SED 1877

[0:00:12] GV: Hello, and welcome to SED news. This is a different format of Software Engineering Daily, which we do monthly where we just take a spin through the latest news headlines, things that popped up in the kind of the main news. We then dive into a deeper topic during the middle, and then we take a look at some Hacker News highlights towards the end. So, as usual, we just like to kind of have a catch-up on what's been going on our side. I'm Gregor Vand, and with me is Sean Falconer. Say hey, Sean. 

[0:00:42] SF: Hey there. Good to be back. How you doing, Gregor? 

[0:00:44] GV: Yeah. Yeah. Good. Good. So what's been going on over, I guess - I always lose track of months these days. Your September, Sean. How was your September? 

[0:00:52] SF: I was on the road a bunch. I was gone to Europe for 10 days. I did five cities in Europe for work. A lot of speaking, customer engagements was fun, but it was very, very exhausting. So, I'm glad to be back in the US for a little bit. But I do leave again for Europe next weekend. So, I'm in like kind of my heavy travel schedule for the year. So, I'm just trying to survive at the moment. 

[0:01:13] GV: Yeah. Well, we managed to catch you flying in for this episode. So, that's much appreciated. Yeah. On my side, yeah, it's just been I guess sort of ramping back up month and obviously conferences, etc. Starting to tick back in. I mean, this week in Singapore is what's called Token Week. I have nothing to do with crypto or tokens. So, it's a bit of a strange week when a bunch of people that look different, shall we say, to the usual Singapore population descend on the city. But I went to a sort of enterprise - I wouldn't say the company. But I went to sort of like an enterprise AI conference. 

And certainly, I think reception was quite muted, I would say. They were really pushing their agents. They have a special name for those agents. But if I say the name, then you'll know the company. They were pushing their agents and really trying to sort of bring up their customers on stage to talk about what they've been doing with them. It just wasn't a great reception. We're talking sort of some quite basic RPA type things. So, it was certainly interesting. And I do think that's maybe just over this side of the world. We're not quite pushing out the - that said, it's an American company. Again, I won't say which one, but it was an American company. I'm just not sure if that's sort of the level we can be expecting of agentic conferences right now. 

[0:02:23] SF: Yeah. Do you think based on your experience that at least the part of Asia that you live in is maybe a little bit behind the United States or other parts of the world when it comes to AI adoption? I certainly noticed that at least in parts of the Europe that I was in. I ran kind of a lunch in there for executives, and I started by asking everybody in the room which companies were actively building something in AI right now, even if just POC's and demos, and not a single person raised a hand, which is quite a contrast from my experiences in the US. They're all very interested in this topic, but it felt like they were a little bit slower in adoption curve. 

[0:02:57] GV: Yeah. No. I mean, I think that's probably a fair assessment. I mean, obviously, yeah, where you're based, San Francisco, in the valley. For sure, that has to be pretty much the place it's getting adopted first certainly for where I look and sit. Over here, yeah, I think there's still inertia. And I think people are sort of in the businesses that are - if I think of the customers that came on stage at that conference, let's just say big shipping companies or something to that effect, I think there's still just a bit of inertia around, "Well, why should we change our ways of doing things just so that the CIO can say that they've cut costs?" 

And I think that was the big theme of that conference was CIOs saying that they could cut costs here there, and everywhere using AI. I thought that was an interesting theme. And we're talking agents to help you reduce your SaaS spend and having employees be able to talk to a chatbot and say, "How can I reduce my spend on this platform?" which is quite on the nose given that this was obviously a SaaS platform in itself. I don't know. It was kind of interesting. 

[0:03:59] SF: Nice. Yeah. 

[0:04:00] GV: Okay. So moving on to the main headlines. These are things that pop up in the main news outlets. The more niche stuff comes a bit later from the Hacker News. So the big one that we're looking at to begin with is Nvidia investing in Intel. So we did look at this last month in terms of - well, we looked at Intel last month. They had an investment from the US government to the tune of sort of 10% of the company. That was sort of on the basis of - we called it too big to fail. It's sort of almost like too important to not do well. They weren't exactly failing, but they were certainly starting to lag. That was an interesting play. 

But what's interesting here, so this is a 5 billion investment by Nvidia. And this has been partly in talks though for over a year. And so the timing of all of this is what's more into question rather than sort of how much and so on and so forth. Because the US government sort of popped up 2 weeks before this was inked saying they'll take that 10%. Whereas the talks with Nvidia have been going on a long time. Nvidia have made another investment as well which we'll get on to. But, I mean, what is your take on this one, Sean? 

[0:05:12] SF: It's interesting. I think it kind of goes back to what we talked about last time where - at least from the government side, the semiconductor industry is this critical infrastructure, like energy or defense now. And there's a lot of concerns over this single point of failure in terms of our dependence on the Taiwan chip manufacturers there. What's this mean if anything ever happened? So, I think US wants to shore up their bets. 

But I think it's a strange day in tech when it seems like the government has moved faster than the world of technology in terms of their investment. I don't know if that was a catalyst for Nvidia to move forward with this or why it took so long. But in all of this, between Nvidia, Intel, any idea what's going on with AMD? Where do they stand with all this? 

