Footgun Cavalcade

[JM]: Dan, are you familiar with the 2003 "Bring Me To Life" song by Evanescence?

[DJ]: Of course I am, because as a 21-year-old in the midst of university, Evanescence was exactly the sort of music that I wasted a lot of time listening to.

[DJ]: Right up there with Papa Roach.

[JM]: I was about to say, well, "waste" is a strong word, and then you mentioned Papa Roach, and then I'm like, okay, I'll let you have it.

[DJ]: Cut my life into pieces, Justin.

[DJ]: That was my last resort.

[JM]: Well, I mention it because for some reason, this song, even though it is well over two decades old at this point, seems to find its way into random covers that I come across.

[JM]: And I just wanted to highlight two of them in particular.

[JM]: One of them is a Gregorian-chant-style cover of "Bring Me To Life", and it is fabulous.

[JM]: And you should drop whatever you're doing and tap on the link that's in the show notes so that you can listen to this marvelous and very bizarre creation.

[JM]: and then immediately tap on the next one, which is infinitely more frivolous, ridiculous, and silly.

[JM]: And that's precisely why it's probably my favorite of the covers that I've come across, because this version of it is played entirely via the Otamatone.

[JM]: And if you're asking yourself, I'm sorry, what word did Justin just spew out?

[DJ]: I am asking myself that, yes.

[JM]: An Otamatone is this completely absurd instrument that comes from some enterprising folks in Japan, and an "otama" is...

[JM]: a Japanese word meaning ladle.

[JM]: And it's called that because this particular instrument, and I'm using scare quotes because, you know, you really shouldn't think of this as a proper musical instrument...

[JM]: This instrument kind of looks like a musical note as well as looking like a ladle.

[JM]: And the part at the bottom is the part that you squeeze.

[JM]: And then the part at the top has like a little face with a mouth on it and it opens and sound comes out.

[JM]: that is somewhat reminiscent of, say, a kazoo so this is the kind of toy you should give to people who you don't particularly like who have kids, because their kids will play this instrument at home and presumably annoy their parents to a...

[JM]: fantastic degree.

[JM]: But in any case, I digress.

[JM]: Go check out this absurd cover of "Bring Me to Life" by folks who very creatively used several Otamatone instruments in harmony together.

[JM]: There's a music video that goes with it.

[JM]: It's fabulous.

[JM]: You're welcome.

[DJ]: I wonder if Evanescence could have imagined all those years ago when they were recording "Bring Me to Life" that someday it would have such an impact on the world at large.

[JM]: I can't imagine this is something that they could have foreseen.

[JM]: This song has even been used as the official trailer song from some game in the Elven Ring franchise.

[JM]: So it is just like this gift that keeps on giving all these years later.

[DJ]: So I just want to say another gift that keeps on giving, Justin, is how much of a boomer you sound like when you mispronounce Elden Ring as Elven Ring.

[DJ]: Now, I admit that as fantasy role-playing video games go, Elven Ring does seem like a more sensible name, but that's not what it's called.

[JM]: Not that I expect anyone to believe me, but I did know that.

[JM]: This was just a on-air faux pas that we are letting roll because we're doing it live.

[DJ]: But just thinking about the longevity of "Bring Me to Life", I sort of get a kick out of imagining that someday Amy Lee will be on her deathbed.

[DJ]: I don't get a kick out of that, and I hope it doesn't happen for many more years, but I'm just imagining that at the end of her life, Amy Lee will be looking back on her accomplishments, and then she'll hear someone's, I don't know, virtual reality, whatever we've replaced cell phones with at that point, playing a weird little version of "Bring Me to Life", and she'll just think, ah, yes.

[DJ]: I've accomplished everything I set out to accomplish.

[DJ]: It was all worth it.

[JM]: Indeed.

[JM]: In other news, the Anthropic Claude models known as Fable and Mythos, which were made available and then quickly yanked, are now available again.

[JM]: But after today, and by the time this episode airs, it will only be available via metered extra usage credits.

[JM]: Which means this was available to Claude subscribers from the time it was restored a week ago,

[JM]: for seven days before going behind the extra usage credit paywall.

[DJ]: Did you have a chance to try this infamous model in the brief period of time that it was generally available?

[JM]: I did indeed try it briefly.

[JM]: I don't think that I used it enough to form any kind of opinion as to whether it is significantly different or better or worse or anything relative to, say, Opus.

[JM]: So I don't think that I'm going to miss it since I never really got a chance to use it long enough to form an opinion about it.

[JM]: What about you?

[DJ]: Yeah, same.

[DJ]: I ran a couple prompts through it within the last week, and I don't know if I noticed a substantive difference.

[DJ]: And what I wonder about is for normal person chatbot sort of use cases, the sorts of things that we were probably using Claude for...

[DJ]: If we've reached this point of diminishing returns where the frontier providers keep coming out with these new models and they're good and then they come out with another one and it's good and then there's another one and it's good but like...

[DJ]: It's reaching a point where for the sorts of stuff you want from it, it's hard to imagine it getting so much better that the entire experience of using the software that's powered by the model is different.

[DJ]: That has happened a couple of times in the last few years, right?

[DJ]: We've talked about the sort of inflection points for LLMs in general around, I want to say the GPT-4 era was one of them.

[DJ]: And like in November of last year is when a lot of people talk about as being that's when...

[JM]: Opus 4.5.

