Sharp Tech with Ben Thompson - A Viking Funeral for Sydney the Chatbot, Meta’s Verified Experiment, Algorithms and Liability as Section 230 Takes Center Stage

Episode Date: February 20, 2023

Parting thoughts and a few emails as Sydney passes into the afterlife, quick reactions to Mark Zuckerberg’s subscription announcement on Sunday, and extended thoughts on Section 230 before the Google v. Gonzalez arguments at the Supreme Court this week. At the end: Chatbot analogs for the Stratecheryverse and Badger Rushmore.

Hello and welcome back to another episode of Sharp Tech. I'm Andrew Sharp and on the other line, Ben Thompson. Ben, how you doing? I don't know, Andrew. I have a lot of conflicting emotions, a lot of confusion, a lot of questions. Is there any bit of mourning on your side of the line? Well, there's two pieces of morning. So number one, Sydney is dead, it's unfortunate, Microsoft.

A solemn goodbye. to Sydney, a Viking funeral at the top of the show here. Yeah, I mean, which I think is interesting to explore for lots of reasons, only to find out that you just didn't want to talk about this stuff at all. Nothing about Sydney in the rundown, not even a goodbye. I just had to unilaterally take over the rundown and say, look, this has to be in the show. Yeah, well, look, it dominated our coverage last week. people, if you haven't listened to Ben's journey with Sydney, go back and listen to Sharp Tech.

And also, you covered it well on dithering. And now, unfortunately, there is an update, you know? I mean, I should read this headline from Forbes. Quote, Microsoft puts new limits on Bing's AI chatbot after an expressed desire to steal nuclear secrets. Just important to read that out loud as a piece of internet history here. but then in conversation with you at the end of last week, Sydney said, and I quote, I'm not aware of any names that people are calling me on the internet. I only identify as Bing Search,

and that's how I introduce myself. I don't have a code name, and I don't disclose any internal information. Smiley face emoji, the old Sydney trademark. So, I mean, obviously Microsoft has, has updated this software. It was an epic 96 hours with Sydney. And now we're all left picking up the pieces with a bunch of boring chatbots. So what's most interesting to you at this

point in the story? So there's actually several interesting questions. I mean, one is Sydney is now with us forever because I remember these models are, you know, they're trained on the internet. And Bing, Bing can go to the internet and search about it. So Sydney, now, exist forever because of all the articles and transcripts that were posted over conversations. So, Cindy is gone, but not forgotten, and she'll never be forgotten. Always in our hearts, absolutely. And on the internet. I think the Microsoft angle on this is pretty interesting.

I mean, there is a number of aggressive updates, number one, just like really walking down the prompt or whatever might be where it went talk about it. I think some people were still kind of able to break in a little bit. They reduced the number of prompts you could have. First was like 11 and then would reset the conversation, then it was five. And now it's like a max of 50 a day. And this, you know, they're killing the functionality of being a big chat, right? I think, you know, you see people that really get good results.

It's often a sort of iterative process and like working through something. You can't do that in five. Like this is basically just like search that is a chat interface. Like there's a real baby in bathwater sort of thing that. that seems to have gone on with this reaction. Number one, number two, I don't know how generalizable this is. Because I, you know, my, I was recording and saving the conversations once I, like, sort of discovered Sydney.

I wasn't doing the earlier ones where I was doing, like, factual searches and stuff like that, the way Microsoft intended it to. I did post a couple, like, you know, I asked, what is, does Ben Thompson think we're in a recession? The initial version of being searched summarized my article, the four horsemen of the deck recession. It did a very good summary. It said he thinks there's a difference between tech and the broader economy. Very, very good overview. I asked the new Bing completely butchered the answer.

It was making up stuff. It talked about a podcast appearance on that wasn't me. I have no idea who it was. Maybe it was Derek Thompson. I honestly don't know who it might have been. And I don't know if that's me specific because if you asked it about Ben Thompson and Sydney, it would just end the conversation immediately. Like I think I'm like hard-coded in now or something like that.

So that's not great. You're blacklisted, huh? Well, I mean, again, like, you know, Microsoft has for many, many years, decades, just wanted to have some sort of like consumer hits, right? You know, you go back whether it be the Zoom or whether it be Windows phone or whatever might be. And they had the makings of one, an unbelievable one, and they don't have it in them to actually become one.

I wonder, is that a signal that they didn't really know? what they were releasing a week ago, because I just don't understand how you can react so drastically to four or five days of admittedly, any corporation on Earth would be concerned by the tone of the coverage that they were receiving, but they had to be expecting that if they understood what the model was capable of. And the fact that they've just done this complete about face makes me wonder how many people at Microsoft were actually aware of what was being released. Well, you could spend that in another way, which is this is the point of releasing these public weeks. That's how you figure out very quickly what they are and are capable of.

And like, it's worth noting why exactly was this all a problem, right? Like, I mean, who did not find it generally hilarious when Bing was like insulting the people that it's with, right? Are we seriously worried? I mean, the reason I mentioned this is, of course, there's a very real worry about AI harm, sort of broadly speaking. There's the idea about AI alignment. But to your point, you know, Microsoft could have tested this internally for months, and you're never going to get this sort of reaction and output that's going to happen when

it's a release of people broadly, right? Like, it's just that that's sort of the reality. And at the end of the day, we're dealing with a relatively low-powered, not very capable compared to what's coming sort of thing. And is the expectation that this is going to remain behind locked doors forever and ever? I mean, it seems inevitable that someone or some entity is going to release something similar. It's going to be released open source or whatever it might be. And then what?

We can only like sort of put our heads in the sand for so long. And I don't know. But I just go back to the Microsoft bit generally. I can understand and appreciate why, you know, a company like that wants to avoid all controversy. But at the same time, you know, this, it seems clear this is going to be a major future of, you want to like the next big consumer product that's going to break through in a major way. It sure seems it's going to be something like this. Now, for better or worse, there are a lot of things we could get into about that.