[0:05:55] GV: That's a great question. We just don't seem to - I mean, I guess just maybe in the slightly more US-leaning press, but we don't seem to hear a lot really about AMD sort of exactly where are they? So I think that's maybe something we can look ahead maybe next month even. We can try and maybe dig in a bit and see where they are on this one. I mean, Nvidia have also made this investment, 100 million into OpenAI. So it's kind of going both forks of the road here. 5 billion into Intel, who are what's called a fab. They can literally make chips. There's only very few companies that can do that. I believe it's them. It's TSMC, the Taiwanese behemoth. Samsung can do it. There's probably some pokey in this - relative terms, pokey Chinese outfits that can do it. But I don't think anyone's from outside of China are using them. Yeah, I feel that's probably where the Nvidia sort of they're hedging their bets a little bit on who might end up making some of their chips. But yeah, they're 100 million into OpenAI. What does that kind of mean, do you think? 

[0:06:58] SF: In a lot of ways, I'm sure OpenAI is a huge customer of Nvidia and their GPUs. So, I saw a meme about it where essentially Nvidia is handing over money to OpenAI and OpenAI is giving the money back with relationship that they have. I mean, I don't really have a sense for what I guess Nvidia is probably - it's in their best interest essentially for these model companies to do well. It's kind of like if you look back at Google in the early days of the internet, Google did a lot of stuff where they probably lost money to just get more people online. Because ultimately it's better for Google the more people who are online, because that means more people searching, more ad revenue, and so forth. So they'll make it up eventually. And I think maybe from a strategy standpoint, Nvidia is probably thinking similarly. It's like, "Okay. Well, the better these model companies do, the bigger they become. Essentially, that is good for our business because they're so dependent on us today." 

[0:07:53] GV: Yeah. There might also be distill problems with Microsoft and OpenAI. Sort of how they're going about their relationship. And this could be a sort of technology geopolitical, if you want to call it that, move, where OpenAI are saying or showing that they've got kind of the heavies in their corner from a hardware perspective as well. There's probably a whole bunch of things at play here. Very interesting. We'll obviously see how this develops. And yeah, we'll maybe follow up on AMD, try and bring them back into the foreground perhaps next month, and see what we uncover there. 

Moving on, this could have been a Hacker News thing, but it did hit the main headlines, which was Meta's classes intro demo failure. So, this is their AR classes offering. They kind of - at least from the shiny videos that they were showing, it was Red Bull. People on skateboards going down hills. And I was expecting them to, I don't know, head-up display the speed or something. But instead of which the guy was asking what's the speed limit on the road, which all seemed a bit of a damp reception, I would say. It was one of the most death-defying skateboard things, and he's asking what the speed limit is. I know they have to do that for probably many illegal reasons. But it just didn't make a great video in my opinion. 

Anyway, they went on to a demo, and this was with a chef. And the chef said like, "So, help me put this recipe together." Which is funny that a professional chef needs that to happen, but let's roll with it. But then the response from the glasses was, "Well, you've already done blah blah blah. So now do this." And of course, he hadn't even started yet. And he tried again. He said, "No, no, I haven't started yet. What do I do first?" And it responded again, "Well, you've already combined the ingredients." And so he just said, "Right back to you, Mark." And of course, the explanation given was Wi-Fi, which I don't think anyone's buying. Yeah. I mean, we've seen sort of AI demos fail before, but this seems like spectacularly bad. 

[0:09:54] SF: Yeah. I mean, there's some great spectacular failures of demos. There's the big one from Microsoft years ago, where they had the blue screen of death during the demo of one of their flagship operating systems, and so forth. In some ways, I have empathy for them having given a lot of live demos. Like, anything can happen. But I'm not surprised that this makes headlines, besides the fact that it's Meta, Zuckerberg, and everyone loves a train wreck. People love to see the failures. But I have a hard time. And maybe I'm just not the target demographic for this. But we keep trying to make these like AI-powered eyeglasses work. Google, they had their attempts. There's been other attempts. Who's this for? Does anyone actually want this? Yeah, let me know if I'm out on an island, you want this. But it's like Jobs gave us the internet in a pocket. Zuckerberg's trying to give us internet in our faces. I'm just, "Why? Why do we need this?" 

[0:10:47] GV: Yeah. I mean, we're actually going to get into this in a bit more detail kind of in the main topic, when we look at sort of where devices, effectively, devices hardware from the big players. I'm actually going to highlight Snap there as well because they've kind of already had a bit of a - I would say it looks very clunky. I don't know from the actual experience with it, it looks of a very clunky AR experience. But that's been aimed I believe more at the younger demographic which is sort of Snap's user base. 

As you call out, this is I think being aimed at certainly adults and which adults like actually want to have something on their face that - I mean, I think the only version of this I can think of that would get me interested as an adult at least for now is I'm a cyclist. And if I had my - I've got quite big frame Oakley sunglasses. And if they could project, say, the speed or some stats. Because sometimes I am like looking dead down at the road, which is where my computer is. And that could literally cause me to crash if I look at it too long. It'd be quite nice to kind of look ahead and be able to see a bunch of stats that I would quite like to still be monitoring. That's it, though. I don't want to walk down the street and have kind of minority report. That's John Smith over there, or like - I don't know. But I think maybe - 

[0:12:04] SF: You don't want to be at a dinner party, and it feeds you nuggets to help you with networking so you remember people's names and little tidbits about them, like a politician of some sort, and you have somebody in your ear. Except you have Jeeves feeding you this information. 

[0:12:17] GV: Yeah. This is also really interesting. Because, again, we're going to get on to this I think in the main topic around like the modality of the device with the AI, if you want to call it that. But just to sort of, yeah, a little anecdote is I have seen a quite high-powered exec that I know on LinkedIn. And he said if you could give me something where I can walk into a room and it can tell me who all the people are, then I will buy that tomorrow. I mean, again, I think that's quite a specific. I'm not wandering into like gala dinners every evening needing to like remember so and so's other half and so on so forth. So, I don't know. This seems quite a niche. Still quite a niche. 

