WEBVTT Kind: captions Language: en 00:00:00.080 --> 00:00:01.773 I want to show you something. 00:00:01.773 --> 00:00:03.656 It’s this radio. 00:00:03.656 --> 00:00:07.150 It’s pretty cool looking, right? I use it pretty often. 00:00:07.150 --> 00:00:10.097 Now, here’s a fun little challenge for you: 00:00:10.097 --> 00:00:15.204 Tell me the model number, what year this radio was made, what vacuum tubes are inside of it, 00:00:15.204 --> 00:00:20.119 and find its schematic diagram for me so I can order new capacitors for it. 00:00:20.119 --> 00:00:22.071 Go on. 00:00:22.071 --> 00:00:26.830 If that seems like a difficult task - it’s very much not. 00:00:26.830 --> 00:00:36.569 So long as you can see this image, you have everything you need on your screen right now to get all that information and much more in less than a minute. 00:00:36.569 --> 00:00:40.417 That is, if you know where and how to look. 00:00:40.417 --> 00:00:44.068 Now, I’m gonna show you where and how to look in just a moment 00:00:44.068 --> 00:00:54.615 but first I should note that this is no doubt going to be the most crotchety,  old-man-yells-at-cloud video I’ve ever released and I won’t hide from that. 00:00:54.615 --> 00:00:59.661 Another thing I won’t hide from is the fact that this is largely an opinion piece. 00:00:59.661 --> 00:01:05.857 If that doesn’t sound like your jam, please feel free to watch any of the other videos on this here website. 00:01:05.857 --> 00:01:12.192 But if you’ll indulge me - that’s actually a large part of what this video is about. 00:01:12.192 --> 00:01:20.412 It’s about the modern internet  and how I think it’s caused a lot of folks to stop looking for things. 00:01:20.412 --> 00:01:25.034 I’ll explain what I mean later  but first let’s figure out what that radio is. 00:01:25.034 --> 00:01:27.536 So - all we’ve got is a picture. 00:01:27.536 --> 00:01:33.064 But I promise you don’t need any fancy image recognition tools to find this radio. 00:01:33.064 --> 00:01:36.676 All it takes is noticing a few of its details. 00:01:36.676 --> 00:01:48.555 The radio is branded “Silvertone” and the tuning dial has two sets of numbers,  one labeled “standard broadcast” and the other labeled “ffsomething modulation.” 00:01:48.555 --> 00:01:53.902 The first word is partially obscured by the dial’s pointer but it’s likely frequency. 00:01:53.902 --> 00:02:00.931 You probably know that frequency modulation is what FM is short for, and this is clearly an old radio. 00:02:00.931 --> 00:02:05.870 So it’s some kind of old FM radio made by Silvertone. 00:02:05.870 --> 00:02:08.153 How can we identify which one? 00:02:08.153 --> 00:02:11.867 Well, let’s see if we can pick it out from a Google Image search. 00:02:11.867 --> 00:02:18.686 A lot of people would call this a vintage radio  so we’ll look for “vintage silvertone FM radio” 00:02:18.686 --> 00:02:24.779 There are lots of pictures of cool old radios  here but look at that - it’s the very same one. 00:02:24.779 --> 00:02:31.231 And with a click on it we’ll see the image description which tells us this is a Silvertone Model 18. 00:02:31.231 --> 00:02:35.474 Wonderful! Now that we have its model number, we can get more details. 00:02:35.474 --> 00:02:42.518 I wanted the radio’s schematic diagram, so let’s do a web  search for "silvertone model 18 schematic." 00:02:42.518 --> 00:02:48.414 The first link is to a website called radiomuseum.org - that looks pretty promising! 00:02:48.414 --> 00:02:57.129 Click through and there’s another picture of the same radio, so we have confirmation that this is indeed a Silvertone Model 18. 00:02:57.129 --> 00:03:02.170 And right on this page we find it was produced between 1950 and 1954, 00:03:02.170 --> 00:03:05.117 plus we have a list of its 8 vacuum tubes. 00:03:05.117 --> 00:03:08.450 And of course there is the schematic available for download, 00:03:08.520 --> 00:03:11.306 though it can also be found on other websites. 00:03:11.306 --> 00:03:19.811 Keeping this information available online is a serious undertaking and I applaud the people who dedicate themselves to maintaining active databases like this 00:03:19.811 --> 00:03:25.023 (as well as those doing the work to back it all up. Especially right now). 00:03:25.023 --> 00:03:27.818 The point of that little exercise was… 00:03:27.818 --> 00:03:29.827 well, exercise! 00:03:29.827 --> 00:03:34.133 What we just did is nothing short of a human superpower. 00:03:34.133 --> 00:03:43.040 From just an image of an old radio you can find out a lot of information using simple search tools and your own observations. 00:03:43.040 --> 00:03:46.443 And if you’re just a little curious, you can keep going. 00:03:46.443 --> 00:03:50.444 Did you notice that the manufacturer was Sears, Roebuck & Co.? 00:03:50.444 --> 00:03:52.044 Yeah, that Sears. 00:03:52.044 --> 00:03:56.303 The one that built the tower I see everyday, weather permitting. 00:03:56.303 --> 00:04:03.735 They made a lot of stuff in-house back in the day, and Silvertone was their brand of electronics and also musical instruments. 00:04:03.735 --> 00:04:07.662 This brown mid-century beauty was one of their radio receivers. 00:04:07.662 --> 00:04:16.446 And if you don’t know anything about antique radios and vacuum tubes and why these old things usually need new capacitors, you can also find all that out! 00:04:16.446 --> 00:04:21.704 A search for “replacing capacitors antique radios” brings you to this lovely website, 00:04:21.704 --> 00:04:26.218 a much more useful resource than whatever AI nonsense Google is synthesizing 00:04:26.218 --> 00:04:29.720 because that’s apparently what search engines are supposed to be now for some reason. 