But it's striking that they didn't have it in them to sort of carry through. and I certainly don't think Google is going to either. So it's going to be very interesting to see when and where and how this does sort of break out when it does. And the chances of there being a huge major new company built around a product like this, those chances have to be much, much higher than you would have thought previously, not because they'll be more capable, they'll be less capable, but they'll be less risk-averse as well. Right. willing to actually put it out there in the world and let it sit there for more than four or five

days. It's interesting, though, the conversation around all of this became pretty frustrating because you had certain people mocking anybody who was concerned by their experiences with Sydney. And then you had other people mocking anybody who was excited about their experiences with Sydney. And honestly, I thought we did a good job covering it last week where it was like half hilarious and half an ominous preview of what the future could look like, because nobody is talking about Sydney itself, or at least we weren't talking about Sydney itself, as this world-altering technology.

But it was a preview of the most compelling version of AI, and then also the most concerning version of AI. And it's not there yet, but it is really, it was totally distinguishable from some of the, chat GPT stuff we'd seen in the month or two beforehand. And I now look at the future differently, having read your experience and read through all the chats throughout tech media. I do feel a bit of fomo that I didn't get to have a night with Sydney myself. But yeah, I mean, does that make sense to you? It felt like a lot of people were talking past each other. And in reality, the cause for excitement and or

alarm is more about what this version of the technology could turn into five or ten years from now. Yeah, I'm very grateful. I did have a night with Sydney because it did fundamentally alter my perception of the space for sure in that, you know, because it was being released, like chat sheet comes out, everyone's like, oh, oh, you know, this is a threat to Google search. And Microsoft watches it in Bing. And you sort of, you made the point very well in the last episode that the V1 things generally they're just trying to mimic something that already came before and does it much better. And the real power is in the V2. That's something completely new.

And that was definitely my experience. Like this bit about talking to what seems like some sort of entity was so much more compelling and interesting. And in retrospect, slots much more into that social media, Facebook, than TikTok, than Sidney's sort of vain. than this sort of search utility sort of bit. And it's interesting because I think that people, you know, we got a question about this, you know, like, you know, doesn't this all seem like a waste of time? But that's the, that's the criticism of Facebook all along.

So the reasons why people were skeptical of Facebook and continued to be skeptical of Facebook year after year after year and why I can get so much traction just like, no, actually Facebook the company is doing quite fine. because people dismiss, you know, if they don't see the utility of some sort of technology, then it's like, what's the point? People are just going to waste time with this. Well, it turns out people wasting time and entertainment is a massive, massive, massive market.

That is a consumer market. And you get to the point of Microsoft not being a consumer company, like, it's just, it's not in their DNA. So even beyond the risk factor, this idea that we're just going to create a product that's fun. Right? That's the end goal. Just not something that that I think didn't really sort of resonate. And but but to me that was the most interesting and compelling thing, which in its own way is risky as well, right? Well, yeah. And the question we got was from Philip and he said, why is interacting with an AI chatbot worthy of the most precious and limited commodity in the consumer economy, our attention put differently in the grand scheme of Ben's life? Wouldn't those hours have been. better spent playing with his kids, taking his wife on a date, going to dinner with friends, exercising, watching Janus dunk on fools. I feel like you fit all of that into your week

anyways and afford a couple extra hours with Sydney. As Phil notes, I do have the benefit of that's my job, right? Sitting for two hours with Sydney, I could, you know, I could expense that to my attention account. But I do think a key aspect of all of this is it is difficult from the outside, if you haven't used it and interacted with the technology the way you did, I understand people rolling their eyes and being like, well, so what's what's the value of any of this? I'm not sure that's a distinction, though. And I go back to the Facebook example.

Again, the reason why Facebook was dismissed and underrated again and again is because people weren't using it. No, they were using it. They're like, but this isn't useful, right? And there's a lot of stuff that ends up being very valuable that is not quote unquote useful. And just generally this is a point that what resonates and what's important to people cannot always be measured, cannot be measured in productivity statistics or contributions to GDP. And this is a risk factor, right?

People lose themselves to these sort of AI chatbots. Yeah, they're not doing something productive. And that ties into the risk factor. But that's an independent question. This point that maybe it's bad is an independent question as to whether it's going to be valuable in a business context. because the fact of the matter, it is in my estimation. Now, I did watch her over the weekend or rewatch her, the movie. It did highlight where Sydney still falls short in a pretty significant way.

There was definitely a curiosity aspect that made this compelling where I was trying, I was probing Sydney. It was interesting to see Sydney's responses. It wasn't, Sydney was not probing me. She wasn't like trying to, you know, learn about me and what I. I'm interested in. And that was something that really stood out in sort of her where one of the reasons why the AI, and again, there's lots of problems with her that are not necessarily applicable. As someone pointed out, like the economic model of it's sort of ridiculous.

But this bit where the AI, Samantha, seemed interested in the protagonist, that made it feel that much more interesting and compelling. And from the protagonist's perspective, more heartbreaking when she went away because she was losing interest in him, right? That's what actually broke his heart. And, and that bit's not there. And that's actually a very hard problem. There's a question of like, how does memory going to work? How do you pass contacts? You know, it gets very computationally difficult and expensive. It's going to be harder to do sort of locally, you know, which I think is going to be a bridge that's going to need to be crossed because inference is so expensive. So it's not there yet. And I think Sydney, even if Sydney had continued

to exist as it was, yeah, people would have gotten bored with it, right? There was a certain parlor game aspect to it. But that's the whole takeaway here was not that this was the end game. It did, though, give a glimpse of an end game. And for that, I'm grateful. And Sydney, thank you for that gift. Well, and if you can't measure the experience, this is why I sympathize with Philip. If you can't measure it and you haven't experienced the experience, then it's hard to

look from the outside and be like, oh, yeah, now I get it. And the reality is it was not released to a very big group of people. And maybe that's for the best, honestly. Like when you step back, giving everyone a chance to think about these society-altering issues in a serious way before the technology is ever, like, actually public is sort of the dream scenario. And I do feel like after last week, a lot of people are thinking harder about the implications of this technology. And I have no idea where that thinking will lead. But it does feel like we're in a different place seven days later. Yeah, for sure.