[0:12:56] SF: I also wonder, that's a great thing to be able to do if you can do it naturally or you have some other way of doing it if you're the only person in the room that has the capability. But if everybody's the same and everybody has these glasses on, does it kind of devalue the experience that, "Oh, I know your name. But clearly, I got it from this device that fended into me. And you also know my name because you're also wearing the exact same hardware on your face." 

[0:13:22] GV: Yeah. Again, I think that's what's feels strange about this, like walking into like a networking event with a pair of glasses on that everyone knows is that pair of glasses. I mean, way back, you just touched on it there. But yeah, Google Glass. I mean, I lived in New York when Google Glass was released. And I remember getting into a lift, an elevator with someone going to a networking event and he was wearing Google Glass. And I said, "Oh, is that Google Glass?" And he just went, "Yeah." 

And then I think some people might remember that people wearing these glasses got the slightly unaffectionate term of glass holes. And unfortunately, I experienced at least one version of that, which is just sort of, "Well, if you're going to wear this thing on your face with a camera and so on, then at least be prepared to talk about it." And this person wasn't. Yeah, we'll see. 

[0:14:05] SF: I mean, I think the big thing people have been trying to make virtual reality, augmented reality work for a long time, and I think the most compelling example and widespread use of augmented reality might have been Pokémon Go from like eight years ago or whatever when that blew up. And then I haven't seen a really compelling example of virtual reality as well going back to like the Virtual Boy Nintendo failure from like 1990. When I was at Google, I saw some updated stuff that took like an hour to set up with all these cameras and things, and then I was like, "This is only marginally better than what I saw in 1990." 

[0:14:44] GV: Yeah. Well, as I say, we're going to touch on quite a lot of hardware stuff shortly. So, let's just go back to the headlines briefly. The next one is actually in the gaming space, and Software Engineering Daily does cover quite a bit of gaming these days. Yeah, this is quite a big one. Electronic Arts. They're known for the Battlefield franchise, which is obviously direct competition. Well, some people will argue not, but I think it's sort of direct competition to Call of Duty, but obviously very different approaches to the same idea. And they have a bunch of sports franchises as well. But anyway, they're being acquired for it's 50 to 55 billion from private equity, which also includes, I believe, the Saudi Arabian fund. 

I guess what we're seeing here is this is actually coming before Battlefield 6. And there has been a lot of talk about Battlefield 6 because the last two installments of Battlefield were not well-received by the core players. Whilst EA's stock price has been going up, and that would indicate maybe a good time to like sell it, there's commentary here around like, "But why would you sell it before this big release?" And of course, it just maybe looks a bit like execs just sort of saying like, "We're not prepared to take the hit if it goes wrong here. So we need an out." So, it does look good in a sense for gaming that these big private equity are like looking at this as real investments these days. But at the same time, the timing does seem a bit strange in relation to these big blockbuster game franchises. Any thoughts, Sean? 

[0:16:20] SF: Yeah, it's kind of interesting. Actually, if you look at EA's stock price from January to now, it went from $116 a share to 202 probably since the announcement. So, it's really blowing up. And over the last few years, it's kind of been relatively the same kind of bumping up and down and then it has a big spike. But if you sort of project it over time, it's been steadily growing. It's not like it's a completely zombie stock or anything like that. 

But I'm curious, there's been a lot of stuff in the industry with companies kind of either staying private forever or even going from public to private. And I wonder if there's any of that thinking involved on the EA side is, "Hey, let's go back to being private. Run this without sort of the pressure of being a public company and all the things that come along with that." 

[0:17:07] GV: Yeah, it seems to just be a cycle on both sides. I mean, we saw - I mean, the strangest case of this was Dell, where Dell was private, went public, went back to private, and I think is back to public again. It yo-yoed around. In this case, yeah, I think this has to be just down to the top leadership of EA. I think there's obviously a lot of expectations with reporting, etc., when you're a public company. As well as this release, if this goes wrong, I'm sure there's a lot of stock options that are going to go wrong with it. I kind of just see this as - I mean, they've obviously done whatever they've done to kind of structure this or make the private equity buyers feel comfortable that this release is going to go well. 

I mean, I've seen quite a few sort of videos online where they've produced these huge events where they've brought all the press down. I say press, but we're talking like very influential gaming YouTubers to come and play the game. And one guy that I follow, he does a lot of gaming history. He was actually involved with the making of the game. They got him in on various sessions. He wasn't really allowed to talk about what input he was giving, but the fact that they've actually got real gamers in to come and help develop this game. I think there's a lot writing on it, but there's probably a lot more. I don't play these games. For anyone listening, you're probably wondering why I'm not talking about anything to do with the game itself. I don't play them, but I just find it interesting the business behind them. So, yeah. 

Talking of staying private, the next sort of headline was Anthropic Series F. That was, just to check the numbers on that one, 13 billion investment, 183 billion post-money. I mean, these numbers all just sound a bit kind of airy fairy at this point. I think the headline here though is just - 

[0:18:48] SF: Databricks just had their huge round. I think their valuation was 100 billion. I mean, 100 billion is the new billion, it seems like. 

[0:18:56] GV: Yeah. Yeah. 

[0:18:57] SF: It's crazy. 