[DJ]: Yeah, and that's when these so-called like coding agents really started to gain traction where it was like, oh, an LLM is not just capable of like generating a computer program.

[DJ]: This is actually becoming a useful way to generate software.

[DJ]: I am really curious if we will see fundamental changes to the models themselves that keep transforming how we use them.

[DJ]: I guess what I'm saying is I wonder if eventually, when is an LLM good enough, essentially, right?

[DJ]: Certainly these labs have their reasons for answering that question with never, and maybe that's because, although I have a hard time saying this out loud, they're trying to create

[DJ]: the machine god, right?

[DJ]: Like, they're like, well, the model's not done until we've created this so-called artificial general intelligence.

[DJ]: All right, whatever.

[DJ]: But for the rest of us that are just using chatbots and coding tools, it kind of feels like these models are getting, at least they're good enough, or merely improving the model does not solve the problems...

[DJ]: that we have with them is maybe another way to put it. It doesn't matter how current the model's training is or how good it is necessarily at producing output from that training. There seem to be these persistent limitations, and I haven't gotten the sense, at least I haven't heard people talking about this, maybe because they haven't had a chance to use it...

[DJ]: I haven't heard people saying, for example, that Fable confabulates way less than the other models.

[DJ]: "Oh, Fable's amazing because it doesn't just make up totally wrong stuff and then present it to you in this confident tone."

[DJ]: That would be a massive transformational shift in how we experience LLMs, but I don't hear anyone saying that.

[JM]: No, I haven't either.

[JM]: And we have been talking for quite a while about the diminishing returns that we seem to be getting from these models.

[JM]: I don't know if you really were to look at it that from this November 2025 inflection point that you just mentioned...

[JM]: From then until now, I don't really know that I would say there have been substantive, like truly substantive improvements in any of the Frontier models that are really worthy of note.

[JM]: So, for example, if we're talking about Claude going from Opus 4.5 to 4.6, hmm.

[JM]: 4.7.

[JM]: Yeah.

[JM]: Okay.

[JM]: I guess it's a little bit better.

[JM]: 4.8.

[JM]: Same.

[JM]: We are just seeing these very small improvements.

[JM]: Sometimes they even feel like almost regressions, or at least that's what I see.

[JM]: Some people talk about where they say like, "You know, actually, I think I got better results from the previous one."

[JM]: So obviously a lot of that is anecdotal.

[JM]: Presumably Anthropic is not in the business of shipping models that are worse than the ones they shipped the month or two prior.

[JM]: But, you know, here we are with Fable and Mythos 5, which have been anointed with this major version number, right?

[JM]: They're not just, it's not just a 4.9.

[JM]: It's got a new name and a bigger number.

[JM]: And presumably that is meant to signify that it is indeed a significant improvement.

[JM]: And I don't know that I'm seeing that borne out by a lot of responses or impressions that I've seen from people.

[JM]: It does seem like these models might be better at surfacing security vulnerabilities.

[JM]: And if that's the kind of thing that you are doing, then...

[JM]: great, you have a better tool in your toolbox to do that work.

[JM]: But this release for me really does underscore what feels like a very noticeable trend of diminishing returns and they will continue and have continued to work on their harnesses

[JM]: and all of the tooling around the model to better utilize it, to make it produce better results for end users.

[JM]: But the models themselves, I get the distinct impression, are not really improving all that much.

[JM]: And that is concerning given how much money is being invested into them.

[DJ]: That's true.

[DJ]: Although another way to look at that is I wonder if there's a light at the end of the tunnel, because really everything we're describing feels very familiar.

[DJ]: It's the way that most innovative products develop.

[DJ]: We've been talking about this in the realm of smartphones for years at this point, right?

[DJ]: Where when the iPhone first came out, every year's iPhone release was

[DJ]: so substantially different and better that it was really very exciting.

[DJ]: Whereas a sort of theme that I've both felt and seen in the tech world over the last five years at least is that every year's new iPhone is like, eh.

[DJ]: Like, yeah, it's fine.

[DJ]: But last year's was fine, too.

[DJ]: Or even better than that, like last year's iPhone was really good.

[DJ]: And this one is also really good.

[DJ]: Camera's a little different.

[DJ]: Sometimes there's like a new thing, but the product is mature.

[DJ]: And I wonder if we're seeing something similar in the large language model powered software space.

[DJ]: as new as that space is.

[DJ]: It sort of makes sense that after this certain period of furious growth and change, we'd start to see things beginning to stabilize.

[DJ]: Now, having said that, like I said before, I think there are lots of incentives, many of them bad, for the big AI companies to keep investing insane amounts of other people's money and all the computer hardware that we would like to be putting in our gaming PCs

[DJ]: into improving models.

[DJ]: But on the other hand, if their businesses start to reach profitability on the backs of the current or near future generation of LLMs, I almost wonder if maybe some of the insane like overinvestment, you know, what people have been calling like an AI bubble will start to diminish.

[DJ]: And that would be a good thing.

[DJ]: Like, I think it would be very nice for a lot of the hype to drain out of this industry and have us just be left with like, yeah, I mean, there's good software for like a chatbot or a coding agent.

[DJ]: And yeah, there's still lots of problematical second order effects with all of these things.