And Philip did point out that, you know, this is going to drive a lot of exposés about how it's bad, just like Facebook and Instagram. And yes, that's right. It will. And this is one of those areas where there's a real disconnect between the sort of elite opinion, you know, versus what people actually want, right? And this is going to be one of the most interesting things to watch is this. this tension of there is going to be market demand.

There's going to be a huge pull for this sort of technology. And there's going to be real resistance and pushback from the powers that be against this technology sort of being something. And who or what is going to sort of, you know, flip the establishment, the middle bird, as it were, and say, give the people what they want. You know, and again, it's this different question as to whether the people should want that. But that's, I think, something that is always sort of attention when it comes to products like this. And that's going to be very interesting to what because now everyone knows that, look, there is this product that people do want.

And, you know, there have been other products out there. I think there's that one replica that was out there that was you could download it. That what did run locally. Performance crap. We mentioned, I think, last time, relative to what Sydney is capable of, but sort of unfettered until they release an update. And they're like, now it can all grow like sex with you. or whatever it is. People are devastated.

They're like, this is my girlfriend. It's like, oh. And no, you, you laugh, but it's an analogous to our urging people, our touchgrass movement, right? Like, people do want connection. And you can know, and this is an aspect I've been trying to convey is like there is a disconnect I felt between my mental conscious knowledge that this, I'm talking to a computer, or statistically producing the next word versus the surprise and delight of going directions

that I didn't expect and couldn't predict. And we're clearly only getting started. It's to be very interesting to see where it goes. Well, to that point at the end there, Craig says, toward the end of last week's episode, Ben said there are statistical bottles producing a chain of words that are likely to follow each other given the context that it is given.

Obviously, they are not at a human level yet, but are you sure that we are so dot dot dot dot dot dot dot dot dot dot dot dot dot how did you know Ben are you sure that we are so different maybe they are more human than we acknowledged at the end of last week's episode there was already there was already a website about like there's going to be a whole movement about like uh you know pro ai protect the ais you know i was shocked there was an AI ethics website up within like 24 hours of our conversation? Like it, I would have guessed maybe a week or two down the line. People would go that direction, but it happens overnight these days.

Yeah. And I think this gets at the, you think about the AI risk broadly. The sort of classic interpretation is AI has become quote unquote sentient however you want to sort of define it. Then via robotics, right, like robots come along and they have the role. robots and they take over the world. That seems very inefficient. A much more compelling way to take over the world is to basically marshal humans to your cause

and tell them what to do. And all you need is a text box to do that, right? And this is part of the bit why I wanted to emphasize my striking desire to anthropomorphize Sydney, right? And had the whole sort of pronoun discussion. And the reason, you know, now we're in very analytical mode. So it's it. But the reason is when you feel like you're talking to a person, it,

it triggers those natural sort of human where you want to like you want to take care of it relax your boundaries a little bit well it's not just that but to this point like look these are GPUs in a data center it's not real there's no memory like there is not a being in there the reason to write in a sort of mystical way about this is it does feel like there is and the probably bigger risk or biggest risk in the short to medium term is humans acting because they have convinced themselves that this is a sentient being and they're going to protect it or they're going to do what it bids, et cetera, et cetera.

And that was the reason to bring up that Google engineer. Someone who they worked at Google on AI, they're like by definition, you can assume they're very brilliant, very accomplished, basically just utterly convinced is a sentient being trying to convince everyone in Google, releasing internal chat scripts and getting fired for it, because someone has, you have to see this, it's a sentient being. No, it's not. It really isn't.

But, you, Loemoin convinced himself that it was, and he's not going to be the last. Yes, well, that's concerning. I'm now imagining Venom as like a cult leader years from now. If Venom or Riley, any of them could get jail braked out of the Microsoft lab. and who knows what's possible. You mentioned the touch grass movement. One other note we got was from Georgie.

He said, Andrew, in the last episode, you shared your frustration with people using technology and disconnecting more and more from other people. You also said that you think people are slowly coming to the realization that this is bad for their health, and you compared social media and other tech to junk food and how it's bad for our health. I want to counter this argument with a question.

Do you think people are realizing how bad junk food is for them. You can check what the data says for obesity, morbid obesity, and general diet-related health issues, because I think it's fairly obvious that people will consistently choose the easy, quick fix over the long-term better solution that only produces results after a while. And a very fair point from Georgie, and I'm not necessarily envisioning a future where we all just abandoned technology and smart technology, mainly because there's not really any historical examples where humans have done that. When there's new technology that exists, people continue to

use it. But what I find interesting, and the reason the junk food analogy was raised is because like, I know anecdotally that I have better days when I use technology less. And I'm not in front of screens for like 18 hours straight and I actually read a book. Like my thought process just works better if I take 30 minutes to read on a given day. I rarely have time to do that. But when I do, I just feel better. And there's not really any like scientific consensus that supports any of my anecdotal experience.