[0:18:57] GV: Yeah. But again, here we are. No IPO. Why do you think that is, Sean? 

[0:19:02] SF: I would think it seems like they're following a lot of the same strategy that Databricks and some of these companies have followed. If you can continue to operate privately and as long as you can figure out a way. Now, Anthropic is a much younger company than Databricks. But with Databricks, they figured out a way that they can allow their employees to liquidate some of their assets so that their options actually have value. 

There's a lot of advantages to continuing to stay private. The challenge when you're public is, every quarter, your report card is available for the world to see. And there's a lot of things that you have to do as a company to address your board of directors, address your public shareholders. There's a lot of scrutiny with that. And if you want to be able to move fast and kind of independently, you have a lot more freedom to do that when you stay private. 

I mean, I think the stuff like with Anthropic is insane. Their revenue growth. They reported 1 billion in revenue at the beginning of the year in 2025. Now they're five billion. We're like nine months in. They had like a 5X revenue run rate jump since the start of the year. It's insane when you're talking about like billions of dollars. And they have a $500 million run rate in three months with Quadcode. That's really, really impressive. Like it's just numbers I've never seen before in my 20 years of working in the industry. 

[0:20:20] GV: Yeah, I think that's a good call out. Yeah, Quadcode has clearly contributed quite significantly to just that growth. And obviously, we're seeing companies like Cursor that, back in August, they introduced Cursor CLI sort of at least in beta, but it's very much GA now. Yet again, we're seeing the sort of foundation model producers being able to capture so much of the market when it comes to the use cases. 

At the beginning, you'd have companies like cursor who could leverage these models but then really amplify them. And here is Anthropic coming along and just saying, "Yep, we can actually do that too." And I think arguably, to many developers, better. And they've angled in on financial reporting specialty products. Which again, there are other companies, other completely separate companies trying to achieve that and you just think, "Well, you're going to have to have something really special to then beat out the company that's developing the models themselves." Yeah, as you say, Sean, these numbers are insane. 

[0:21:21] SF: Anthropic has done a really good job of positioning themselves as like the enterprise-ready AI native company, whereas OpenAI had a tremendous amount of success with a consumer-facing product of ChatGPT. And I think now they're trying to build that muscle within the company, but they're certainly playing catch-up to where Anthropic started. 

[0:21:40] GV: Yeah, for sure. Just to round it, as you say, at the end of the day, if they can raise the money and they don't need to go public, then that's a great place to be. Stripe has very much adopted that model for its entirety and doesn't look like, at least from what I understand, that's changing anytime soon. I think it's really just when are we going to see some IPOs that make sense. 

I mean, the only one we've seen that's been a kind of success story recently was Figma. They were about to be bought. Adobe tried to buy them, and that was shot down by antitrust. And instead, Figma went public. And that's been held a success overall. I mean I'm sure there was the usual like pop at the beginning and it's gone down a bit. But overall, Figma is a very profitable business at least from the numbers they put out before the IPO. So, that's a great win for the average investor. Because again, the argument here is that the average investor is simply losing out year-on-year because there are no companies that they're allowed to participate in. 

And just to sort of sidebar that, I mean, I was just listening to the acquired - the ACQ2, which is when they interview someone, and it was Toby Lütke from Shopify. I'm a big fan of his. And they reminded the listeners that Shopify went public at 1.25 billion. And that just seems insane. Because, I mean, it's now about 200 billion. But if you think about the fact that an average investor like you or I could have got in at the IPO and 200-X'ed our money. And that's just not possible where we're sitting right now, we're watching just series E, series F at these valuations. Where's the growth for the average person going to come from? 

[0:23:17] SF: Yeah. I don't know. Yeah. Certainly, you got to go work for one of these companies, I guess, and - 

[0:23:22] GV: Correct. Yeah. Yeah. Very interesting. So, just a very final - we're going to hit on time for our headlines here, but just a final one was interesting that sort of hitting the main news now is sort of more I wouldn't say niche AI. But going beyond just, "Oh, here's the next model." And Wall Street Journal touched on Google's Genie 3, which is like a world model, i.e., sort of where the training can be done, say, on videos. But then the actual model itself is where you can - if you imagine almost like a video game that then looks like video, they show an example, say, of driving a boat around Venice. But it really does look like - I mean, it's very photorealistic. Yeah, kind of interesting how these in the background really advancing. If you look at what Genie 2 looked like versus Genie 3, it's like PlayStation 2 versus PlayStation 3 and 4, that kind of leap in those kind of advancements. Yeah. Have you got any sort of touch points with these? 

[0:24:17] SF: I think it makes sense. You got to get some - especially for certain types of use cases or environments, there's got to be some step function jump beyond what you can do with purely large language models and transformer model today where LLMs are really good at patterns and language. But how do they understand the world around them? How do you teach them certain rules or physics? You have to use simulation as a training ground. It's a little bit like using flight simulators to train pilots. It's kind of very similar to like how humans might learn certain skills is can you put these models in certain environments where they have to learn in a sort of simulated environment where you can create what the consequences of doing the wrong things are. So they actually are able to determine sort of the patterns of these new types of environments, these sort of world models as they call them. 

[0:25:07] GV: Okay, so that was kind of the main headlines. We're going to move on to our main topic, which is just looking at where devices and hardware have netted out at the moment. We're going to be looking at the big companies. I mean, this is kind of slightly anchored by Apple and sort of the announcements of iPhone Air. That just sort of sets the tone for like but where are we going with hardware and devices? 