00:04:30.600 --> 00:04:33.120 Ope, there’s the old man yelling at clouds. 00:04:33.120 --> 00:04:34.773 Told ya. 00:04:34.773 --> 00:04:42.137 I am sure a lot of you knew how useful the information in that image was and how you could use it to find the radio, 00:04:42.137 --> 00:04:48.154 and to those of you who did this video probably seems incredibly unremarkable so far. 00:04:48.154 --> 00:04:59.640 But I believe quite strongly now that this is a skill which as time marches on people are forgetting they have and thus don’t think to use even when it could help them.   00:05:00.600 --> 00:05:03.283 I want to reiterate the language I just used there: 00:05:03.283 --> 00:05:10.272 I am not saying people don’t know how to do this - anyone can do this, and hopefully you learned how in school. 00:05:10.272 --> 00:05:12.648 It’s pretty basic research. 00:05:12.648 --> 00:05:23.633 What I am saying is that it appears as though an increasing number of people seem to operate in the world without realizing these are things they can do themselves. 00:05:23.791 --> 00:05:26.324 And that really concerns me. 00:05:26.324 --> 00:05:30.687 Now I’ve spent enough time on forums to know about “let me google that for you” - 00:05:30.687 --> 00:05:35.051 a snarky response to people who ask easy-to-answer-with-a-web search questions. 00:05:35.051 --> 00:05:37.791 I was even on the receiving end of that once. 00:05:37.791 --> 00:05:41.105 But that’s not quite what I want to talk about. 00:05:41.105 --> 00:05:47.948 I want to talk about how we decide what we want to see, watch, and do on the internet. 00:05:47.948 --> 00:05:57.452 Because, well, I’m not sure we realize just how infrequently we’re actually deciding for ourselves these days. 00:05:57.452 --> 00:06:04.113 That’s right, this is video about the problems of recommendation algorithms on social media and the internet at large. 00:06:04.113 --> 00:06:11.083 But I’m not gonna be focusing much on what exactly those algorithms do. I think we all know by now. 00:06:11.083 --> 00:06:16.765 Instead I’m going to be focusing on something which feels new and troubling. 00:06:16.765 --> 00:06:27.066 I’m starting to see evidence that an increasing number of folks actually prefer to let a computer program decide what they will see when they log on, 00:06:27.066 --> 00:06:30.041 even when they know they have alternatives. 00:06:30.041 --> 00:06:34.283 Since this feels like a new phenomenon, I felt it needed a name. 00:06:34.283 --> 00:06:38.399 I’ve chosen to call it “algorithmic complacency.” 00:06:38.399 --> 00:06:43.842 Now, I recognize that I am stepping into a discussion that many many people have had 00:06:43.842 --> 00:06:49.751 and so I don’t want to claim that this is an original thought or even an original term. 00:06:49.751 --> 00:06:57.950 But as I’ve worked to define and articulate it, I’ve come  to believe that it’s a serious problem that needs our immediate attention. 00:06:57.950 --> 00:07:06.105 Succumbing to algorithmic complacency means you’re surrendering your own agency in ways you may not realize. 00:07:06.105 --> 00:07:13.727 And as the internet and real life continue to blend together, that can end very badly. 00:07:13.727 --> 00:07:20.437 Since I believe there’s significant gravity to this topic, I want to present a convincing argument to you. 00:07:20.437 --> 00:07:24.553 And that requires that we first look backward in time. 00:07:24.553 --> 00:07:33.664 Think for a moment about what your experience on the internet is like these days and, if you’re old enough, how it differs from a couple of decades ago. 00:07:33.664 --> 00:07:39.068 The internet used to only exist through a web browser on a desktop computer. 00:07:39.068 --> 00:07:41.438 Maybe a laptop if you’re fancy. 00:07:41.438 --> 00:07:47.192 [dial tone and dialing in background] And your computer had to pretend to be a telephone and shriek at other computers through a phone line 00:07:47.192 --> 00:07:51.496 just to transmit and receive data at blazing slow speeds. 00:07:51.496 --> 00:07:53.629 It was a dark time. 00:07:53.629 --> 00:07:59.658 Back then, Google was just a happy little search  engine which helped you find websites, 00:07:59.658 --> 00:08:06.082 and when you found a cool website which you liked you’d use your web browser to bookmark that website. 00:08:06.082 --> 00:08:10.213 That would make sure you could get back to it later without having to search for it again. 00:08:10.213 --> 00:08:12.598 Like writing a note to yourself. 00:08:12.598 --> 00:08:21.785 In other words, the internet was still very manual and you were in charge of navigating it and curating your own experience with it. 00:08:21.785 --> 00:08:28.559 That also meant the internet didn’t do anything until you asked it to do something. 00:08:28.559 --> 00:08:35.588 You might set your browser’s homepage to your local newspaper’s website so you could get a bit of a news update each time you logged on 00:08:35.588 --> 00:08:40.068 but other than that, information didn’t just come to you. 00:08:40.068 --> 00:08:45.317 Finding news, products, and information on the internet was entirely up to you. 00:08:45.317 --> 00:08:53.157 You had the world’s information at your fingertips, but you needed to use your fingertips and your brain to get to it. 00:08:53.157 --> 00:08:59.710 And it was also up to you to gauge the  trustworthiness and reliability of the information you found. 00:08:59.710 --> 00:09:04.516 If you’re over the age of 30 you probably remember what this was like. 00:09:04.516 --> 00:09:09.865 Overtime, though, things have steadily become a lot more automated. 