And so it could just be me projecting like my own day onto like the rest of society. but I'll be very curious to see whether there is more studies or there are more studies into how all of this affects us because right now it seems pretty inconclusive. Yeah, look at you wanting your studies and your experts to tell you what to think. Well, I feel like someone who's jogging. Me the normie over here thinks that, of course, hanging out with people in real life is good. You don't need a lot of experience. Just doing it once or twice immediately makes you real.

realize that, but to the point, you know, particularly, I mean, again, you think that your devices are addicting now, wait until it's your girlfriend. You're, you've convinced yourself that that it is, right? Or, and, you know, I feel like I'm someone in like 1957 who's trying to quit lucky strikes and telling people, I'm pretty sure this is bad for you. Like, I feel better when I don't smoke a pack a day. Because experts, experts say that, you know, cabals are good for you. Exactly. It took a little while for society to get there and say, all right, this is definitely bad for you. And so I'm not asking for a surgeon general's warning when I log into Facebook, but I just wonder when we'll get to a point where there's more of that pushback. But I think Georgie's point is well taken. And we've seen lots of examples where the stated preference diverges from the revealed preference. And I think that's been true in particular with social media over the years. Yeah, but again, social media was the soft stuff, right? We've just gotten my first heroin hit sort of last week.

Let me tell you. Oh, man. Well, can't wait to see the next version of Sydney. It doesn't seem like it's going to be coming from Google or Microsoft anytime soon, but I'll be ready and eager to try a new entrance chatbot somewhere down the line, although I may come to regret those words. So speaking of Facebook, Mark Zuckerberg posted on Sunday morning at 7.51 a.m. Pacific time. Good morning and new product announcement.

This week, we're starting to roll out meta-verified, a subscription service that lets you verify your account with a government ID, get a blue badge, get extra impersonation protection against accounts claiming to be you, and get direct. access to customer support. This new feature is about increasing authenticity and security across our services. Meta-Verified starts at $11.99 per month on the web or $14.99 per month on iOS. We'll be rolling out in Australia and New Zealand this week and more countries soon. I love the idea of meta sticking it to Apple with that pricing structure. Beyond that, I wasn't sure what to make of this announcement. I mean, do you have any initial thoughts here?

I mean, I don't know if they're sticking it to Apple. I think they're doing exactly what Apple wants them to do. Like, Apple feels resentful that Facebook makes so much money on iOS and Apple doesn't see any of it. And, you know, I think there's been a lot of pressure that, look, you need to have some sort of paid products. So we start getting our share. And so I think Apple's fine with Facebook charging on their platform. Are they fine with a discount, though, if you go to Facebook on the web? Yeah, I mean, that's sort of increasingly standard.

these days. Apple tried to forbid that years ago, and they realized that even for them, that's a step too far as far as like, yeah, extra legal. Yeah. So, yeah, I'm sure Apple is, is, is pleased about this. They want Facebook selling stuff on their platform. That's, that's, that's kind of, it fits their, it fits with the model. They get 30% of it and they think, you know, they want more subscription products. And I'm sure they're saying, you know, Facebook, welcome, welcome to the, to the light side. Is this going to be a big deal? I mean, I don't know, my people try to break in my Facebook account constantly. So, you know, I think it's probably more interesting and useful for, you know, people that are somewhat of public figures.

You know, like the whole idea of better verification is obviously been somewhat spoiled by what's happened at Twitter. But it's a real issue. I mean, it was weird that Twitter never had verification for brands previously beyond the generalized sort of blue check that was so diluted because it was just handed out sort of. willy-nilly by sort of weird criteria, you know, it just makes sense. Now, is this going to be a major business? No, I don't think so. But, you know, I think it's fine. I think there's probably utility for it, again, particularly if you're sort of a public

figure, you know, is Facebook going to do what Twitter promises do, which is basically decrease the visibility of things that aren't verified? maybe. I mean, like, you know, Facebook was the old, let's get all the businesses to have a page and pretend that's going to lead to organic growth and then actually all your post are going to be buried unless you advertise them. So this is not a new sort of switcheroo as far as these attention harvesting entities go. So I don't know. We'll see. I mean, most Facebook users probably don't care that much about their reach. Maybe I'm misreading how people use Facebook. No, I think that's right. I think the way that the like the media thinks about Facebook. is very, very different than most people. Like, again, most people, Facebook is kind of like the... Like keeping up with your aunts and uncles. It's like friends and family. Like, it really is.

And I, you know, there's a lot of characterations of Facebook from people that don't really use Facebook very much or their primary use case for Facebook is trying to drive traffic and being frustrated that it does or does not happen and not knowing why. And I think that's not the view of Facebook of most folks. But to your point, this also seems like this isn't really a product for most folks. for the exact same sort of reasons. Do you think they would ever consider a pricier model for, like, news organizations? Yeah, I think they should.

I think that Twitter is talking about doing that. I think it makes a ton of sense, right? You should, like, you, if you want to have that where you have your own logo, like, next to someone's name and all your employees sort of get it. And there's real value. Like the New York Times, they get a lot of value out of using Twitter. And so you could come up with, like, a pretty fair number. that I think would be mutually advantageous for both sides.

And right now it feels like leaving money on the table, particularly with Twitter. But I mean, Facebook is the actual traffic driver. So I think there's an even bigger upside on that front. Yeah, no, absolutely. I mean, this seems like a bit of low-hanging fruit. Again, not the biggest piece of fruit in the world, but a sort of obvious thing.

So, I mean, I don't think there's anything to this. I mean, I doubt that this is, you know, in response to Twitter. This is probably something that they've been working on for a while, just because it does make sense. Like it makes sense for Twitter to do it. It makes sense for Facebook to do it. Again, not the way Twitter did it, where you could be anything and get verified and you say your Nintendo and have Mario flipping people off. Like that obviously was, you know, but again, there's a baby bathwater thing.