[DJ]: But I mean, I've been desperate for a long time already for so-called AI to just become like

[DJ]: software, to just become normal technology. I think in fact there is an interesting newsletter literally called something like "AI is Normal Technology", and like that's the world I want to live in instead of it being this, and like instead of I think there's an awful lot of -- and some people would argue it's warranted -- but an awful lot of melodrama to Anthropic being like,

[DJ]: "We've created the most powerful model in the world."

[DJ]: And then the government being like, "No, you can't give that to anyone."

[DJ]: "It's too powerful."

[DJ]: And they're like, "Okay, well, we don't like it, but we won't give it to you because it's so good."

[DJ]: "We'd love to give this so good thing to you because it's so good, but we can't."

[DJ]: In a way, and I'm not saying this, I think this is true, but like it's almost the ultimate marketing stunt for Anthropic is to be like, we built a model so good that the American government won't even let us show it to you.

[DJ]: Like, okay.

[JM]: You couldn't buy that kind of marketing.

[JM]: Absolutely.

[JM]: It's perfect for them.

[DJ]: Well, I mean, in the current administration, you probably could buy that kind of marketing, but point taken.

[JM]: Yeah.

[JM]: And this whole idea in their announcement that going forward, we are going to be partnering with governments regarding frontier LLM security.

[JM]: It's just more marketing.

[JM]: This idea like these things are so powerful that don't worry, we're going to be partnering with governments going forward so that we will hopefully avoid these kinds of disruptions where we introduce a model and then it gets yanked because some government is concerned about it.

[JM]: And then we

[JM]: do a bunch of backroom negotiation, and then we ship it again.

[JM]: And I mean, it's just the absurdity of this being the kind of regulation that we're seeing with large language models, because there's a lot of potential regulation that I think is warranted, but this is like probably dead last on my list of the kind of government regulation that I want to see in the large language model space is this...

[JM]: saying, oh, yeah, this is kind of powerful.

[JM]: We're not really sure we want everyone to have it.

[JM]: Like, oh, great.

[JM]: That's when you're going to step in?

[DJ]: Well, not to be cynical, but yeah, of course.

[DJ]: Like, of course, regulation is going to move faster when it's considered a matter of national security than when it's protecting consumers or protecting people's well-being.

[JM]: Very true.

[JM]: Doesn't mean I have to like it, but very true.

[DJ]: No, no, indeed.

[DJ]: Well, we should move on to a happier topic like the price of computer components.

[JM]: Indeed.

[JM]: Following up on the topic of RAM prices, I came across this amazing screenshot of someone's interaction with ChatGPT.

[JM]: And it seems the query was regarding a list of computer components that the human was trying to run by ChatGPT to see whether or not

[JM]: this was a good combination of components to build their new computer.

[JM]: And the response this person got back from ChatGPT was the following.

[JM]: "That is a very solid build, but I spotted one thing immediately."

[JM]: "The RAM price is wrong."

[JM]: "Your screenshot shows G.Skill Flare X5 32 gigabytes DDR5 6000 CL36 (which is a description of the kind of RAM) for $419.99."

[JM]: "That has to be a pricing error or a PC part picker picked some weird seller."

[JM]: "32 gigabyte DDR5 RAM should be around 80 to $120, not $420."

[JM]: "If you actually paid $420 for 32 gigabyte RAM, I'd call the police."

[DJ]: Justin, do you want to tell him or should I tell him?

[JM]: I feel like the next line in this response should have been ChatGPT saying, we're all trying to find the guy who did this.

[DJ]: Yeah, seriously.

[DJ]: The reason this is so wonderful, of course, is not just the ChatGPT doesn't know what year it is, but that ChatGPT is the reason that RAM is so expensive.

[JM]: Right.

[JM]: And the guy who did this is the CEO of the company that created ChatGPT.

[JM]: So the guy who did this is your boss, ChatGPT.

[DJ]: You can imagine if, not that this would ever happen, like Sam Altman is giving a demo of ChatGPT and someone types in this query and GPT gives the response on the screen and as it's printing, and then GPT just keeps going on about like,

[DJ]: That price is totally insane.

[DJ]: Like, it must be ridiculous.

[DJ]: And Sam is sitting there going like, um, um, uh, uh-huh, like pulling out his collar and just being like, like, like doing the zip it thing across his lips.

[DJ]: Like, that's enough.

[DJ]: That's enough, ChatGPT.

[JM]: We should find whoever's responsible and tar-and-feather them.

[JM]: "Shhh! Ixnay on the ard-and-feather-tay."

[DJ]: Well, I feel better about not being able to afford computer hardware anymore.

[JM]: That makes one of us.

[JM]: Well, speaking of disappearing hardware, Sony has announced that they will no longer produce physical game discs for PlayStation as of January 2028.

[JM]: And as of that date, new games will be available on PlayStation Store and at retailers in digital formats only, Sony said in a blog post announcement.

[JM]: And it seems that during Sony's fiscal year, ending March 31st, 2026, digital downloads accounted for 78% of full game unit purchases.

[JM]: And I have thoughts regarding Sony deciding that they are going to end the sale of physical copies of PlayStation games and this move toward a licensing-only sales model.

[JM]: The bottom line, as far as I'm concerned, is that Sony wants more profit and control over game prices.

[JM]: So don't believe them, for example, when Sony says that the reason is the general preference for digital media or to adapt to consumer trends.

[JM]: Don't believe them when they tell you that it's about anything else, because in the end, it's really about more profit and control over game prices.