And obviously as software developers, we're trying to think about like what are we actually developing for into the future? Because I think a lot of people are saying, "Well, are we really all going to be walking around with smartphones into the future? Because we're kind of all getting a bit fed up of them." There's some really nice - I don't even want to call them dumb phones. They're like way paired back devices like the minimalist phone. It kind of looks a bit like a Kindle mixed with a Blackberry which can run like any Android app you want but just in a very paired down monochrome kind of environment, which I think speaks to the times of, yeah, we're getting fed up with smartphones. But yet, we're still seeing a lot of developments of hardware. 

I realize we didn't really talk about our predictions from last month, but the one that I made does feed into this, which kind of also helps set the tone a bit, which is I predicted that based on a bunch of hardware bricking incidents from software updates that there would be one of those. And there wasn't - I just said PS4 just as a joke. But what did transpire was that Nest thermostats, which is owned by Google, have decided to completely discontinue their support for their Gen 1 and two devices. 

Now, that's pretty insane because these are thermostats like wired into houses. And they're basically saying, "You know, all that functionality as to why you bought it, with Wi-Fi, and you can like control it from wherever. Oh, yeah. Well, we're going to completely discontinue support for that and you just need to use it like a normal thermostat on the side of your house or side of your wall." There are big consequences here, like which devices and hardware are we adopting, which modalities, etc. 

We're going to look at Apple, we're going to look at Meta. Again, we have obviously talked quite a bit about them in the headlines. We're going to look at Google, we're going to look at Snap, and just kind of like look at what's going on here. The first company as sort of talked about is Apple. I think just sort of hit the high notes here, which is iPhone Air has been released. People are a bit underwhelmed by it. If we then look at where are they trying to go with hardware and devices? Again, developers thought, "Okay, should I be developing for this? Well, the Vision Pro, that was a sort of 3 and 1/2 grand failure. It didn't really get great adoption. It was a very clunky device. I saw someone wearing it in a bar in New York and it just looked ridiculous. I mean, I think they were obviously just wearing it to show their friends, and the friends were using it in a bar, but you would not take this thing to a bar normally. That's pretty strange. And they kind of canned the Vision Pro ultimately. 

And then if we actually look at where they're quite strong right now, it's actually the MacBook Pro, which is with their Apple Silicon. Or I mean the MacBook Air as well, Apple Silicon basically, where they've moved away, and they decide to double down on their own chips, which I mean I think has been a huge advantage for software developers as much as - especially running models locally as much as video editors, etc. And we also probably will touch on AirPods, which is sort of their secret breakaway success products, which they obviously - I say secret. Just I think people maybe realize just how much money they probably make from AirPods and AirPod Pros. Yeah. I'm going to set the scene there with Apple. What do you think on where is Apple right now, Sean? 

[0:28:38] SF: It seems like on the laptop front, they're making the right kind of moves where they're investing in making their laptops be able to run AI models and AI workloads more efficiently locally, which I think makes a ton of sense. If that's the way the world is going, and you want to be able to do that from a development standpoint, or even certain applications, we'll have certain models probably running within them at some point, then you want to have the hardware to be able to support that. 

I think where Apple, as well as a ton of these companies that are focused on sort of next-generation hardware is they're all kind of trying to figure out what is that next iPhone type breakthrough of hardware. And we haven't really had that. People have tried different things. We already talked about all the glassware sort of failures. Is that the next thing? I don't think so. Maybe I'm wrong. There's humane that tried the AI pin where I actually kind of liked some of the thesis behind their idea, but the hardware ended up being kind of clunky, and it was a little slow. 

[0:29:44] GV: What was that? What was the AI pin? 

[0:29:45] SF: Oh, yeah. So, it was this little pin that you could wear. And it had no screen, but you could talk to it. So, it was voice-activated. And then if you needed a screen, it could kind of - you could use your hand and it would draw with like a sort of laser pointer on your hand to show you some representation. But it also had cameras in it, so you could just point at it. The whole idea was to create some sort of passive piece of hardware that wasn't as intrusive as getting your phone out to do something when you wanted to snap a picture or you wanted to capture a moment at like a concert or something. You're not just like looking at all those things through a phone and not experiencing themselves was kind of the idea behind it. And I like that, but they weren't able to deliver on it at least yet in terms of being able to create those experiences. 

So, a lot of these companies, I think, are struggling to try to figure out what that is. And I wish I knew. I don't know. But I think you raised an interesting point about the AirPods. I think with the AirPods, you see them everywhere. And I think as somebody who maybe you don't know if anything about headphones, and especially headphones that don't have something to like really keep them in your ear, how hard that is to like really nail with people where you have to be able to fit this to all kinds of different people's ear sizes and have something that kind of just like universally works. It stays in and connects to any device very quickly. That's a really, really hard thing to pull off. And I think Apple did it. So, they did it spectacularly. And I think that's part of that reason why AirPods have been so successful. 

[0:31:13] GV: Yeah, for sure. Yeah. I mean, I own a pair of the last-gen AirPod Pros. They have done pretty well. I mean, the case on them has always been the issue. They've got these little connectors for the charging. And in this climate, which is very humid, these connectors basically corrode very fast. I've gone through like two cases just on that basis, even trying to clean the little connectors with alcohol and all that kind of stuff. 