00:09:09.865 --> 00:09:13.727 And also a lot more in-your-face. 00:09:13.727 --> 00:09:19.451 Many people these days experience the internet  primarily through their smartphones and mobile apps 00:09:19.451 --> 00:09:25.638 in a neatly-packaged ecosystem created and curated by giant tech corporations. 00:09:25.638 --> 00:09:33.349 Even though those apps are often just bespoke web browsers that only take you to a very specific website and keep your traffic inside a walled garden 00:09:33.349 --> 00:09:35.841 while also collecting lots of data about where you go and what you do, 00:09:35.841 --> 00:09:43.519 they still represent a radical shift in how we use and experience the internet as a resource. 00:09:43.519 --> 00:09:45.913 Platforms became the new name of the game. 00:09:45.913 --> 00:09:53.252 We’re not surfing the web and looking for cool and useful things anymore, we’re hanging out on platforms. 00:09:53.252 --> 00:09:56.854 Like, well, this one. 00:09:56.854 --> 00:10:04.466 We’re so used to this now that we imagine apps less as a piece of software which enables connection to people and information 00:10:04.466 --> 00:10:07.947 and more of a place where we spend time. 00:10:07.947 --> 00:10:16.084 Internet historians remind us that this is not our first rodeo. We escaped the walled garden that was AOL, after all. 00:10:16.084 --> 00:10:22.882 But I think most would agree  that today’s internet is distinctly different and intense. 00:10:22.882 --> 00:10:30.239 It has become so integral to our lives  that it’s shaping how we view the world and how we operate within it. 00:10:30.619 --> 00:10:37.866 And most troublingly, we are largely no longer in control of what we see. 00:10:37.866 --> 00:10:44.079 Recommendation algorithms end up putting content in front of our eyes using methods almost nobody really understands 00:10:44.079 --> 00:10:47.496 (but probably have something to do with maximizing revenues) and, 00:10:47.496 --> 00:10:52.119 well, I think it’s breaking our brains. 00:10:52.119 --> 00:10:58.377 When you have that finely-tuned, algorithmically-tailored firehose of information just coming at you like that, 00:10:58.377 --> 00:11:03.082 you might feel like you’re having a good time and learning some interesting things, 00:11:03.082 --> 00:11:08.916 but you’re not necessarily directing your own experience, are you? 00:11:08.953 --> 00:11:14.575 Is your train of thought really your own when the next swipe might derail it? 00:11:14.575 --> 00:11:18.331 Now, I am by no means the first person to ask that question. 00:11:18.331 --> 00:11:26.376 And, full disclosure, I make my living finding information, packaging it into videos which I hope are entertaining and insightful, 00:11:26.376 --> 00:11:32.185 and putting them online for people to hopefully stumble across when the YouTube algorithm recommends it to them. 00:11:32.185 --> 00:11:40.148 So I recognize the awkwardness of this particular person talking about this particular thing on this particular platform. 00:11:40.148 --> 00:11:44.761 But here’s what I think might be new, or at least under-discussed: 00:11:44.761 --> 00:11:52.419 I am seeing mounting evidence that an increasing number of people are so used to algorithmically-generated feeds 00:11:52.419 --> 00:11:59.971 that they no longer care to have a self-directed experience that they are in control of. 00:11:59.971 --> 00:12:10.705 The more time I spend interacting with folks online, the more it feels like large swaths of people have forgotten to exercise their own agency. 00:12:10.705 --> 00:12:14.859 That is what I mean by algorithmic complacency. 00:12:14.859 --> 00:12:23.264 More and more people don’t seem to know or care how to view the world without a computer algorithm guiding what they see. 00:12:23.264 --> 00:12:27.739 That’s a pretty bold claim I just made, and that’s gonna require evidence. 00:12:27.739 --> 00:12:34.467 I will say up front that my evidence is not scientific, so set your expectations accordingly. 00:12:34.467 --> 00:12:39.230 I do have data for one particular thing but I want to talk about that second. 00:12:39.230 --> 00:12:46.661 First, I want to talk about my experiences on new forms of social media and how that has informed this argument. 00:12:46.661 --> 00:12:52.020 I have long given up on “traditional?” social media 00:12:52.020 --> 00:12:57.520 but I’ve been indulging in some of the alternatives like Mastodon and, lately, Bluesky. 00:12:57.520 --> 00:13:00.384 I’m not trying to sell you on using them, to be clear, 00:13:00.384 --> 00:13:08.194 but both of those platforms are far more manual than anything the likes of Meta or Twitter might have spun up. 00:13:08.194 --> 00:13:11.449 I think this is great and quite refreshing! 00:13:11.449 --> 00:13:18.044 I follow accounts that I’m interested in, and I never have to worry about whether an algorithm won’t show me their posts. 00:13:18.044 --> 00:13:25.031 I still discover new accounts all the time, but through  stuff the ones I’m following have shared. 00:13:25.031 --> 00:13:30.722 That’s a much more human experience than letting an algorithm decide to put stuff in front of my eyeballs 00:13:30.722 --> 00:13:35.727 because it will keep me engaged and increase user time on platform. 00:13:35.727 --> 00:13:39.461 It’s also a much more social experience. 00:13:39.461 --> 00:13:47.545 Yet to a lot of new users that are migrating to these platforms, the need to curate your own experience is very frustrating. 00:13:47.545 --> 00:13:55.197 Several of whom have told me this directly and even said they’d prefer not to have to put any work into social media. 