 0:20:45.479 called the right to repair, which is, if you purchase 0:20:45.560 --> 0:20:49.280 some technology, whatever it may be. Maybe it's a stereo system, 0:20:49.320 --> 0:20:53.119 maybe it's a harvester for a farm, but you should 0:20:53.160 --> 0:20:55.960 be able to make reasonable repairs to that technology all 0:20:56.000 --> 0:20:59.000 on your own. The limits of your repair should really 0:20:59.040 --> 0:21:02.800 be up to your expertise and knowledge. But a lot 0:21:02.880 --> 0:21:07.280 of companies have made repairs really hard, from using proprietary 0:21:07.400 --> 0:21:11.240 screws and other fasteners, to making it intentionally difficult to 0:21:11.320 --> 0:21:15.200 access components, to including terms of service that discourage people 0:21:15.240 --> 0:21:17.800 from opening up a piece of technology in the first place. 0:21:18.280 --> 0:21:21.240 Then you've got the black box problem. This is when 0:21:21.240 --> 0:21:23.840 it's difficult or impossible for someone to see how a 0:21:23.840 --> 0:21:28.560 particular piece of technology works, and sometimes you have firmware 0:21:28.600 --> 0:21:31.879 there that prevents you from even doing any work unless 0:21:31.920 --> 0:21:38.840 you first run certain diagnostic tests and processes with official equipment. 0:21:39.440 --> 0:21:41.600 Now we've definitely seen this sort of thing in the 0:21:41.640 --> 0:21:44.719 automotive world, where it's much harder to work on modern 0:21:44.800 --> 0:21:48.120 vehicles compared to cars from just a few decades ago. 0:21:48.640 --> 0:21:51.919 While working on cars does require skill and knowledge, the 0:21:52.000 --> 0:21:54.760 barrier to entry was lower in the past. I mean, 0:21:54.760 --> 0:21:56.720 you still had to learn how to do it, but 0:21:56.960 --> 0:22:01.399 car companies weren't making it harder to do this. However, 0:22:01.520 --> 0:22:05.119 these days, with computerized systems and specialized service codes, it 0:22:05.160 --> 0:22:07.160 could be really hard to figure out what's even going 0:22:07.240 --> 0:22:10.920 on with a vehicle and even harder to conduct repairs. 0:22:11.359 --> 0:22:14.040 And the same is true with farm equipment. So farmers 0:22:14.040 --> 0:22:17.600 have become frustrated when trying to get hold of software 0:22:17.600 --> 0:22:20.720 and firmware from companies like John Deer in order to 0:22:20.800 --> 0:22:25.120 diagnose and repair problems with complicated equipment. Because of that, 0:22:25.440 --> 0:22:28.879 these farmers are starting to turn to pirate firmware for 0:22:28.920 --> 0:22:30.760 their equipment in order to be able to make those 0:22:30.800 --> 0:22:34.600 repairs themselves or to have independent mechanics do it for them, 0:22:34.960 --> 0:22:38.120 and companies like John Deer don't want that to happen. 0:22:38.520 --> 0:22:42.199 These companies purposefully restrict access to stuff like firmware and 0:22:42.240 --> 0:22:47.440 software because that represents another potential revenue source, which is 0:22:47.560 --> 0:22:50.679 repair and maintenance. See, if you make a tractor, you 0:22:50.720 --> 0:22:54.159 can only sell that one tractor once. But if you 0:22:54.280 --> 0:22:58.679 also offer repair and maintenance services, while you can keep 0:22:58.760 --> 0:23:02.280 making revenue off of at initial sale for as long 0:23:02.320 --> 0:23:04.399 as possible, I mean, especially if you make it really 0:23:04.440 --> 0:23:07.160 hard for anyone else to do those kinds of repairs 0:23:07.200 --> 0:23:10.920 and maintenance. And that's the heart of the problem. Farmers 0:23:10.960 --> 0:23:13.400 who understandably want to get the most out of their 0:23:13.400 --> 0:23:16.080 equipment would like to be able to do that kind 0:23:16.080 --> 0:23:19.240 of maintenance and repair themselves or on their own terms, 0:23:19.480 --> 0:23:22.800 and thus limit their costs. And this gets right into 0:23:22.840 --> 0:23:25.320 that right to repair issue, which is something we're seeing 0:23:25.359 --> 0:23:28.000 in different parts of the world as groups of consumers 0:23:28.320 --> 0:23:32.520 protest the business model that relies not just on sales 0:23:32.560 --> 0:23:37.360 of hardware, but having a whole sales department that's related 0:23:37.400 --> 0:23:41.240 to fixing and maintaining that hardware. And I should add that, 0:23:41.320 --> 0:23:44.280 according to the Drive, most of the farmers aren't actually 0:23:44.320 --> 0:23:47.960 happy about having to resort to pirate ID software and firmware. 0:23:48.080 --> 0:23:50.760 They would much prefer to go through official channels. They 0:23:50.760 --> 0:23:53.840 would prefer to purchase a license from John Deer directly. 0:23:53.880 --> 0:23:56.280 It's just that's not on the table. I see a 0:23:56.280 --> 0:24:01.080 lot of parallels with other industries actually, including the music industry. Historically, 0:24:01.320 --> 0:24:04.760 if companies make some aspect of their products difficult to access, 0:24:05.200 --> 0:24:08.600 people will find workarounds, and so it's generally just a 0:24:08.640 --> 0:24:11.040 bad idea to even go that route in the first place. 0:24:11.640 --> 0:24:15.000 On Wednesday, Facebook announced it would reduce the amount of 0:24:15.040 --> 0:24:18.879 political subject matter popping up in people's news feeds as 0:24:18.920 --> 0:24:23.119 part of an effort to reduce quote inflammatory content end quote. 0:24:23.160 --> 0:24:27.479 According to Tech Explore, it's going to start small, aiming 0:24:27.520 --> 0:24:31.840 at a sliver of users in places like Canada, Indonesia, 0:24:32.080 --> 0:24:34.840 and Brazil, before trying it out in the United States. 0:24:35.600 --> 0:24:40.119 Asta Gupta, the Product manager director at Facebook, said during 0:24:40.280 --> 0:24:43.240 these initial tests, will explore a variety of ways to 0:24:43.359 --> 0:24:47.320 rank political content in people's feeds using different signals, and 0:24:47.359 --> 0:24:51.200 then decide on the approaches will use going forward. Now. 0:24:51.240 --> 0:24:55.480 Apparently that means Facebook will no longer recommend politics themed 0:24:55.520 --> 0:24:59.600