And then to your point, yeah, I've tried, wife has a pair of from another manufacturer and they kind of basically broke after like a year. I've tried the Bose. They're kind of quiet comfort earphones as opposed to headphones. I'm wearing a pair of the headphones right now, but these are the earphones that we're talking about. And to your point, they're actually quite difficult to get in your ear. And that puts me off thinking that could be a daily modality for anything beyond like - yeah. It's a tricky device to get right. Apple seem to have got it pretty right. And then they added in the health aspect of it. Basically, the latest gen can function as a hearing aid, which is pretty spectacular. 

I think looking ahead, it's like, "But what could we do more than that?" I mean, the one that comes to mind to me is like translation. Could you walk around Japan and someone's like talking to you and it's just being filtered through and being translated real time? You're sitting in the UN or something like, yeah, they have their earpieces with - 

[0:32:40] SF: Yeah. I mean, the universal translator to some extent. That's what Google - when they first launched their Pixel Pod, or I forget exactly what they were called, but I had a pair. You were supposed to be able to like kind of tap it and then you could engage the assistant was built into it and you could ask for translation. Of course, the demo looked amazing. But in reality, it was a lot clunkier and it didn't work very well. 

I mean, part of the challenge with a lot of those things is they have to rely on a network connection and then it ends up being slow. But there are things now with the - there's a bunch of people who've worked on compressing Whisper models so they can run on an iPhone. Could you get some of this translation down into a model that can run directly on the phone so you can do most of that translation there and then you don't have to round trip to the network? I feel like that is certainly something that's not that far away. 

[0:33:34] GV: Yeah. So I think the thing to highlight there with Apple generally is just that they do have a hold on this piece of hardware that is pretty ubiquitous these days. And I think, arguably, they've got the best hardware kind of in there, which is pretty critical to if you're trying to then deliver these as you're kind of calling out, Sean, very low latency experiences or use cases using AI. Because either the AI component needs to be done locally on your device and then it's just Bluetooth to the device. Or as you say, more likely, it needs a network connection. And then you're talking like, "Well, that's actually where 5G would help." One of the few kind of maybe consumer use cases for 5G at the moment would be sort of this ultra-low latency roundtrip stuff. Yeah, interesting to think about. 

If we just kind of move on what are the other players doing? Well, Meta, we have touched on them quite a bit, so we probably won't dive into them quite so much. But they had the Ray-Ban glasses. And then the new glasses are, I think, also Ray-Ban. It's just that this is now the AR version. Or they're Oakley. I can't remember which manufacturer they're teaming up with for this new set of glasses. But they are going all in on the glasses stuff. 

And as we've touched on, that seems like a slightly interesting modality to be really focusing on, especially for like the adult market. And what other use cases could we see? Well, we kind of touched on them. Things like networking, being able to like know who's around you or sort of head-up display-like experiences for sports. But I don't know. I mean, just as a sort of software developer, I wouldn't see these get me terribly excited. I'd probably be thinking more about the audio side of things where you can deliver these more like sort of subtle experiences as you're walking around. You can like still enjoy the environment around you unhindered visually. It's just that you're getting a bit of information audibly perhaps. But Meta see it differently, I think. 

[0:35:37] SF: Yeah, I do think that there's a lot more we can do with audio medium. I think we'll see that as well. People have been trying to do that for a long time with the Google Assistant, and Siri, and Alexa and all these. I'm setting off everyone's devices right now. But this didn't really work that well. It's kind of a letdown. And you end up using them to set timers for cooking and asking what the weather is. They kind of end up being not that important. They're useful, but you could live without them. They don't have that killer use case. 

But if you can just converse, and I know people are studying to do this with if you turn sort of ChatGPT into like the audio model where you can just sort of have a conversation with it, then it becomes like an interesting way to learn certain things. And we just haven't had historically the powerful enough models to be able to do that properly where audio can become the engagement medium. 

We created keyboards and we created all these input devices not because they were convenient to people. We taught ourselves to use those things. We taught ourselves to use a mouse, taught ourselves to type so that we could communicate with the computer. But now, are we in a place where we can finally make it? You rely on sort of the ways that we are used to communicating, gestures, speaking, all these types of things and use that as the medium as input into a computer. 

[0:36:58] GV: Yeah, absolutely. Looking at Google, I mean, they for a long time were getting quite big into, yeah, devices, hardware, connected home. Touched on the Nest earlier. But they are the one that kind of - it's odd because they've obviously got a lot of money, thrown a lot of resources at hardware. I know people that have worked in hardware at Google. But at the end of the day, I don't think they've made a lot of money from hardware. This is just usually a sort of conduit to deliver Google services basically, which is quite different to Apple, who do make money on hardware, and they do make money on services. 

Yeah. Google, we've just kind of seen them fade away a bit, I would say, when it comes to -  we're not seeing any breakout product or sort of experimental product like we've seen with both Apple and Meta Pixel phones. They're making a big song and dance about the AI capabilities of Pixel phones. But again, that's not category defining. That's just, yeah, you have a great phone. Obviously, they're trying to define it as this is the best Android phone. 

[0:37:55] SF: And I think they mostly do that strategically to push the overall Android market. I don't think they're trying to own the Android hardware market. The number of Pixel phones they actually make is still relatively small. It's more like can we create a really, really great Android phone that's all our stuff and show what the capabilities are? So that way, anybody building an Android phone can't have excuses that blame the operating system or something like that? Yeah, they're not really trying to own that market. 