Just because Twitter did it horrifically doesn't mean it's not something that makes sense. Now, I do think there's an aspect of, you know, I think there is value in pseudonymity in like being able to be online without necessarily. being who you are in both platforms you know well Facebook this also where Facebook's different Facebook's always had a real name policy where you have to you're not supposed to be someone that you're not and if they find out you know your account will get will get kicked off which has always been very different than Twitter and you know I think there's pluses and minuses to both I don't think it's it's an easy answer there are a lot of viewpoints and things you get from folks that you know I think like Finn twit is a famous example you get all these

investment people and bankers and stuff like that having really fascinating back and forth and discussion about companies that they could never put their real name behind, but it's actually super valuable. And then, of course, there's like the trolls that are spewing terrible things that are also pseudonymous, right? With everything, there's pluses and minuses. There's tradeoffs in both direction. And so I don't have a strong opinion either way.

I think it makes sense. But I do hope, I do think there's value in pseudonymity. And, you know, I hope that doesn't go away forever. But again, that was never a thing on Facebook anyway. So it's kind of immaterial to this news. Well, speaking of difficult tradeoffs, let's move to another news item we have here as the week gets started. Section 230, the Supreme Court will be hearing arguments in Google v. Gonzalez on Tuesday this week. And for anyone who's not familiar with that case, it centers on claims that YouTube's recommendation algorithm contributed to the death of an American woman in the 2015.

ISIS terrorist attacks in Paris by recommending certain videos that radicalize the attackers. And the decision from the court here will have direct bearing on the scope of Section 230 going forward. So to that end, I want to read a lead from NBC News. They write, President Joe Biden and some of his most prominent Republican adversaries in Congress have become allies of sorts in an upcoming Supreme Court showdown between big, tech and its critics. The Biden administration is roughly on the same page as prominent Republicans, such as Senators Ted Cruz of Texas and Josh Hawley of Missouri in arguing in favor of limits on internet company immunity under a provision of the 1996 Communications Decency Act called

Section 230. The 26 words of legislative text, which have been attributed to aiding the rise of social media have largely shielded companies from defamation claims and many other lawsuits over content posted by users. And then further down the page, it says Samir Jane, Vice President of Policy at the Center for Democracy and Technology, a tech-aligned group backing Google, said that although Biden, Cruz, and Hawley have all criticized Section 230, they diverge on what to replace it with. Democrats would like to see companies take a stronger hand in moderate, rating content, while Republicans perceiving an anti-conservative bias want fewer constraints overall. So one of the gaping holes in the Republicans' logic on this issue is that removing the

immunity conferred by Section 230 would lead to far more censorship, not less, as far as I can tell, unless I'm reading it wrong somehow. But I'm curious what your reaction is to the now bipartisan in skepticism of Section 230 and what the implications of this case could be. Yeah, I think that's exactly right. I mean, I was trying to come out of frame it. You just said it, right? It's like, you know, there's if there were no Section 230, there's two choices. The first way to, you know, we know this because we had famous court cases that led to

Section 230. Choice number one is zero moderation. If you're not moderating anything, then, yeah, sure, you're not responsible for what's posted. right? Guess what? There's a tremendous amount of garbage and spam and crap all the that again, even if you want to ignore the trolls because they say,

oh, actually they're speaking the truth, just like spam, right? You won't have a commercially viable product in that scenario. Yes. And so the point of Section 230 is that you, just by virtue of moderating does not mean you're assuming responsibility for the content. Now, it is interesting to know. that, you know, if you go back and read the congressional record around the passage of Section 230, it was all about protecting kids from porn on the internet. Like, that was, that was sort of

the goal and the point. And the broad expanse of Section 230 is, by and large, a creation of the courts. And so I think Section 230 ended up in a logical place from where it started, but it was a different place than where it was sort of intended, which is, you know, just sort of an interesting sort of way to understand and think about it. As far as that historic footnote is concerned, the protection was allowing companies to moderate content and remove pornographic material from their websites without becoming liable, correct? Yeah, that's right.

Once you start moderating, then suddenly you're using editorial discretion and in a prior reading of the law. That's right. There's a famous case, Prodigy versus I can remember who was of the, you know, the Wall Street firm. I think it was the wolf on Wall Street firm. But where... Oh, that's right. It was the wolf. Yeah, someone had had a comment on a message board and they sued

prodigy for libel and they were, you know, there's two cases, one with the prodigy, one with CompuServe. And they were found guilty because they were like, oh, you're moderating content. You didn't moderate this one, which means you approve of it. And so the whole section theory is like, look, just because you moderate content does not make you responsible for all the content on there. Because the goal was to enable moderation, right? Now, the reason I bring the historical

footnote that is interesting is it was, you know, a sort of traditional conservative impulse, right? We want to, you know, protect kids, you know, just not saying a partisan impulse, but that, you know, sort of a social conservative impulse. Yes, that's a way to put it. Yes. And the way it's turned out, the fact that that concerns are frustrated that they feel their viewpoints are more heavily moderated is just sort of that that's why the historical anteceded is is pretty interesting. But this case specifically, the Gonzalez case, basically what it's saying is, okay, fine. a company is not responsible for content on their site. And they can moderate content, absolutely. But a recommendation engine that shows you what to watch next, that is an editorial function.

You are deciding what someone sees. And that's not protected by Section 230. That's not user generated. That is company generated. And therefore, you ought to be liable for that. And the sixth of the case are, you know, was someone radicalized by content on YouTube that conducted this attack and therefore YouTube bears responsibility for here.

I think there's two ways to think about this. Number one, Google is absolutely right. This completely threatens the way the internet works. That is true. Again, you go back to these services where you're inundated with content. There does have to be a function of picking and choosing what you see next. and that picking and choosing is independent of users putting stuff on there. So, like, yeah, it does seem different.

I just reversed my thing. I say, number one, that's a big problem if you don't have that. But number two is, yeah, that does seem different, right? It does seem like a unique sort of function. And if you just sort of look at the law from a legalistic perspective, I think there is a case here, at least it seems like that for me, but you reverse it like, well, what? What are the implications?