[JM]: I saw someone post a screenshot in a Reddit thread of some game in the Metal Gear franchise that on the PlayStation Store sells for $70.

[JM]: And in this same image on the right-hand side is clearly a screenshot of the same game in disc format being sold on Amazon for $25, or significantly less than half the price that

[JM]: Sony's charging when you buy the digital download on the PlayStation Store.

[JM]: That is why they're making this change.

[JM]: And even if 78% of people are buying digital downloads, that still means that 22% of people prefer to have the physical disc version of the game.

[JM]: That's a significant number of people.

[JM]: This is not like 2%.

[JM]: This is almost a quarter of all purchases.

[JM]: And really what this ends up resulting in

[JM]: is the end of ownership for PlayStation games.

[JM]: And sadly, this is just part of a trend we've been experiencing for a long time where increasingly the concept of ownership has just been chipped away at whether it's music, movies, games, books, you aren't buying these products anymore.

[JM]: By and large, you are merely licensing them.

[JM]: And that license can be yanked anytime the seller wants to.

[JM]: And sadly, the current generation of young folks may not realize before it's too late that there was another way to do this, that you used to be able to just hand a book to your friend and say, here's this book that I enjoyed.

[JM]: You might enjoy it too, or a movie or a music album or a game.

[JM]: So this sucks.

[JM]: And unfortunately, not much we can do about it, except -- as I am wont to say -- to vote with your dollars.

[JM]: And if you decide that your dollars could make a difference, find another way to spend them than to give them to the Sony PlayStation store.

[DJ]: This is one of the reasons my preferred way to play video games is on a PC, because there are still ways to get your hands on legitimately purchased yet DRM-free copies of video game software.

[DJ]: This whole thing you mentioned about how digital media has eroded the whole concept of being able to own media is something I think about a lot, and it's been around for a long time at this point.

[DJ]: As you say, this is just one more example of an ongoing trend.

[DJ]: And I'm a big fan of pushing back at that trend.

[DJ]: You mentioned something like books.

[DJ]: And aside from the annoyance of just being an individual and realizing that you're renting all your media instead of owning it, there is actually really something lost...

[DJ]: in a world where you tell someone, Hey, I just read this book and it really meant something to me.

[DJ]: I think you should read it too.

[DJ]: And instead of just like handing them your copy of that book, their only recourse is to go purchase...

[DJ]: Let me put that differently:

[DJ]: Go *rent* a version of the book themselves, because you certainly can't share the ebook that you bought from (God forbid) Amazon with them.

[DJ]: That's not allowed.

[DJ]: I'm a big fan of finding alternatives to this stuff when we can.

[DJ]: We talked a while ago about quitting Spotify, for example, for possibly an alternative service like Qobuz because you can purchase DRM-free music from them.

[DJ]: You can also get DRM-free music from places like Bandcamp.

[DJ]: So I realize it's not necessarily feasible to tell someone, hey, you know that PlayStation that you scraped together like $70,000 or however much they cost now to purchase?

[DJ]: You should throw that in the garbage and buy a gaming PC, which, as we've been saying for the last few weeks, you can't afford anymore.

[DJ]: Instead...

[DJ]: But to the extent that there are alternatives available, I really do think we should try to support them as much as possible.

[DJ]: One of the things that this digital distribution potentially kills off are things like a secondary market where you could buy used video games.

[DJ]: But to the extent that that still exists, man, I don't know, maybe go out and buy like the version of the PlayStation from five years ago.

[DJ]: It still has like hundreds of awesome games for it.

[DJ]: Have you played all of those?

[DJ]: If you haven't, you know, maybe that's a one way that you can do your video gaming without supporting what Sony is doing here.

[JM]: Yeah, if you're the kind of person who doesn't need to play the latest and greatest games because those apparently, as of January of 2028, won't be made in disc form.

[JM]: If you just want to play what is probably not hundreds but thousands of games that are available in disc form, if you don't already have a PS5, get one while you can that has a disc drive and you could spend the rest of your life probably playing all the different disc-based games

[JM]: up through January of 2028 and never play them all.

[JM]: So that is indeed another way to go.

[JM]: And if you'll forgive one detour into this broader realm of renting things versus buying things, I got an email message today from whoever the current publisher of the Mac utility Bartender is,

[JM]: saying that they're launching a new version called Bartender "Pro", which is designed to subsume Bartender 6 and add some other tool that puts stuff inside the notch of a MacBook Pro.

[JM]: They're essentially adding additional functionality.

[JM]: And instead of launching Bartender 7 or instead of launching a separate product with a different name, they're re-envisioning the entire thing into a, you guessed it, subscription.

[DJ]: Re-envisioning is a weird euphemism to describe what they're doing.

[JM]: Well, every announcement these days is just chock full of euphemisms.

[JM]: So I figured I would play along.

[DJ]: Well, yeah, it's because people use ChatGPT to write them.

[JM]: And it was just so unsurprising, you know, to see this email come through because we've talked about Bartender in the past.

[JM]: I stopped using it as soon as the owners changed hands and some weird things happened.

[JM]: And I lost any goodwill and trust that I had regarding the folks that make it.

[JM]: Whether that's deserved or not, one can certainly debate.

[JM]: But when I saw the news today, it certainly underscored my personal decision to switch to a different tool because I'm just not interested in this rental model of software where you stop paying and the software just stops working.