00:13:55.197 --> 00:13:58.894 Now, I must admit that sentiment alone concerns me. 00:13:58.894 --> 00:14:03.011 I’m one of those weirdos who think the most rewarding things in life take effort, 00:14:03.011 --> 00:14:05.042 at least outside November, 00:14:05.042 --> 00:14:13.225 so I would not expect to just walk into a new place and have a good time without doing a little ol’ fashioned exploring. 00:14:13.225 --> 00:14:16.296 However, I can sympathize with it. 00:14:16.296 --> 00:14:23.116 For one, I have to recognize that I’m a popular YouTuber and I can just show up somewhere, say “hey guys” 00:14:23.116 --> 00:14:31.118 and expect a lot of people to start following me very quickly, and that helped re-establish connections I formed in other places. 00:14:31.118 --> 00:14:36.645 That’s a privilege I have and skews my perspective a lot, which is only fair that I acknowledge. 00:14:36.645 --> 00:14:41.229 My experience online is not normal. 00:14:41.229 --> 00:14:48.161 But observing Bluesky grow from an extremely niche platform to a still-niche-but-quickly-growing one 00:14:48.161 --> 00:14:53.330 has presented a very interesting case study which I feel compelled to share. 00:14:53.330 --> 00:14:58.133 Bluesky, for those who don’t know, allows for the creation of custom feeds. 00:14:58.133 --> 00:15:01.046 That’s one of its central features which is really cool! 00:15:01.046 --> 00:15:06.731 But baked-into it as a default when you create an account are two feeds with different purposes: 00:15:06.731 --> 00:15:09.596 Following and Discover. 00:15:09.596 --> 00:15:15.238 The Following feed is a reverse-chronological feed  of the posts from accounts you follow. 00:15:15.238 --> 00:15:18.382 Exactly how Twitter used to work. 00:15:18.382 --> 00:15:24.718 But the Discover feed is algorithmic, like those “for you” pages everyone keeps going on about. 00:15:24.718 --> 00:15:27.535 On Bluesky it’s a pretty basic algorithm, 00:15:27.535 --> 00:15:36.040 but its job is to pick posts from accounts across the platform and essentially promote them so people can find new accounts to follow. 00:15:36.040 --> 00:15:39.977 It’s by no means a bad idea to have that feature! 00:15:39.977 --> 00:15:44.440 But as someone with a pretty large following on Bluesky now, 00:15:44.440 --> 00:15:53.841 I can tell the instant a post of mine ends up on the discover feed because the replies get real weird real fast. 00:15:53.841 --> 00:16:02.794 What was previously a post with a nice discussion going on underneath between myself and various people who know who I am and what I mean when I say words 00:16:02.794 --> 00:16:07.512 becomes littered with strange, out-of-context, often antagonistic replies 00:16:07.512 --> 00:16:16.133 as if the only possible response to seeing a post of any kind online is to loudly perform a challenge against it. 00:16:16.133 --> 00:16:20.640 Now, some of these are bots, a known problem on Bluesky. 00:16:20.640 --> 00:16:29.097 But through forming a habit where I check profiles before replying to see if they show signs of being a bot, it’s clear that a lot of them are not. 00:16:29.097 --> 00:16:39.893 They’re people who just start talking with absolutely no idea who they’re talking to and with no desire to figure out the context of the discussion before they write their reply. 00:16:40.593 --> 00:16:46.669 This may feel like I’m veering off track a bit or just complaining about people, which is fun, 00:16:46.669 --> 00:16:55.193 but this is less about calling out the behavior of individuals and more about recognizing the incentives which promote that behavior. 00:16:55.193 --> 00:17:03.535 Algorithmic feeds on social media are unfortunately quite good at fostering something known as context collapse. 00:17:03.535 --> 00:17:12.211 To understand this, imagine you’re dining in a restaurant and you’re close enough to a table of people to hear snippets of their conversation. 00:17:12.211 --> 00:17:15.429 You don’t know who any of the people at that table are, 00:17:15.429 --> 00:17:23.318 but if you manage to overhear them talk about something you’re really interested in you might feel tempted to join their conversation. 00:17:23.318 --> 00:17:30.185 But in the context of a restaurant setting, that’s considered very rude so it rarely ever happens. 00:17:30.185 --> 00:17:39.036 On social media, though, the same kinds of quasi-private conversations between parties who know each other are happening all the time, 00:17:39.036 --> 00:17:47.050 but since the platform is just one big space and it might decide to put that conversation in front of random people, 00:17:47.050 --> 00:17:53.155 that social boundary of etiquette which is normally respected is just not there. 00:17:53.155 --> 00:17:56.380 And lots of conflicts happen as a result. 00:17:56.380 --> 00:18:00.184 A really common one you might accidentally step into on social media 00:18:00.184 --> 00:18:06.324 happens when you stumble across a conversation among friends making sarcastic jokes with each other 00:18:06.324 --> 00:18:13.144 but since you don’t know who those people are you don’t have the context you need to recognize they’re joking. 00:18:13.144 --> 00:18:20.751 And so, if you reply with a serious critique, well that’s a social misfire which some will react poorly to. 00:18:20.751 --> 00:18:23.968 And that’s a pretty mild form of context collapse. 00:18:23.968 --> 00:18:30.112 It can be much, much worse when people want to discuss things like politics. 00:18:30.112 --> 00:18:36.187 And unless we realize recommendation algorithms  are what’s fostering these reactionary conflicts, 00:18:36.187 --> 00:18:41.200 they’re going to continue so long as we  use platforms in the ways that we do. 00:18:41.200 --> 00:18:50.080 It's for all these reasons that I believe algorithmic complacency is creating a crisis of both curiosity and human connection.   