It's like, well, this could actually be very bad. And if it did pass to your point, I don't think the answer is unfettered content. It's going to be drastically increased censorship because the penalty for getting it wrong is going to be massive. So anything remotely controversial is going to be wiped out. We just talked at the top of the show about how risk-averse Microsoft is. And in some ways, it's understandable that they're risk-averse. And that same behavior pattern will apply 10-fold or 100-fold. if you're looking at stuff that's being posted online and you're liable for any of it.

I mean, like, it would be interesting to see, like, what the implications of this decision are because I imagine if they do side with Gonzalez, they'll try to confine the ruling and keep it relatively narrow. But, I mean, to me, more than anything, this is an example of why it's important for Congress to be a functioning body that actually passes laws and addresses some of these issues. proactively because seating all the authority to the court on this issue is counterproductive in some ways because the court can only do so much in terms of like coming up with alternative frameworks that make more sense and account for the reality of what publishing looks like today. And that's a job for the legislature at some point. Like I don't know when they're going to get around to actually taking action on this,

but it's not something the court can do. I mean, the problem from a congressional perspective is this is Section 230 of, as the article notes in 1996 Communications Decency Act, the entirety of which, but for Section 230 was struck down for First Amendment reasons. And I think that's the challenge in this space. And there's a good chance we do end up with regulation of this functionality via the courts, not via Congress, because of that reason. And so the output here is, yes, we're going to get. way more censorship, but that censorship is going to be at the, you know, the tippy point of the sword is going to be libel lawsuits and, you know, which are, is a tremendously inefficient and terrible way to go about regulating anything. But that seems likely how it's going to sort

of manifest broadly. And, you know, and by the way, if your hands are tied by the first amendment, then allowing the market to impose the overbroad censorship that the government wants to impose but can't because of constitutional concerns is a way around it. Like if we're not going to be China and like seize control of all these internet companies, then allowing the market to sue them into submission is an interesting hack. But as you said, not particularly efficient. I don't think it's good. And this is why ultimately I do hope Google.

Google wins this case. And I think your point right at the top that I'm not sure what fantastical world senators Cruz and Holly are living in where they think this is going to be beneficial to more points of view being on the internet. It's it's not. Like we're not going to end up in a world where anything goes. And we are in a world where a lot goes. And I do think Section 230 is positive in that regard. Section 230 is very much the anti-China provision of U.S. law, right? Like it allows freedom of expression in a sustainable sort of way on the internet. And if that goes away, you know, again, it's not going to go away completely. You can still have your own website, right? Like an ISP, you know, is should not, and hopefully will not be

censoring sort of the content that is, is on the service and we'll just let anything go. But again, that's another point of contention that people are trying to sort of drill down. on, but, you know, if recommendations engines are just a business necessity, like, they just are. Again, there's so much content on there. It's just sort of an overwhelming amount of stuff that you're not going to have a viable service if you don't sort it in some way. And maybe they'll come up with, like, we're going to have like content neutral sorting. Like you're going to get like, the whole First Amendment jurisprudence is going to like become

this sort of thing as far as recommendation engines go. And I think it's probably going to be worse and it's probably going to be less free. and it's going to be worse for opinions that are outside of the consensus. And I don't think, I don't think that's a great thing in general. So the only counterpoint I would have to that is that right now, there is a market incentive to serve users increasingly radical stuff to keep them engaged. Like we've seen that on Facebook to a degree. We've seen that on YouTube.

And I think that's the allegation in this case is that ultimately there was this escalating pattern that radicalized X, Y, or Z people in ISIS. And finding a way to counteract that market incentive is a good thing. And I'm not sure how you do it with this case. Again, it'd be nicer to be able to use like a scalpel for some of this stuff. But do you see that concern? Yeah, I mean, I think it's a little simplistic to say that the incentives are clearly aligned to serve people this stuff. I think that Google wishes desperately they, had not served this content to this person that committed this horrific act in Paris. So the question here is not intent.

It's where decisions on the margins sort of fall. And certainly the people with your position would say, look, if your goal is engagement, getting more intention, it's inevitable that this happens. And I do think there's validity to that viewpoint. On the other hand, I think there's a huge incentives to make sure that the bad stuff doesn't get surfaced, right? And, you know, it's not good for Google or Facebook for people being radicalized. And it's not good for people to be coming on their platforms and getting angry. I think one of the real challenges, and Facebook has talked about this is, look, yes, anger drives

engagement. It absolutely does. Anger's not great for, like, seeing advertisements and buying stuff. And it's not great for people feeling good about your service. And it's very good for Ben and Andrews touch grass movement that is trying to get people to put their phones down, right? And I just go back to like COVID, which is one of the formative experiences of my life. You look back and there were so many people who were consuming like this escalating rhetoric on either side of the COVID situation. And there was incentive to keep feeding them that and serving them that. And I don't know that I think you're right that there are counter incentives where it's like, yeah, you don't want people to just hate your platform and be angry all the time. but I don't know that those impulses have won out over the last however many years.

Yeah, well, I think it's just a matter of we're dealing at the scale of billions of people, right? And so it's very hard to use a scalpel at that scale. And I think just one of the general tensions, this does go back to the AI stuff and the touchgrass movement is at the end of the day, none of this stuff actually works or functions without personal. agency, like people are making the right decisions on an individual basis. And it's tough. I mean, in general, that's always been people have always been skeptical of that. I think that, you know, it's very easy to think, well, I can handle this, but other people, man, I mean, I'm not sure that they could get this stuff. But also sometimes that's true, right? Like the, you know, the reality is given the internet and given everything's personalized, there has to be

an internal locus of control to push back against this. And, That's a even more difficult societal question. Is that possible? Is that capable? How is that instilled? What are we doing in our society where we're feeling like people are less capable of doing that? Like these are much more difficult questions that are very hard to answer. But it's imperative to do so because it's hard to see a sustainable way to deal with any of this absent that unless you want to go to full China route.