[JM]: I've described in detail what I think is a much better model for software, which is to have something that you buy, you get a certain period of time where bug fixes and updates are included.

[JM]: And if you want to, beyond that deadline,

[JM]: pay for, say, another year of updates, you have the ability to do that, but your software doesn't just magically stop working the second you stop paying, which is presumably how this new rental-based application is being distributed.

[JM]: And there's many other tools that provide this functionality.

[JM]: And as you can imagine, I would suggest you use one of those.

[JM]: All right, moving on to other news, Xbox "CEO"...

[JM]: and of course, I'm putting that also in scare quotes because you know how I feel about there being multiple CEOs inside of a single company.

[DJ]: But wait, Justin, I thought you said we were both the CEO of the Abstractions podcast.

[JM]: Co-CEOs is already a bit weird, right?

[JM]: Because like, how can you both be *chief* executive officer? But having CEOs as different department heads?

[JM]: I mean, come on.

[DJ]: Yeah.

[DJ]: Co-CEOs involves a lot of arm wrestling.

[JM]: Lots of arm wrestling.

[DJ]: So the CEO of Xbox though.

[JM]: So the head of Xbox, Asha Sharma, has announced its biggest restructuring ever, in which she is firing 3,200 Xbox employees and divesting a bunch of studios that Xbox had acquired over time.

[DJ]: Wonder what her bonus is like this year, though.

[JM]: Guaranteed to be seven figures easy.

[DJ]: I mean, if you're the CEO of Xbox.

[JM]: And you might ask yourself if you know anything about the financials of this company, how is a division of Microsoft that is raking in $5 billion in revenue per quarter or $20 billion annually in so much trouble that they need to take these drastic measures?

[JM]: This is a company that is making $600 million in annual profit.

[JM]: So what's the problem?

[JM]: Why do all this?

[JM]: And the answer is, if you do some back of the envelope math, $600 million divided by $20 billion results in a 3% margin.

[JM]: Just to put that in perspective, Costco, which is a huge retailer in the United States and perhaps other countries, instead of $5 billion per quarter, makes $70 billion per quarter in revenue...

[JM]: and has famously low margins.

[JM]: And they have 4% margins as a company that makes its business in volume.

[JM]: So 3% margins for Xbox is not great.

[JM]: And the question is why?

[JM]: Why are their margins so low?

[JM]: And I think the answer is short-term vision

[JM]: caused a multitude of own goals.

[JM]: For example, they didn't have to buy so many studios.

[JM]: They didn't have to let their core business languish by instead pushing a so-called Game Pass subscription that most video game players don't want.

[DJ]: Wait, wait, wait, Justin, you're saying that instead of purchasing software, people don't like merely renting it through a subscription.

[DJ]: I mean, given the last couple of things we were just talking about minutes ago, that's... Oh, yeah, no, you're right.

[DJ]: That totally adds up.

[JM]: And on top of that, they didn't have to raise prices on that subscription, which they did, which of course caused people to unsubscribe.

[JM]: So this is a series of completely avoidable bad decisions, a virtual footgun cavalcade by senior management that they then euphemistically call a "reset" that they disingenuously blame on shifting industry trends and rising component prices.

[DJ]: So aside from firing a bunch of people, I'm curious, what is it that they're going to do differently with Xbox?

[DJ]: Are they going to sell a bunch of studios and go back to just paying those studios to make games for them and then they'll sell the games to you on a disc?

[DJ]: Somehow I doubt it.

[JM]: It's an excellent question, right?

[JM]: Like, what is this new strategy going to look like beyond divesting a bunch of studios that they overbought, firing a whole bunch of people, flattening layers of management where the number of reports hierarchically go from somewhere between five and seven to three.

[JM]: Okay.

[JM]: I guess reducing layers of management...

[JM]: might help, but there's nothing in their announcement that talks about strategically what they're going to do that's different that's going to help.

[JM]: So yeah, I'm with you.

[DJ]: I assume the answer is nothing.

[DJ]: I mean, this all seems like basically they're hoping revenue will stay where it is, but their expenses will go way down.

[JM]: It does seem to be the solution du jour amongst management consultants and newly hired CEOs like this one is to just trim the fat with the idea that that will somehow help margins, which of course in the short term it will, but I don't know that it actually improves the long-term health of the company and the growth of its core business.

[JM]: So if Xbox was spun out as a separate company, I for one,

[JM]: would pass on buying the stock, but that's just me.

[DJ]: This does remind me of a notion...

[DJ]: I think I heard the writer Ben Thompson articulate this, that arguably from a stockholder's perspective, companies should die eventually.

[DJ]: Companies should have a life cycle.

[DJ]: A company reaches a certain point where the best thing they could do is return value to their shareholders so their shareholders can take it and go invest in a different company.

[DJ]: But that's not the way the human mind works.

[DJ]: So companies, among other things, end up with a narrative the same way that humans have a narrative for their own lives.

[DJ]: And so no company in existence ever sees its next best move as being liquidating itself.

[DJ]: No company that's been around for 20 or 30 or 40 years and has maybe reached a point of stagnation is ever going to say, look, we've had a great ride.

[DJ]: Thank you so much to our shareholders for believing in us.

[DJ]: We've made whatever returns we can.

[DJ]: We're giving them back to you now so you can go invest in whatever the new exciting company of the future is.

[DJ]: Bye.

[DJ]: And then the people who work at that company all go get jobs at other companies and they go do other things.