00:18:50.080 --> 00:18:55.382 I would even go so far as to say it’s  fostering lots of other disturbing things, too. 00:18:55.382 --> 00:18:59.138 Ever notice how a lot of folks these days need to have a simple 00:18:59.138 --> 00:19:00.154 “good or bad,” 00:19:00.154 --> 00:19:01.405 “black or white,” 00:19:01.405 --> 00:19:05.521 “best or worst” understanding of a topic or issue? 00:19:05.521 --> 00:19:11.655 It seems to me like algorithms which promote content through a simple lens of positive or negative engagement 00:19:11.655 --> 00:19:16.348 would reinforce those binaries and contribute to polarization. 00:19:16.348 --> 00:19:21.838 And as people learn about new products through the slot machine of social media feeds, 00:19:21.838 --> 00:19:28.077 they can develop a learned helplessness where they will wait to be sold on a solution for their problems 00:19:28.077 --> 00:19:38.802 rather than be introspective and explore what their problems actually are and how they might be able to come up with their own solutions which don’t cost any money. 00:19:38.802 --> 00:19:45.554 Introspection will reveal a lot of the problems you think you have are being put in your head by influencers - 00:19:45.554 --> 00:19:50.361 you weren’t unhappy until they told you you should be. 00:19:50.361 --> 00:19:55.354 And, well, I can think of lots of other stuff which has disturbed me for quite a while 00:19:55.354 --> 00:20:02.670 but is now past the point I can ignore as a quirky consequence of connecting large numbers of humans together. 00:20:02.670 --> 00:20:08.215 Social media algorithms don’t nurture human connection - they exploit it. 00:20:08.215 --> 00:20:14.104 And we’re so used to this reality now that I’m not sure many of us care to get off this train. 00:20:14.104 --> 00:20:16.312 But I think we should. 00:20:16.312 --> 00:20:19.417 So now, let’s talk about YouTube! 00:20:19.417 --> 00:20:26.618 I’m sure a good number of you have thought “this is pretty rich comin’ from a guy who makes his living on a platform which does all these things.” 00:20:26.618 --> 00:20:30.345 Well, there’s a funny thing about YouTube. 00:20:30.345 --> 00:20:34.783 Its recommendation algorithm is entirely optional. 00:20:34.783 --> 00:20:36.747 You know that, right? 00:20:36.747 --> 00:20:40.699 You… you do know that, right? 00:20:40.699 --> 00:20:49.099 There’s this feature on this website that has been here pretty much  since it became a thing and it’s called the subscriptions feed. 00:20:49.099 --> 00:20:54.044 It’s not hiding, it’s at the bottom of the mobile app and on the sidebar of the desktop site. 00:20:54.044 --> 00:20:58.600 In fact it's got its own URL and if you’re feeling old-school you can bookmark it! 00:20:59.240 --> 00:21:04.848 This is a completely manually-curated feed which has nothing in it but the videos 00:21:04.848 --> 00:21:11.008 (and shorts, for better and worse) from the creators you have chosen to subscribe to. 00:21:11.008 --> 00:21:14.374 That means it’s entirely in your control. 00:21:14.374 --> 00:21:22.078 This feed has been in plain sight the whole time and, here’s why I am using the term algorithmic complacency, 00:21:22.078 --> 00:21:24.485 nobody cares to use it. 00:21:24.485 --> 00:21:28.532 And that I can back up with hard data. 00:21:28.532 --> 00:21:35.440 YouTube gives us a shocking amount of information regarding audience metrics and in 2024, 00:21:35.440 --> 00:21:42.504 less than three percent of this channel’s views came from the subscriptions feed. 00:21:42.504 --> 00:21:44.966 Almost nobody is using this feature 00:21:44.966 --> 00:21:54.169 and yet it’s the most reliable way to keep track of the things you have explicitly decided you want to watch through hitting the subscribe button. 00:21:54.169 --> 00:22:03.220 The fact that it’s easy to get to, has existed in plain sight almost since this platform was born, yet almost nobody is using it anymore is… 00:22:03.220 --> 00:22:04.722 puzzling. 00:22:04.722 --> 00:22:07.821 I want to stress my use of the word puzzling. 00:22:07.821 --> 00:22:12.567 It’s not my intent to be judgmental here and I’m sorry if it has sounded like that. 00:22:12.567 --> 00:22:19.918 Different people use websites differently and the main “home” feed on YouTube, which functions like a “for-you” page, 00:22:19.918 --> 00:22:24.466 usually does a good job of surfacing new videos from creators you like. 00:22:24.466 --> 00:22:31.999 Subscribing to a channel is a signal to the recommendation algorithm  to boost that channel’s videos in your home feed. 00:22:31.999 --> 00:22:38.151 And it’s not like that algorithm has no value - there are some creators I’m not even subscribed to 00:22:38.151 --> 00:22:47.527 yet I see most of their new videos since the algorithm has figured out I keep watching them so I must want to see the new ones, and it puts them in my home feed. 00:22:47.527 --> 00:22:57.152 Through that feed I’ve also stumbled across countless new channels and have been reminded of videos I watched years ago and will happily watch again. 00:22:57.152 --> 00:23:05.230 Plus of course, I have found so much cool stuff through the recommended videos that appear alongside whatever I happen to be watching. 00:23:05.230 --> 00:23:07.880 I’m glad YouTube does that. 00:23:07.880 --> 00:23:10.395 Wait, nuance? 00:23:10.395 --> 00:23:11.688 On the internet? 00:23:11.688 --> 00:23:13.059 That’s illegal! 