Leo [00:00:24]:
Hey, Gary.

Gary [00:00:24]:
Hey, Leo. How's it going?

Leo [00:00:26]:
So, yeah, it's. So we had to record it kind of at the last minute. Normally we end up recording like, you know, like a half an hour or so before our formally scheduled time, but today it was right down to the wire. In my case, it's because I was in charge of nine corgis and. We have a visitor and I was. I was left in charge, which. It raises its own set of questions, but I had to. I had to wait for folks to come back from their lunch.

Leo [00:01:03]:
So we are here. So today we are recording on Tuesday, 18th November. This was an exciting morning for a lot of the Internet. It's funny, I saw at least one claim posted on the Internet that the Internet was down, which I found wonderfully non sequitur. Ish to, you know, hey, the Internet's down. Let's run to the Internet and post that.

Gary [00:01:37]:
Yeah, really? Well, that. Actually, to be fair, I'll let you continue. But that actually happened a lot.

Leo [00:01:42]:
But yes, yeah, once upon a time. Yes.

Gary [00:01:46]:
The.

Leo [00:01:47]:
The issue, for those who weren't aware of it, is that one of the large service providers used by a large number of online services and online infrastructure, Cloudflare, had a problem this morning, and they have since both acknowledged and, of all things, apologized and having basically committed to letting us know in detail in the upcoming days exactly what went wrong. It was not an attack. It was not some kind of a DDoS or anything like that. It sounds like it was a classic whoopsie. This configuration setting that we just deployed to our infrastructure exposed a bug, and that bug was kind of sort of catastrophic. One of the things that happens when this kind of thing happens, and this is like the third in probably about four or five weeks, right? We had Azure go down. We had. I forget, was it.

Leo [00:02:46]:
Oh, it was Amazon on the east coast went down and now Cloudflare. And everybody's immediately jumping on, oh, my gosh, we're relying on these huge service providers for everything. And I think with the assumption that that's a bad thing. Right. And I, when I think about. Makes me wonder a. What's the alternative, right, for everybody and their brother to be duplicating the same kind of infrastructure in a gazillion different data centers around the planet.

Gary [00:03:22]:
Yeah, yeah.

Leo [00:03:24]:
The other comment, of course, is that, no, everything didn't go down. For example, I didn't notice it. Right. I just happened to stumble across some news reports saying that, hey, there's a major outage, these services are affected, and they were all services that of course, I wasn't using at the time. But the bottom line is that as much as we depend on these service providers, they are not 100% of what it takes to run the Internet. They are big pieces of it, to be sure, but they are not everything. The Internet was not down. I just kind of wonder if this kind of situation for any.

Leo [00:04:10]:
Well, first of all, I believe that large service providers like aws, like Azure, like Google Data, like Cloudflare and so forth, not only are they inevitable, they're the right thing to be doing, they are the right thing to have providing these very large scale services. Now if there were only one of them, I would probably have a different opinion. But, but the fact is, you know, I just rattled off four and that's not even all of them. It's just, it makes sense. It's an economy of scale kind of thing for the services that they're providing now, you know. So I do believe that using large service providers like this is inevitable. But we've both been in technology for way too long. We realize that there's another thing that, that is also inevitable and that's the fact that something's going to break sooner or later.

Leo [00:05:07]:
There will be a hiccup. The real measure of what happens when these failures happen is not that the failure happened. My measure is how quickly did they acknowledge it, how quickly did they resolve it, how quickly did they take responsibility for it and how quickly then do they, are they or how open are they about exactly what went wrong under the hood, both as an explanation to their customers, but also so that the rest of the industry can learn from their mistakes. I'm just not on this bandwagon that says these big providers are evil. They're, they're, I mean, I'm sure they have, they have their, I'm sure they have their agendas, their business plans and so forth. But in terms of concentrating all this power in these large companies, like I said, I think it's the right thing to be doing.

Gary [00:06:08]:
Yeah. And yeah, I agree. As long as there's not just one or even just two, but there's a bunch of them, you get a kind of redundancy, you get competition. Them trying to, you know, figure out, you know, better ways to do things. But what I was referring to earlier was the whole like, you know, people going to the Internet to post that the Internet is down. I guess Twitter, some of Twitter was affected by this and a popular site for detecting or for seeing if sites were down or was down.

Leo [00:06:41]:
Yes, down. Downdetector.com. yeah, that was ironic.