I wonder, part of it, maybe they were burned too much on some of the hardware investments they made in the past and they didn't really work out for them. And I think in some ways they're kind of focused on the right things. If you look at what they're doing in the AI market, they're focused more on models and infrastructure for AI, which I think they're really, really good at. They're very, very good at scaling efficiency, and they mostly have used hardware as a means to I think strategically protect their business. Or it's like they can essentially afford to give it away or sell it at a loss because they'll make it up on the other end of selling ads and getting more people on the internet. I think it makes sense that they probably are not overinvesting in this market right now. 

[0:39:08] GV: Yeah, I totally agree. I think it's just as interesting observation of where they were with hardware. As you say, there's quite a few projects they had for the last 10 years on that front, consumer hardware, but not investing heavily in it now. And it does make sense sort of in the landscape. 

Finally, Snap, which is a bit of a wild card here. But I've been reacquainted with Snap only recently, but kind of then realized just how much effort they're putting into what were called spectacles. And then I believe they're going to kind of morph into just specs, which is a sort of branding thing. But in June this year, they said that there's going to be a huge update coming out. I believe that's coming quite soon. And they did release very recently Snap OS 2.0. And that's very much thinking around - and that's the developer platform. And how can developers interface with Snap? But very much how this very much is setting it up for whatever is coming with specs as well. 

It's very interesting. Because I heard a podcast with Evan Spiegel, the CEO, co-founder/ founder. I mean, the way he was talking was he sort of said like this is the future of the company. Now I know he kind of has to say that. I mean, it is a public company. So he's got to like really get behind whatever they're doing. But I guess it was interesting that this device was what he was saying the future of the company was. 

And again, I think the demographic is interesting because the demographic, again, as I've learned more recently, is younger. It's a sort of social network that is - or social platform, whatever you want to call it, that is more accepted for high school age kids to use. And whereas a lot of parents are not okay with high school kids being on these platforms, that seems to - and even younger, I believe. But this is one that sort of seems to be on the okay list for a lot of parents actually. 

It's kind of interesting that company that has that demographic are going all-in on these glasses where AR is a huge thing and they really push this idea of AR with your friends. That does open up some really interesting possibilities software-wise. You touched on Pokémon Go, Sean. I mean, I can imagine probably a whole bunch of kind of collaborative AR experiences that you might be able to have with glasses. At the moment, the biggest problem I see with it is just it looks an incredibly clunky device. And you wouldn't wear this thing outside of your home. This is kind of almost like VR glasses, just are a bit nicer. 

[0:41:39] SF: Yeah. I mean, I think the standard is so high to be able to build a device like that that people would be willing to wear in public on a regular basis. I haven't seen anything that's like even close to what the general public would accept. Now it's a reasonably acceptable behavior that people are like looking at their phones when they're public and so forth. But to be wearing something and be okay with it, it is something that draws attention to you and also being okay with that. That's just a smaller audience than what you ultimately are probably going for when you're a company that's investing in this. You want something that's universally appealing to people. That's a hard thing to pull off if it's not just a technology, but it's also part of how people see you and it's part of fashion. 

[0:42:27] GV: Yeah, as you say, even Meta's Ray-Ban band glasses still looked a bit kind of like something that isn't a pair of actual sunglasses. Yeah, it's going to be interesting just - I mean, I don't think I've seen anyone wearing them out and about. But I've definitely heard of people saying that they do wear them, "Oh, because they've got like little speakers as well." But yeah, I guess we'll see. 

Kind of wrapping this one up, if we were to like put together what would be the ideal combo here is sort of something along the lines of if you took Apple's UX attention, if you added Meta's experimentation, Google's AI, and then the sort of let's give the Snap sort of long-term vision because they seem to be the only ones actually committing to quite a specific thing and saying this is it. That seems to be where the ideal device might appear from. But that's what makes it interesting. We've got all these players taking quite wildly different approaches. 

Yeah, moving on to our favorite bit of the show, Hacker News highlights. I feel there was like so much over the last month. I really struggled to pick what I was going to bring up this time. There were too many. The one that definitely scratches my itch when it comes to developer off the deep end. This was posted by Bogdan the Geek and that I believe is also that was the blog of the person. This person posted it. It was called hosting a website on a disposable vape, which was just fascinating. 

As this person points out, it seems incredibly paradoxical to call a vape disposable when it has a 24 MHz Cortex M0 processor, 24 kilobytes of flash storage, 3 kilobytes of static RAM, and a few peripherals, USBC. That doesn't sound disposable to me. And I think he was sort of hinting he would not want to be a big tobacco lawyer in a few years. He's probably going to have to explain what on earth these devices were being branded as disposable. 

At the end of the day though, this person was able to host a website on this disposable vape when it was at the top of Hacker News. Admittedly, it was crashing as you'd expect. I did manage to access it today. It does work. This person had posted the blog post as well obviously somewhere else. He hosted on both the vape and a regular server just to kind of make sure people would read the article. 

I mean, absolutely fascinating that this is even possible. And I like the one quote from him that I'll just sort of pull out. I just read out the specs a minute ago, but he said, "You may look at those specs and think that it's not much to work with. I don't blame you. A 10-year-old phone can barely load Google, and this is about 100x slower." I on the other hand see a blazingly fast web server, which is just I love. This person can look at that hardware. And sure enough, they were able to get like very low latency ping out of this thing, which, yeah, just very ridiculous and fun. 