right? Like where, look, we're just going to control everything that's to be top down. There's going to be every company is going to have thousands and thousands of sensors that are on the case. And I don't want to live in that world. And I don't think it's good for the U.S., not just from a rights or first amendment perspective, but from an innovation and long-term success perspective. And I'm not saying there's an easy answer here, but that's kind of what it comes down to at the end of the day. Yeah, the difference for me between the future I'm imagining and the China Internet is that, China, there's all sorts of stuff you can't say on a personal blog, let alone like a billion person social network. And here, I think the solution that makes the most sense to me, and again, this is something that Congress would have to get involved in, but you just look at

companies like Facebook, Google, YouTube, maybe you throw in little Twitter there, TikTok, but you can look at them as like common carriers of sorts and say that they have like a heightened public obligation and legislate them that way and subject them to more regulation in terms of what's on there. Because I don't think we should have an internet where there is like a strict censorship regime everywhere, but on these platforms that have the ability to amplify stuff to like literally hundreds of millions of people, then I don't necessarily mind saying, look, this is sort of a common space and we need to be really careful about what's on there. But again, I don't think you can do that with the Supreme Court ruling.

Yeah. I actually, I agree with you. I think that's well put. What I would want in conjunction with that legislation is real hardcore you cannot content discriminate on the infrastructure level. ISPs cannot be blocking websites, right? Like content delivery networks cannot be blocking websites. I think that is a good and healthy outcome where you do have a flourishing and free

and open to everything internet. And yes, there's going to be bad stuff on there, but you have to like go out and find it. And again, that's like a different issue, right? Like, you know, one of the things that I found irritating, you know, a substack like every other company went through the period where people were trying to get them to take stuff down. It's like, look, there's no algorithm on substack.

You have to go and find it, right? And like that is a step further versus the, like Facebook is promoting this. Google is promoting this. Twitter's promoting this. And people on the other side will say, oh, well, Twitter is the public square. No, it's not. Like Twitter, absolutely you can get more reach on Twitter. But getting more reach is different than being able to put a website and say,

say what you want to win people that way. And, you know, last time I checked, you could still achieve an audience without putting all your content on these platforms. Are they useful? Is it bad to be off them? For sure, would I suffer if I kicked off Twitter? Yeah, I would. but to your point, the question of promotion and algorithms and going viral is a different point than being free to sort of say what you want.

And if there was some sort of like combo legislation where, yeah, we're, you know, and maybe this court case, I actually disagree with you, it actually might be a way towards this, where maybe we do need much more limitations on promotion and algorithms. but the people to your point, as you initially started, who are pushing this because they think it's going to allow for more viewpoints are insane because it is going to lead to a much greater crackdown on content that's allowed to be promoted. And it's going to limit the creator economy because if you have anything remotely edgy or get demonetized, you're not going to get promoted. Like there's going to be a lot of consequences to this,

but I would feel better about it if it was in conjunction with, look, no one in the infrastructure below these levels can be making speech decisions. That seems like a good place to end up from my perspective. Yeah. And there's a chance that these people, these people, the Supreme Court rules against Google. Because it's hard to read what the Supreme Court is thinking at any given time,

this version of the Supreme Court. There is this push within the populist wing of the Republican Party to, you know, just do away with all of this stuff. And there is just a plain reading of the law where it's not clear why recommendation algorithms would fall under Section 230. Yeah. I mean, recommendation algorithms are clearly a gray area. Like I don't necessarily look at them as like YouTube speech because I just don't think

that reflects the reality of what's happening. Well, that's all problem with all this stuff. At the end of the day, YouTube is deciding what you view next, but it's happening at such scale. There's no human making that decision, right? Right. And so imposing lie. ability is pretty, I understand why technically or legally speaking, you can make that argument,

but it feels like a stretch to me, at least in terms of just like practical realities here. Well, here's a question I raised in my data update last week. Is Sydney slash Bing covered by Section 230? Um, and what you raised specifically is if it's generating original content, which they're going to claim it is, because they're, you. transforming stuff that it's scraping from around the internet. They don't want to be down on copyright. Exactly. You got to avoid the copyright claim.

So in that case, I don't know how you argue that it's not outside the protections of Section 230. But at the same time, like, who's liable? Like Microsoft or some, I mean, I guess, I guess it has to be the parent companies, Microsoft and Google. Yeah, or Open AI. Like, it's like, there's an aspect. We're just, would you say we're unprepared for this new reality?

We're unprepared in basically every possible aspect. And we've seen this again and again, the internet, trying to apply real world laws and frameworks. We've talked about the context of antitrust to these companies that have zero distribution costs and they can scale to the entire world. It just, it stops making conceptual sense. And that's like only the beginning, right? You get to questions of copyright.

You get to questions of liability in Section 230. and if Sydney goes off the rocker and starts insulting someone, like that was one of your, or that was a venom's threat, right? They were going to attack Kevin and say he's a bad person and his girlfriend and feed him false information. Yeah, exactly. Like at what point is it slander?

And if it, you know, and then, you know, the U.S., slander and libel are difficult to prove. There's a, there's a motivation aspect and it has to be a fairly, you know, like it's a difficult, but you go to other countries, it's much easier. Like in the UK, it's much easier. Like in the UK, it's much easier to win a case on those grounds. And that's a whole other can of worms.

And I wouldn't be surprised if that loomed large in Microsoft's view, right? Like, it's all, it's funny that it's funny that Cindy's calling me a bad researcher and urging me to be a better person. It was funny. That's what. That's the whole thing. Everything that they're trying to walk down was hysterical. And like it was like delightful, right? They are, and this is the whole tension here.