Gary [00:06:44]:
That was funny. Yeah. Yeah, I thought. I thought something was down today, so I went to Down Detector and Down Detector was down was like, posts that I saw. So, yeah, that was kind of funny. I wasn't as lucky as you and did not notice it because I woke up this morning immediately to a few people saying, I can't see your content at Patreon. And fortunately, one of them said, the error message is something, something Cloudflare. And that immediately made me think, well, my first thought was, most likely they've got some security software or something installed that blocking Cloudflare.

Gary [00:07:22]:
Or they decided they saw an ad and they looked at it and said, oh, that's from Cloudflare. Okay, I don't know what that is, but I'm blocking Cloudflare. But then the fact that I got a couple other people saying, I'm having trouble viewing your Patreon content made me think that, okay, this is, you know, something I could. It's not me. It is an Internet outage. I immediately went to Bluesky and did a quick search for Cloudflare and sure enough, it was listed there as something that was down. And then I looked around and it was even funny because I assumed that it was probably news because news loves to jump on these things. So somebody.

Gary [00:08:02]:
I actually told somebody who messaged me and I said, oh, by now you've probably seen the news that there's a major outage. And they said, no, I haven't seen anything on the news. So I did a search and it was just every major news outlet, them a couple links. I was like, here's like every major news outlet talking about it. So. So, yeah, it was. It was kind of interesting that I did notice, fortunately, didn't seem to last long. Patreon was never mentioned as one of the sites that was mainly affected.

Gary [00:08:31]:
I would see articles that said, like, including these. This list of sites. And I was like, oh, and Patreon wasn't on there. I think that was because they. I don't know if they completely rely on it because I noticed if I went as a creator to my page to see if it was working. It worked at first, but if I refreshed a couple times as a test, eventually I got to the Cloudflare interesting. Which tells me that, well, it probably is. They are using Cloudflare for their video serving because it was always my video pages that had the problem.

Gary [00:09:05]:
And it was simply that Cloudflare, all of Cloudflare, wasn't down. So it's like, I'll hit this server, I'll hit this server, oh, this one's down. And then you get the error.

Leo [00:09:16]:
What I'm remembering now from the admission of guilt, so to speak, is that the configuration change they made was related to their bot detection algorithms. So that made me wonder if, like, if you're running a service like Patreon and a lot of sites will now use Cloudflare to, to prevent bots, not just random bots, but they're preventing AI scrapers and all that kind of stuff. If implemented properly, that could be just a switch. Right. So that, say at Patreon you notice that Cloudflare is hicking up, maybe you flip the switch and you run without bot protection for a while until Cloudflare gets its act together.

Gary [00:10:02]:
Yes.

Leo [00:10:03]:
I don't know if that, if that is, if anybody's implemented that way, but that certainly would be a possibility.

Gary [00:10:08]:
That would be great if they did, because it is a problem. And you know, I think I've talked about it for, on the show about how it seems like the standard Cloudflare, you know, you're, are you a bot?

Leo [00:10:23]:
Are you human?

Gary [00:10:23]:
Yes. It became kind of ubiquitous as the main one for a while. And if you ever think that, oh, I'm in a situation or maybe I will turn that VPN on, like you're in a hotel or something like that. And he decided to do it, something I do more and more rarely now, you just hit, you're just hit with it constantly. One of the reasons being is that, you know, there's good, there are legitimate reasons to use VPN and there's plenty of illegitimate ones. And so basically what a lot of sites do is they probably check a box. I'm guessing there's a box that they check when they configure it that basically says, oh, if it's a vpn, then have them confirm that they're real. They just say, yeah, that's probably part of the standard checkboxes for how you configure it.

Gary [00:11:14]:
It's very easy for them to know it's a VPN because the VPNs aren't private. They're not like a mystery where the IP addresses are. There are lists of like, here are all the VPNs and if you're using one VPN provider, I'm using another. Some of them are using actually the same third party service to actually be the vpn. It's very easy. It's part of the nightmare. It's like, oh, you're doing stuff, you're getting stuff done. Then it's like, I should turn on my VPN because I'm sitting in this hotel lobby.

Gary [00:11:42]:
Then all of a sudden, every page is like, confirm you're human. Confirm you're human. That's all Cloudflare. And it's probably connected to this in some way.

Leo [00:11:53]:
It's unfortunate that they don't have the ability to remember your answers to, you know, when you give it to one site, since it's Cloudflare. Right. This single point could collect the information.

Gary [00:12:03]:
I could see the problem. Right. It's probably. They probably would like to do that.