00:23:13.059 --> 00:23:20.242 In the interest of time I’ve removed a big section on the flaws with the subs feed and how I think YouTube should address them 00:23:20.242 --> 00:23:23.901 (mainly - please let us remove shorts if we aren’t interested in them 00:23:23.901 --> 00:23:29.979 and please lets us organize the subs feed a little bit so creators who upload a lot don’t end up cluttering it. 00:23:29.979 --> 00:23:38.457 Maybe condense their daily activity into a single tile which we can then click on and expand. Just a thought). 00:23:38.457 --> 00:23:40.904 But the reason I wanted to talk about it 00:23:40.904 --> 00:23:51.024 is that it feels indicative of a growing disinclination to grab the reins of the internet and be the person steering your own experience. 00:23:51.024 --> 00:23:52.705 That’s really what bothers me. 00:23:52.705 --> 00:24:01.687 There has to be a balance between cool stuff you  stumble upon and stuff that you’re actually interested in and matters to you. 00:24:01.687 --> 00:24:05.617 Algorithmic complacency, if not noticed and acted upon, 00:24:05.617 --> 00:24:12.243 means you’re allowing other people who are not you  to decide what matters to you. 00:24:12.243 --> 00:24:16.183 I should not have to spell out why that’s dangerous so I won’t. 00:24:16.183 --> 00:24:28.247 But I will spell out that it’s very easy for people who wish to weaponize this reality to craft a narrative which is not overtly obvious and so might slip past your defenses. 00:24:28.247 --> 00:24:34.317 If you can reduce the presence of algorithmically-curated feeds in your life, you’ll be less susceptible to that. 00:24:34.317 --> 00:24:41.149 And if you build up your own network of people and resources you trust,  you’ll know when you’re being bullshitted. 00:24:41.149 --> 00:24:47.019 And in case you haven’t noticed, people are trying to bullshit us all the time now! 00:24:47.019 --> 00:24:49.841 And algorithmic feeds make this worse. 00:24:49.841 --> 00:24:59.652 They don’t just exist on YouTube and social media, they exist in popular news aggregator apps and that’s meant a lot of really stupid articles keep floating around 00:24:59.652 --> 00:25:06.471 because all that matters to many institutions which make news these days is clicks and ad money. 00:25:06.471 --> 00:25:09.365 Look at this stupid thing which ended up in the Google News app. 00:25:09.650 --> 00:25:15.131 "The end of Walmart and Target in the US - A new retailer appears and is much cheaper." 00:25:16.428 --> 00:25:18.516 I know what an Aldi looks like. 00:25:18.516 --> 00:25:21.228 That’s a picture of the inside of an Aldi. 00:25:21.228 --> 00:25:30.104 And, uh, if you’ve somehow not heard of Aldi or stepped inside one that doesn’t mean Aldi is a new retailer which just appeared. 00:25:30.104 --> 00:25:36.732 This article is a waste of time for the vast majority of people who might be tricked into clicking on it. 00:25:36.732 --> 00:25:46.109 You may have noticed that the publisher of that article was El Diario 24 which, as far as I can tell, might actually be a fake news source. 00:25:46.109 --> 00:25:48.493 How did it get in Google News? 00:25:48.493 --> 00:25:50.624 That’s a great question for Google. 00:25:50.624 --> 00:25:58.568 But even mainstream publications have gone wildly off the rails as they chase metrics rather than the truth. 00:25:58.568 --> 00:26:03.127 The New York Times of 2025 is publishing opinion pieces 00:26:03.127 --> 00:26:14.314 where bloviating morons go through the fun little exercise of what the political ramifications of turning Canada into the 51st state would be for Democrats. 00:26:14.314 --> 00:26:19.633 Folks, since they can’t say this in plain language for some reason, I guess it’s up to me. 00:26:19.633 --> 00:26:22.791 Canada is a sovereign nation. 00:26:23.550 --> 00:26:27.517 It’s a foreign country which has the right to self-determination. 00:26:28.054 --> 00:26:33.037 The United States cannot simply turn Canada into a US state - 00:26:33.037 --> 00:26:36.431 Canadians have clearly indicated they do not want that, 00:26:36.431 --> 00:26:42.844 which means for us to force the issue would be to declare war with Canada and invade. 00:26:43.603 --> 00:26:46.537 No sane person should want that 00:26:46.537 --> 00:26:55.593 and it’s a shameful embarrassment that the New York Times would even entertain this as a possibility and legitimize the awful, inevitably bloody idea. 00:26:56.225 --> 00:26:58.240 How on Earth did we get here? 00:26:59.560 --> 00:27:01.833 I can tell you how I think we got here: 00:27:01.833 --> 00:27:09.337 big news publications have become just as dependent on algorithms to find their readers as their readers are to find the news. 00:27:09.337 --> 00:27:13.997 Which means they’re more concerned with being enticing than being honest. 00:27:13.997 --> 00:27:15.605 Which is a damn shame. 00:27:16.301 --> 00:27:18.151 OK, reel it in. 00:27:18.151 --> 00:27:19.331 [breathes] 00:27:19.331 --> 00:27:21.254 Reel it in. 00:27:21.254 --> 00:27:28.186 I’m about to wrap up this video but before I do, I want to explore one more related but different angle. 00:27:28.186 --> 00:27:35.329 I struggled with whether I wanted to call this phenomenon algorithmic complacency or automation complacency. 00:27:35.329 --> 00:27:40.562 The reason I struggled is that there isn’t a clear distinction between those two things. 00:27:40.562 --> 00:27:49.757 Algorithms are a kind of automation, so you could say everything I’ve been talking about is the result of automatically-curated feeds 00:27:49.757 --> 00:27:52.258 and the video wouldn’t change. 00:27:52.258 --> 00:27:55.685 But automation in itself is not necessarily bad. 00:27:55.685 --> 00:28:02.921 Lots of menial labor tasks have been replaced by automation and this has largely been a great thing. 00:28:02.921 --> 00:28:08.015 The actually-important word in this discussion is curated. 00:28:08.015 --> 00:28:10.960 It’s one thing to automate, say, an elevator. 