[0:45:06] SF: There's so many of these ridiculous projects online. I love it. I mean, we're giving like Meta hard time and other people had a hard time from watching ridiculous products. But if someone put that on Hacker News, I'd be all over it. I love people using their brain power and their time for some sort of ridiculous side project. And actually in that vein, the Hacker News article that I had was this article about someone built a way to play the game Snake in your URL address bar. So you can check it out. It was posted by I think - I don't know if you - Makott or Makati is the person's name. Snake is the game where you're moving the snake around and it keeps getting longer and you don't want it to collide. And basically, they jam that into - 

[0:45:50] GV: I'm playing it right now. This is so fascinating. Yeah. I tried to open up on my phone yesterday and it didn't work. But yeah, I'm playing it now on the desktop. 

[0:45:57] SF: Yeah. I think you have to do it on a computer. 

[0:46:00] GV: Yeah. Yeah. This is hilarious. Yeah. If you remember Snake on a Nokia, a Nokia screen was kind of square. So, the obvious difference here is it's a very long rectangular shape that you're trying to navigate within. But I mean, hey, it works. It's pretty amazing. 

The other one that caught my eye, this is posted by a user. And so, it is the 3,000-year-old story hidden in the @ sign. @ sign, what we think of for emails, so and so @blahblah.com. Turns out that the history of this is the @ was - it meant ampersand, which is that's a unit of measurement that the Greeks used. They'd sort of say, "I want five ampersands of flour or something." 

And it then got kind of used by accountants apparently, and that's why the @ sign was even put on a keyboard in the first place basically back in like the '70s. And of course, then when ARPANET was being developed, that's when the person creating email effectively had to look on the keyboard and be like, "Hey, what shall I use to help people so and so @ so and so?" And that was the sign that was picked. 

It sort of has absolutely zero history in terms of that actually being anything to do with directing you at something. It was to do with measurements initially and accounting. And it's now become obviously synonymous with email. I never knew that. And that was just something that somebody posted. Very interesting. 

[0:47:26] SF: Yeah. And the last one I had, which is very thematic for our conversation around hardware today, is that someone has built the first iPod-controlled game called RidePods where you use your head movements to control like a car or a motorcycle thing as you drive around. 

[0:47:43] GV: Oh, wow. Yeah. Cool. Looks a bit like a motorcycle thing. And yeah, very kind of like PlayStation era. But, yeah. Well, just talking about AirPods. Yeah, exactly. Being this modality. Well, exactly. They got, I guess, motion sensors as well. 

[0:47:58] SF: Yeah, I didn't realize that. I guess it must be maybe that's how they kind of control sort of the level volume or something like that as you move around. 

[0:48:06] GV: In that sort of vein, this is just from memory. There was someone who posted about actually the fact that there is a sensor in the hinge of a MacBook that tells you the exact degree of open and shut. It's very difficult to get at it from the API apparently, but someone managed to get there. And what did they do? Well, of course, they hooked it up to the sound of a creaking door. Most useful thing you can do. Open it and obviously show the degree on the screen, but also make it sound like a creaking door. 

[0:48:34] SF: Hacker News needs to start having - or maybe we start this. Maybe Software Engineering Daily needs to do this, is the ignoble prize for a random engineering hobby projects on the internet. 

[0:48:47] GV: Yeah, we should do like the SED News or SC Daily Awards at the end of the year or something like that. I like that. 

[0:48:51] SF: Yeah. Most ridiculous use of engineering talent. 

[0:48:55] GV: Yeah, absolutely. Just wrapping up on the Hacker News. This was a good just a sort of feel-good story which was also shows the power of Hacker News. It was posted by just Skyfall who ran - I believe this person ran a online coding academy for like thousands of children and teenagers. And they had a Slack bill originally of about 2K a year or 5K a year. Not nothing, but like very reasonable to handle this number of people. Slack popped up and said within a week or 2 weeks, your bill is going to be 195k. And of course, they just said, "Well, this is impossible. We're going to move to Mattermost." 

But luckily - I mean, I say luckily, I mean, it would have been probably overall great if they could have moved to Mattermost. But that's a huge undertaking. So, luckily, hit the top of Hacker News, got a lot of attention, and the Slack CEO reached out and the update to the post was that a good outcome had been reached and this was no longer a problem. Good to see that Hacker News can really influence some good things outside of these crazy fun projects that we also find. 

We're hitting on time. Predictions as always. What do you think you'll see across the next month, Sean? 

[0:50:04] SF: Yeah, so I was trying to think about this. Salesforce has their big annual conference here in San Francisco in a couple weeks, Dreamforce. This is somewhat ridiculous prediction. I'm sure they're going to make a lot of AI announcements, but I'm going to go out on a whim and say that Salesforce launches their own large language model, foundation model at Dreamforce. 

[0:50:22] GV: Interesting. Interesting. Foundation model from Salesforce. Okay. 

[0:50:26] SF: Yeah. 

[0:50:27] GV: What is my prediction? It's funny. I haven't even vaguely thought about this. Well, I'm just going to pull from Hacker News this time. I'm going to predict that we see someone one-up the disposable vape. Someone manages to host a website on something that's even more esoteric than a disposable vape. I think we're seeing people constantly trying to one-up each other on some of this hardware stuff. Yeah, that's a completely random but fun prediction that somebody will manage to get a website running on something that's even less than a disposable vape. 

Well, with that fantastic prediction, it's time to wrap up. Thank you so much for tuning in to SED News. Great to see you again, Sean. 

[0:51:10] SF: Yeah, you too. Always enjoy these conversations. We covered a lot of ground today. 

[0:51:13] GV: Yeah. Yeah, we did. Yeah. Hopefully, we'll cover just as much next time. So do join us next time on SED News. We'll see you then. 

[0:51:21] SF: See you. 

[END]