[DJ]: But that's never how it works.

[DJ]: So you end up with incumbent companies like Microsoft and Xbox inside of it who there's really nothing that they're institutionally capable of doing to reinvent themselves as a bold new thing.

[DJ]: All they can really do is this kind of thing where it's like, oh, well, how do we squeeze more efficiency out of this organization without fundamentally changing it?

[DJ]: So I agree with you that there is probably no destiny for Xbox that looks very, very different from the current one, no matter how many people they lay off.

[JM]: I think potentially there is, and that would be a relentless focus on the customer and the product that you're selling to that customer.

[JM]: And I think it's probably clear at this point that I'm arguing that that really hasn't been their core focus.

[JM]: And that's why despite raking in $600 million in annual profit, they're making 3% margins and feel the need to take these drastic measures.

[JM]: I agree with you.

[JM]: I don't think they're going to reinvent themselves to the degree where they're actually going to pose a serious competitive threat to PlayStation.

[JM]: They've had years to do that.

[JM]: They haven't been able to do it.

[JM]: I see nothing that indicates that they will be able to do it.

[JM]: But they could still be the number two option. There's only two major consoles... I guess three if you count Nintendo and the Switch, so that's probably not fair for me to say there's only two major ones, but they still have the ability to be a big player here, and it's going to be up to them to see whether they can...

[JM]: innovate enough and focus enough on customers to remain a going concern.

[JM]: Like, I mean, a big part of this, I think, is just it being part of Microsoft.

[JM]: I think if it were spun out as a separate company, it would probably be a different story.

[JM]: But that's part of the perils of operating as a division of a larger company.

[JM]: And I think it shows here.

[JM]: All right, moving on, I wanted to talk briefly about a seasonal thing that happens every year, and that is the graduation commencement speech season.

[JM]: And this year was particularly fun because of the times where people got in front of a stage of people and talked about AI stuff to an audience who understandably is concerned about their role in society and the business world...

[JM]: and their ability to earn a living in this strange and uncertain future that lies ahead of us.

[JM]: So, for example, when someone got up in front of an audience and said that "the rise of artificial intelligence is the next industrial revolution" and the crowd erupts into boos,

[JM]: the speaker paused and says, "Okay, I struck a chord."

[JM]: "May I finish?"

[JM]: As if this was a surprise to the speaker.

[JM]: Like, that's the part that amazes me is this happened over and over again this past commencement speech season.

[JM]: And it came as a surprise to the people saying it.

[JM]: Are you serious?

[JM]: You didn't read the room?

[JM]: You didn't see this reaction coming?

[JM]: What are you doing on the podium then if this is your level of awareness with the audience that you're talking to?

[JM]: I'm just flabbergasted by it.

[DJ]: Yeah, I don't have a lot of sympathy for someone who's so out of touch with the students to whom they are speaking that no matter how excited they might be about artificial intelligence, they haven't taken the time to really understand how the audience feels about it.

[DJ]: Don't you think that what students look for in something like a commencement speech or really throughout their education as a whole is, hey, I'm stepping out into this world that is all very new to me and I want to be excited and I'm also kind of scared.

[DJ]: Share your wisdom with me.

[DJ]: Like I remember when I was in university back when dinosaurs ruled the earth and

[DJ]: One of the most utterly valuable exchanges I had and I was taking a computer science degree was one of my professors.

[DJ]: I think he actually wasn't even like a officially a professor.

[DJ]: I think he was an instructor like he didn't have a Ph.D.,

[DJ]: But he was a guy who had worked in the software industry, which is what I wanted to go and do.

[DJ]: And when you take a computer science degree, at least this was the case when I was doing it, you'll learn a lot of stuff that is useful, but not directly applicable to being a working software engineer or software developer.

[DJ]: And so there was at least, I think it was like one lecture near the end of term or something in this course that was about like embedded systems programming that this guy said, all right, look, I'm not going to lecture today.

[DJ]: You guys just ask me questions you're curious about working in the software industry, and I'll answer them from my experience.

[DJ]: And that was by far the most value per minute of any course I took from my degree.

[DJ]: That's what students want, right?

[DJ]: And so students who are so...

[DJ]: Students who have still bought into the notion that the thing they should do with their young lives is go to university, even as university has gotten arguably less valuable and way more expensive over the years.

[DJ]: I say less valuable, I mean, in a tool as like a tool for differentiating you as you join the workforce anyway.

[DJ]: So you've got students under those circumstances and to some degree what they want to hear is you made the right choice and it's going to be okay.

[DJ]: And instead they have people being like, have you guys used AI?

[DJ]: It's totally rad.

[DJ]: Or I mean, I assume that's what the, I didn't actually read these commencement speeches.

[DJ]: I assume they went something like that.

[JM]: Here's how one of them went.

[JM]: Some CEO got on stage and among other things said, "AI is rewriting production as we sit here."

[JM]: And when the crowd began to boo in response to this, the speaker pushed back saying, "Deal with it."

[JM]: "Like I said, it's a tool."

[JM]: You know, the same thing as saying, "Don't worry, kids, it's going to be okay."

[JM]: "Deal with it."

[DJ]: I mean, they were right.

[DJ]: They did show that they were a tool. "Deal with it."

[DJ]: Imagine like, do you really think that's going to go down in the legacy of, of like epic commencement speeches is using the expression, "Deal with it"?