00:28:10.960 --> 00:28:13.120 Or an inventory system. 00:28:13.120 --> 00:28:15.644 Or a telephone switchboard. 00:28:15.644 --> 00:28:21.878 But it’s a very different thing to automate what information people see. 00:28:21.878 --> 00:28:26.167 There are situations where that sort of automation is necessary. 00:28:26.167 --> 00:28:30.004 There’s an ever-increasing amount  of information being stored online, 00:28:30.004 --> 00:28:38.020 so when you’re looking for something specific amongst that sea of information, keywords by themselves aren't enough. 00:28:38.020 --> 00:28:39.543 The demonstration we did at the beginning 00:28:39.543 --> 00:28:50.150 relied on Google’s search algorithm determining context from the keywords we gave it so it could sort everything it found by relevance to that inferred context. 00:28:50.150 --> 00:28:54.089 And it’s usually really good at that, as we saw. 00:28:54.089 --> 00:28:56.072 But it’s not always. 00:28:56.072 --> 00:29:05.504 I’m sure by now you’ve had the really frustrating experience  of Google latching onto the wrong context and producing a lot of irrelevant results, 00:29:05.504 --> 00:29:11.361 and it can be extremely tedious to refocus the algorithm on the correct context. 00:29:11.361 --> 00:29:16.191 This happens a lot when search queries include a word with many homonyms. 00:29:16.191 --> 00:29:24.693 But we’re rapidly moving away from a paradigm in  which search queries present a list of sources for us to look at, cite and verify, 00:29:24.693 --> 00:29:27.803 and are now being pressured into a new reality 00:29:27.803 --> 00:29:35.190 where large language models synthesize responses to queries which are statistically likely to produce a useful output 00:29:35.190 --> 00:29:39.228 but which do not provide us with sources of verifiable information - 00:29:39.228 --> 00:29:45.292 or at least obfuscate them to the point that many people are not going to check them. 00:29:45.292 --> 00:29:50.448 I don’t think enough of us have put much thought into what that means. 00:29:50.448 --> 00:29:58.161 It means we’re careening towards a future where people just trust computers to do their thinking for them. 00:29:58.161 --> 00:30:03.691 And the thing is, we already know that’s often a bad idea. 00:30:03.691 --> 00:30:08.804 Take, for example, how we navigate the actual physical world. 00:30:08.804 --> 00:30:16.295 If you drive a car, I am certain that by now you’ve used some kind of GPS-navigation app to figure out how to get places. 00:30:16.295 --> 00:30:21.411 I do, too, don’t think I’m about to say "oh we should go back to paper maps." 00:30:21.411 --> 00:30:22.795 Gross. 00:30:22.795 --> 00:30:30.195 But when you use a mapping app to navigate somewhere, how often are you prioritizing the fastest route? 00:30:30.195 --> 00:30:33.298 You probably have that set up as the default, don’t you? 00:30:33.298 --> 00:30:35.175 I do. 00:30:35.175 --> 00:30:41.936 But do you ever question whether that is actually the  most logical way to get somewhere? 00:30:41.936 --> 00:30:49.772 It often isn’t because arrival time is only one of many,  many variables which might be important to you. 00:30:49.772 --> 00:30:53.443 If I ask Google Maps to take me from my home to my office, 00:30:53.443 --> 00:31:02.393 it is going to suggest a route which first requires going the wrong way to then hop on a tollway on which I have to pay a roughly $1 toll 00:31:02.393 --> 00:31:08.658 and have to overshoot my destination to hop on a second expressway and backtrack. 00:31:08.658 --> 00:31:17.501 That suggested route requires an extra 4 miles of travel and  an extra $1 each way all to save me exactly 00:31:17.501 --> 00:31:23.428 one minute over taking much more direct state routes and surface streets. 00:31:23.428 --> 00:31:26.170 If I mindlessly did what Google suggested, 00:31:26.170 --> 00:31:38.085 over a year I’d put an extra 2,000 miles on my car and spend an extra $500 on tolls just to save not even one full work day of time. 00:31:38.085 --> 00:31:44.118 Google is suggesting a terrible route just because it’s one minute faster. 00:31:44.118 --> 00:31:50.414 The only reason I know it’s a terrible route is because I live here and I know what it’s suggesting is asinine, 00:31:50.414 --> 00:31:55.893 but when I don’t know where I’m going I trust it to make the best decision. 00:31:55.893 --> 00:31:59.626 But we don’t necessarily agree on what is best. 00:31:59.626 --> 00:32:01.523 That’s the problem. 00:32:01.600 --> 00:32:06.514 Side-note, the other bad thing about taking the tollway is that when I do, 00:32:06.514 --> 00:32:10.162 I don’t get to see my neighbors and what they’re up to. 00:32:10.162 --> 00:32:13.643 I actually really enjoy driving through town, 00:32:13.643 --> 00:32:16.009 seeing new businesses as they pop up, 00:32:16.009 --> 00:32:18.114 homes getting built and remodeled, 00:32:18.114 --> 00:32:20.649 admiring Christmas lights over the holidays, and 00:32:20.649 --> 00:32:31.590 just seeing people to remind me that other people exist and they live real lives and they are connected to me because they’re my neighbors. 00:32:31.590 --> 00:32:33.361 I’m not the most social person 00:32:33.361 --> 00:32:42.956 but even I don’t like the isolated feeling that a commute on a tollway where there’s nothing to look at but other cars and sound-isolation walls gives me. 00:32:42.956 --> 00:32:51.616 I feel way more connected to my community when I can actually, ya know, see it and the people who define it. 00:32:51.616 --> 00:32:57.278 That’s a preference of mine which I grant but it’s also a priority of mine. 00:32:57.278 --> 00:33:04.286 I am being mindful now to put human connections above technological connections. 