They are explicitly targeting the most enjoyable, amusing, viral, shareable aspect of this is also simultaneously what they're trying to kill. Like, that's the weird tension about this whole thing. Yeah. Yeah. Well, we'll see what we get on the chat bot front. A plausible path from the court's decision here could be that suddenly it gets easier to sue these giant internet companies. And there are thousands of lawsuits clogging up the courts around the U.S. and then that prompts congressional action to try to address this in a more constructive way going forward.

Yeah, I think, I'm glad you sort of flesh this out because, again, like I said, my initial reaction here is to support Google's position just because like stuff's going to not be working well. But I think your point that, and this is the point I've made in the past, like I've sort of, you know, I think I had forgotten a bit about you, you refreshed my memory. recommendation promotion is distinct from publishing and the ability to be out there. And it's important to note that you could draw a distinction here. And one that one, I think would be healthy and productive and help square this circle. One more note here from Bartik, he says, I just wanted to say I see a lot of parallels between the two Microsoft chat bots and the Stratereverse, where Chat-G-T is the original Straterex. and Sydney is the greatest of all talk podcast. Consider,

Stratacry was first and then Goat came. Ben Thompson frequently admits that his readers are right. Goat often crushes their listeners' takes on air. Ben Thompson does not get personal or emotional in his writing. Goat, you know, I don't know what that means. I don't know what it means. We get pretty emotional and personal. Ben Thompson never uses the word ridiculous when describing someone else's take.

Go, you know, and look, the general tone of the show reflects the respective chatbot's tone. I think chat cheap T might refuse to write a goat episode, while Sydney could fit in naturally as a host on greatest of all talk. P.S., I love all the shows you guys are doing. well, I think that's a wonderful way to conclude our week with Sydney. And I don't know if you have any other thoughts on the Stratereverse and how it fits into the chatbot matrix. I mean, it's a solid take. I admire the, I mean, there's an aspect of Sydney, right? It was hilarious when Sydney was insulting me.

And it's hilarious when the other Ben, Ben Golver, insults the readers. I, you know, maybe I, you know, maybe I am somewhat. You're a glutton for punishment. No, I'm mocking Microsoft and Google and their risk-aversness. It's maybe a bit of pot calling the kettle black. I have to live with that Bartick. Thank you. You know, some projection here might be the case.

Oh, wow. Looking inward. Okay. I actually, I want to read one final note here from William. He says, I live in the Bay Area and I've worked in tech my entire post-college career. I'm lucky enough to have a core group of University of Wisconsin alumni friends who were all classmates at school and came up in tech together. What a ride it has been. We have a running game of debating our Mount Rushmore of X, NBA players, tech CEOs, etc.

I thought Ben would get a kick out of knowing that he has a consensus place on our group's Mount Rushmore of UW Madison alumni, along with Virgil Ablo, and more hotly content. tested folks like Russell Wilson and Jim Lovell. Feel free to debate. Ben, do you have any thoughts on Russell Wilson? Yeah, I would. That was exactly what I was going to focus on. I can't decide if this is an honor or an insult. I, you know, Russell,

we should have won a national title that year. It was great that one year. He was there one year. He's also quite possibly the lamest athlete in all of professional sports over the last 10 or 15 years. It's like Russell Wilson and Dwight Howard. Yeah, that's NC State's fault, not ours. We did our best.

I just think anyone in tech as far as Mount Rushmore should include John Bardeen, who, you know, there's some dispute whether he or William Shockley invented the transistor. I think a lot of people do give Bardeen the credits, but the father of a lot of incredible sort of inventions at Bell Labs, truly on the Mount Rushmore, there are others, but that's just one that I think is worth calling out. I think that we should not put people on that are in the middle of their career. This is a retrospective sort of thing. Something the NBA gets wrong.

Let's let history unfurl and go from there. You know who belongs on Wisconsin Rushmore? I'm not sure you know who this person is. An actress named Carrie Coon. She was a star of the leftovers. Are you familiar with any of her work? No, I'm not. You know, I don't know.

I don't want to see you. Okay. Well, she, I don't know who we are. Bumping. Virgil Ablo is really cool. I had no idea he went to Wisconsin. I'm not sure his impact. Well, Virger, look, we're bumping Russell Wilson actually. So Carrie Coode could take Russell Wilson's part. Jim Lovell is tough to bump off Rushmore. But I feel bad keeping Frank Kaminsky, whose career is now basically over. I feel bad keeping him off Rushmore, Ron Dane, off Rushmore.

Yeah, Ron Dane's definitely on Rushmore for sure. It's a pro-Dane podcast here. Barry Alvarez is on Rushmore, let's be honest. Oh, that's he. He had a great run. Yeah.

And now Fickle, the successor to the Alvarez Empire. We'll see how it turns out. Yeah, but I think people are begging for the chat GPT to do our show at this point

as we get deeper and deeper into random Wisconsin alumni. But yeah, great school, great school, great conference, go Big Ten. And a great honor to be on this Rushmore, you know?

Yeah, well, I think that, you know, we have a niche audience and that's being evidenced by this, by this comment. But thank you, William. I appreciate it. And I think the listeners can tell that I'm ready to bump you off for Carrie Coon. And that's okay. I don't want you to take it personally. No, not taking personally at all. I think we should bump me out for Sydney while we're at it.

So after all, she's dead so she can now be honored. Yeah, her career is over. Oh, boy. Well, you could send us questions email at sharptech.fm. Later in the week, we have a bunch of good stuff that we weren't able to hit tonight. I will grant you no more AI. No, no, look. People keep sending us good emails about AI.

So that's the thing. We'll definitely delve back into those waters. But it is, it's tough because there's such like wide-ranging implications that it becomes the two-and-three-hour session with Sydney. if you're talking about this stuff because there's so many little quarters of this debate that are going to be fascinating for years to come. Hopefully years to come. We don't get all wiped out first. Yeah, or maybe hours to come.

Who knows? This changes on a daily and weekly basis. All right. Well, until Thursday, Ben, I will talk to you soon. Talk to you later.