[JM]: there is a place to have some empathy for the feeling that the folks in the audience are feeling.

[JM]: And this is that place.

[JM]: And saying "deal with it" shows such a ridiculous lack of empathy.

[JM]: And again, it's the surprise, right?

[JM]: It's like they, wait a minute, I don't understand.

[JM]: Why are you all reacting this way?

[JM]: I saw a comic that perfectly encapsulates this and I'll put this as the chapter art for this chapter.

[JM]: The image is of an alien with three eyes wearing a commencement cap and the laureate gown with a gaping fang maw where the mouth would normally be saying to the audience,

[JM]: "And as you head out into the world, your fresh, meaty torsos will be ripped apart and roasted to feed your new alien overlords."

[JM]: "Wait, why are you all booing?"

[DJ]: Yeah.

[DJ]: Why?

[DJ]: Why are you upset?

[DJ]: None of you are going to be able to find jobs, but I am going to make another $20 million by firing most of the people who I employ.

[DJ]: Hold on.

[DJ]: Why aren't you as happy as I am about this?

[JM]: Yeah, I, Speaker, am going to clear a multiple seven-figure bonus this year, so I'm thrilled.

[JM]: Why are you all booing?

[DJ]: Yeah, you've all been tricked into beginning your adult lives with a totally unsupportable load of student loan debt.

[DJ]: But meanwhile, the transfer of utterly all wealth into the hands of a tiny number of people is accelerating rapidly because of these amazing new technologies.

[DJ]: Isn't that exciting?

[DJ]: I mean, you can't.

[DJ]: One of the reasons this is so fascinating, aside from the awesome cartoon that you shared, is just you couldn't ask for a better microcosm of class conflict, essentially, than these particular commencement speeches.

[DJ]: Class conflict and generational conflict.

[DJ]: Everything that everyone is so riled up about is here, here in this one room with this idiot at the podium.

[JM]: Yeah, who knows where the whole large language model fueled stock market feeding frenzy is going to go.

[JM]: But assuming it hasn't cratered by graduation commencement speech season next year, it'll be interesting to see whether folks have learned their lesson or whether speakers next year double down and show an equal or commensurate degree of cluelessness.

[DJ]: Speaker rolls up to the podium like, I've brought this portable furnace and I would like you all to get in an orderly line so that I can feed you into it.

[JM]: This is the furnace that powers our data centers.

[DJ]: Yes.

[DJ]: Yeah, exactly.

[JM]: And speaking of the aforementioned feeding frenzy, Allbirds Incorporated, a trendy apparel company that was known for making wool sneakers, apparently, and which was valued at upwards of $4 billion in its heyday, ran the course of many other trendy businesses and apparently demand for its wool sneakers cratered.

[JM]: So right before they were poised to close down for good, they decided to...

[JM]: Any guesses?

[JM]: Yes, of course, you already know the answer.

[JM]: They decided to pivot into generative software infrastructure.

[JM]: They're going to make an "AI" play just like everybody else.

[DJ]: What was that I was saying a minute ago about companies refusing to return value to shareholders?

[JM]: It's weird. It's weird how suddenly we're talking about something that should have happened and didn't.

[DJ]: So hold on a second, Justin. This is a company that makes shoes...

[JM]: That *made* shoes.

[DJ]: Well, yeah, no, okay, but I'm saying that, like, everything about this company was presumably set up to produce clothing and now they're gonna... what?

[JM]: What are they doing?

[JM]: It seems like they secured a $50 million convertible finance facility, which I assume is a fancy word for a loan or a line of credit of some kind, to pivot to "AI" computing infrastructure.

[DJ]: I mean, that's quite the pivot.

[DJ]: That's not like a five to 10 degree pivot.

[DJ]: So are they, they're like clearing out all the shoes to make way for computers or have they found some way to turn wool sneakers into AI?

[DJ]: Some sort of sneaker-based tensor platform of some kind?

[JM]: Who knows what they're actually going to do?

[JM]: I don't think they know what they're going to do.

[JM]: But as you said, instead of just gracefully going out of business, as clearly they should have, they announced this new plan.

[JM]: And I'm sure you will be shocked to hear that the market reacted to this news by sending their stock soaring over 400%.

[DJ]: Right, of course.

[DJ]: Well-deserved.

[DJ]: I mean, that's a buy-and-hold, right, Justin?

[DJ]: Because that's just going to go up and up and up, obviously.

[DJ]: Of course.

[DJ]: Not actually investment advice.

[JM]: No, this couldn't possibly be a sign of a speculative bubble. And them chasing investor money to pour into the 2026 GPU and LLM infrastructure market rather than trying to revive their floundering fashion brand is, I'm just going to go out on a limb here, not going to pan out well.

[DJ]: I certainly hope it doesn't because it just feels like such a, I don't know, nakedly cynical ploy.

[DJ]: I can't really think of a more charitable description than that.

[DJ]: It just underscores this idea that companies don't exist for any reason other than to generate economic value.

[DJ]: And I'm sure there are people hearing that going like, yeah, what other reason could they possibly exist for?

[DJ]: And I guess I would put forward the notion that there's some purpose of value to humanity other than just having a stock price that goes upwards.

[DJ]: So there's just selling in particular about a company like this company made this particular kind of clothing.

[DJ]: Presumably they possessed some sort of values around that, like there was some reason that they wanted to do that and not something else.