00:33:04.286 --> 00:33:08.433 See, what I do here, is I make connections between technologies - 00:33:08.433 --> 00:33:17.054 that way you can learn how they fit together and how best to use them, and maybe you can use one concept in conjunction with another concept to make a third concept! 00:33:17.054 --> 00:33:18.411 That’s what I’m doing here! 00:33:18.411 --> 00:33:22.295 Trying to empower you to make your life better. 00:33:22.295 --> 00:33:29.344 Technologies which make human connection harder or even just more random are, uh, 00:33:29.344 --> 00:33:30.586 bad! 00:33:30.586 --> 00:33:33.799 That I do think is black-and-white true. 00:33:33.799 --> 00:33:42.032 Any piece of technology which gets in between humans who wish to help each other is frustrating at best and exploitative at worst. 00:33:42.032 --> 00:33:50.958 We ought to know by now the real reason those systems get put in place  is so that we need fewer humans in helpful roles. 00:33:50.958 --> 00:33:56.413 And I think there is absolutely no question that this is going to get worse 00:33:56.413 --> 00:34:04.115 until we all start looking inward and begin questioning how we operate in this world and why. 00:34:04.115 --> 00:34:11.228 Silicon valley seems hellbent on creating machines which can do our thinking for us. 00:34:11.228 --> 00:34:14.338 Why should any of us want that? 00:34:14.338 --> 00:34:23.910 I certainly don’t want that - I don’t learn anything unless I do the mental work to create a complete framework of understanding in my mind. 00:34:23.910 --> 00:34:30.199 I don’t talk about things I don’t understand because that’s the fastest way you can make a fool of yourself. 00:34:30.199 --> 00:34:35.432 And it can be dangerous when you have a platform like I do. 00:34:35.432 --> 00:34:41.720 I will never trust a computer program to be able to understand anything in the way a human can, 00:34:41.720 --> 00:34:45.772 nor will I trust it to find information for me. 00:34:45.772 --> 00:34:51.796 If I have to vet everything it’s finding, then I end up doing the same work I would have done myself. 00:34:51.796 --> 00:34:59.481 And if I don’t vet what it’s finding, then what I’m really doing is saying I don’t want to be responsible for what I do. 00:34:59.481 --> 00:35:07.406 It frightens me that even though we’ve all seen the consequences of what a social media recommendation algorithm can do to shape our viewpoints 00:35:07.406 --> 00:35:14.286 that we are somehow falling for the temptation of machines which can offload our thought processes. 00:35:14.286 --> 00:35:17.320 The thing which makes us human. 00:35:17.320 --> 00:35:23.819 If that’s not the purest form of lazy anti-intellectualism, I don’t know what is. 00:35:23.819 --> 00:35:30.545 On that cheery note, let me make sure you know  that I share the same frustration with the AI hype cycle 00:35:30.545 --> 00:35:40.358 as the AI researchers who are actually doing work to create tools to solve real problems, like early detection of cancers from images or blood screens. 00:35:40.358 --> 00:35:44.683 That is valuable research which will undoubtedly save lives. 00:35:44.683 --> 00:35:52.642 It’s beyond frustrating that the only kind of AI that is in the public consciousness at the moment is the one that does some very impressive tricks 00:35:52.642 --> 00:36:00.525 but any honest person will tell you needs intense  supervision because it will hallucinate and produce bad outputs. 00:36:00.525 --> 00:36:02.918 It seems blindingly obvious to me 00:36:02.918 --> 00:36:11.496 that the stakes are way too high to hand over our decision making to a computer which cannot be held responsible for the decisions it makes. 00:36:11.496 --> 00:36:17.534 And that’s, disturbingly, what I think is the real reason for wanting to push this future. 00:36:17.534 --> 00:36:19.932 But you’ve put up with me long enough. 00:36:19.932 --> 00:36:25.005 Thank you for watching and I hope I don’t sound too far off my rocker. 00:36:25.005 --> 00:36:34.987 I made this channel not just to share cool stuff I find but to show all the amazing ways we solved our problems in the past. 00:36:34.987 --> 00:36:39.594 There are invaluable lessons there which we forget at our peril. 00:36:39.594 --> 00:36:44.003 I don’t talk much about computer technology because it doesn’t really interest me. 00:36:44.003 --> 00:36:48.166 And it’s becoming less and less interesting as time goes on. 00:36:48.166 --> 00:36:53.672 Outside of hardcore enthusiasts or people doing research on vast amounts of data, 00:36:53.672 --> 00:36:57.796 let’s be honest: computing is a solved problem. 00:36:57.796 --> 00:37:05.155 I can use my desktop computer from 2017 to make  these videos for you and it would not get in my way at all, 00:37:05.155 --> 00:37:12.352 and the $600 nearly base-spec M2 Mac Mini I bought to dingle around on can do it just as well. 00:37:12.352 --> 00:37:19.920 Video encoding is getting more efficient and we need less bandwidth and storage to send videos like this across the world 00:37:19.920 --> 00:37:24.660 and that’s despite internet connections getting faster and faster. 00:37:24.660 --> 00:37:32.237 So I think it’s pretty clear  that that reality is why silicon valley is doing the stuff it’s doing these days. 00:37:32.237 --> 00:37:38.236 It has to justify itself as a center of innovation in a world where it’s running out of runway. 00:37:38.236 --> 00:37:40.713 And the best answer it’s got is 00:37:40.713 --> 00:37:44.265 “euhhh computers which pretend to think.” 00:37:44.265 --> 00:37:48.366 Forgive me, but I want to think for myself. 00:37:48.366 --> 00:37:51.000 And I think you should, too.