[DJ]: And so the idea of just, well, the only thing that's valuable anymore is being somehow involved in AI.

[DJ]: So even though that has nothing to do with us, our fashion brand isn't working anymore, so we're going to do this totally different thing.

[DJ]: I don't know.

[DJ]: Again, I would be happy for the principals of that company to come on our show and tell us why they're like, yeah, yeah, I mean, the fashion thing, it's done, but we're seriously and legitimately excited about this and believe we can do...

[DJ]: Make a particular impact.

[DJ]: Like, sure, I would love to hear that, but it just does feel like the most cynical thing imaginable where it's like, all right, well, our stock's about to drop off of off of like stock exchanges.

[DJ]: What do we do?

[DJ]: I don't know.

[DJ]: Put AI in the name of our company.

[DJ]: It worked.

[JM]: Well, you will be unsurprised to hear that my favorite take on this whole story is by Matt Levine, a columnist for Bloomberg, who described the situation in the following way.

[JM]: I feel like if you ran a big technology company and you were looking to expand your artificial intelligence capabilities and needed to rent access to graphics processing units and a GPU as a service slash AI cloud company came to pitch you...

[JM]: and you said, so tell me a little bit about your company.

[JM]: And the company said, well, two weeks ago, we were a sneaker company, but we have since pivoted to AI.

[JM]: You might say, huh, thanks, but no thanks.

[JM]: We're going to go with someone with a bit more AI experience and, you know, an actual data center.

[JM]: Maybe that's wrong.

[JM]: Maybe the sneaker guys are great at AI, but you might worry.

[JM]: But if you said, so tell me a little bit about your company.

[JM]: And the company said, well, two weeks ago, we were a sneaker company called Allbirds, but we have pivoted to AI.

[JM]: Your reaction might be different because in this hypothetical, you run a big technology company and you probably spent years wearing Allbirds.

[JM]: I used to love Allbirds, you might say.

[JM]: High five.

[JM]: And then you might sign a long-term cloud hosting agreement with former Allbirds because like many tech executives, you have a nostalgic fondness for their brand.

[JM]: I don't think that this is that likely, but worth a shot.

[DJ]: Matt Levine really does have a knack for stating totally ridiculous nonsense in such a straightforward way.

[JM]: Yeah, and as he is often very good at, he goes into great detail in terms of the financials behind this pivot, what the terms of the deal are for the investors, and then underscores, okay, let's look at how this potential kind of trade could be replicated and whether you could make money by doing it...

[JM]: over and over again, which is inherently funny thing to look at, right?

[JM]: Because this whole story is ridiculous.

[JM]: But the idea that like, if you had a hundred million dollars lying around, could you just keep doing this?

[JM]: And the answer apparently is maybe, maybe you could make a bunch of money pursuing this ridiculous strategy, which is just, I don't know, inherently hilarious.

[DJ]: So you mean you could, you could like buy the stock of like a pizza company that isn't doing well and then flip them into AI and the stock would jump.

[JM]: Right.

[JM]: He even lays out like a nine-step plan.

[JM]: Step one, go find ten near-defunct public companies that still have stock exchange listings that aren't doing very much and ideally have some sort of nostalgic name recognition.

[JM]: Offer each of them $50 million, 10% upfront, to pivot to AI and sell you a convertible bond on very favorable terms.

[JM]: Maybe they'll say yes.

[JM]: What do they have to lose?

[JM]: They're going out of business anyway, right?

[JM]: You put up $50 million upfront, $5 million per deal that you sign.

[JM]: Each of them announces the pivot.

[JM]: The stock goes to the moon and each $50 million deal is now worth $300 million on paper.

[JM]: There's a delay, of course, to get shareholders approvals, to close the financings, to sell the shares in each deal, and not all of them will actually be worth $300 million by the time you're able to convert your bonds and sell your stock, but that's okay.

[JM]: You've only put up 10% of the money, and you don't have to put up the rest if a deal doesn't work out.

[JM]: But if one does work out, you close that deal for another $45 million, get a pile of stock, sell it for $300 million, and triple your total investment.

[JM]: If more than one works out, you use the $300 million from the first deal to fund the others, etc.

[JM]: If they all work out, you make billions.

[DJ]: This is an excellent example of the adage, the rich get richer, because only when you already have $100 million can you do this stupid, worthless nonsense to turn it into billions of dollars.

[DJ]: Let the rest of the world burn, but you will at least have made a killing telling people that a bunch of useless companies are AI now.

[DJ]: So to go all the way back to the beginning of this episode, did I mention how much I'm looking forward to AI just being another boring technology like anything else?

[DJ]: For, among other reasons, the fact that it will make financial industry nonsense like this a little less palatable or not palatable, but a little less maybe possible.

[JM]: I, too, yearn for the day where this can just be yet another piece of software, which it very obviously is.

[JM]: And until that day arrives, we'll just sit around listening to Otamatone-powered covers of "Bring Me to Life".

[DJ]: I bet your AI can't make those.

[JM]: Only one way to find out.

[JM]: I'm just kidding.

[JM]: I don't want to find out.

[JM]: All right, everyone.

[JM]: That's all for this episode.

[JM]: Thanks a lot for listening.

[JM]: You can find me on the web at justinmayer.com and you can find Dan at danj.ca.

[JM]: Reach out with your thoughts about this episode via the Fediverse at justin.ramble.space.