Leo [00:12:07]:
Right.

Gary [00:12:07]:
Because there would be less work for the servers.

Leo [00:12:09]:
Right.

Gary [00:12:10]:
There's probably some other security measure that there's that.

Leo [00:12:15]:
And there's also what I just described would allow a malicious actor to enable VPN access across a multitude of sites by simply having a person answer the first one.

Gary [00:12:28]:
Yeah, yeah. So there you go.

Leo [00:12:30]:
That's unfortunate.

Gary [00:12:32]:
Yeah. But anyway, you know, so it is interesting because last. On our last episode two weeks ago, I talked about a thing that was bugging me and what did I call it? I called it the. What was it shared. Oh, false consensus. The false.

Leo [00:12:54]:
Right.

Gary [00:12:54]:
Which was a. The idea there was that somebody would have a problem with their computer device, whatever, and then immediately assume that everybody else was having the same problem.

Leo [00:13:07]:
Yeah. Or which other person confirmed that they've seen it too.

Gary [00:13:11]:
And then. Yeah. And that's it. Yeah. And everybody must have it. And it's a problem with their iPhone, their Mac, their Android, whatever it is. And not realizing that, oh, it really isn't. It's something to do with their configuration or a small number of people's configurations.

Gary [00:13:25]:
Don't assume that you can't fix it, that you have to wait because everybody experienced the problem. It's false consensus is the closest kind of logical fallacy or whatever to that. There's the opposite, which I saw with this, which was assuming that. That the problem is. Or assuming the problem is with you, when in fact it is with everyone.

Leo [00:13:50]:
Right.

Gary [00:13:51]:
And I saw that because most of the people that contacted me assume something was going wrong for them only, and they wanted my help to see if they could fix it. Am I. Maybe I'm not logged in. Maybe I let my Patreon subscription expire, like, you know, all these things. And so they assumed the opposite. And even I felt guilty to it. It's fine to fall guilty. It's fine to examine all of the possibilities.

Gary [00:14:16]:
And one of the possibilities was, oh, somebody's set up a, you know, some sort of block and they didn't realize it and it's their problem. And so they. But the problem was everybody's. It was. There was nothing that they could have done. They had to wait for somebody else to fix it.

Leo [00:14:30]:
It's funny because that reminds me of a very common. It's not a question, but it's kind of an introduction to a question that I get all the time. What did I do wrong? You didn't do anything wrong. That's a bad assumption. The software misled you or the software has a bug or Cloudflare is down, or, you know, any number of different things that, that have nothing to do with the choices you made. Yes, sometimes you do things wrong, but it is absolutely not a go to assumption. In fact, almost all the assumptions people make as go to assumptions are typically wrong because there you're, you know, yeah, sure, maybe 10%, 20% of the time.

Gary [00:15:12]:
But yeah, it is, it is interesting. I mean, I guess the solution is really that maybe we should be teaching. Maybe this is worth a video at some point. If I could bring it all together, is that, you know, work. The problem is the thing, you know, you have to go and say, here's, here's what I'm experiencing right now. Let me come up with some ideas that might be wrong, but do not stop at the first idea, right? Say, okay, this could be a problem with the operating system. This could be a problem with my Internet connection. This could be a problem with some setting I've got.

Gary [00:15:45]:
Let's go through all of this stuff and come up with other problems working. The problem could assume, okay, if this is a problem everybody's experiencing, there might be nothing I could do about it and I might not be able to tell. So skip that one for now. Let's go look at my settings. Let's go look at this. Am I logged in? Am I logged in with the right account? Let me reset my router, Let me, you know, just let me restart my computer. All these different things you can do and you build up a few things that it's like, that's not easy for me to check. So I'm going to push that aside and then see, and then maybe you come back later and say nothing else worked.

Gary [00:16:22]:
So now let me reach out to support. Now let me go and reinstall the operating system or whatever more extreme thing you're going to do. Or maybe now let me check the news to see if there's some sort of Internet outage that I haven't noticed.

Leo [00:16:35]:
The. The problem, I think is that assumptions are easier and assumptions are quicker because what you just described is a process and can sometimes be a time consuming process that honestly a lot of people just aren't prepared to. To deal with either from a time perspective or just from a knowledge perspective, understanding what all the possible f what all possible things could fit factor into the behavior that they're seeing. But yeah, it was everybody. It was everybody this time.

Gary [00:17:09]:
Yeah. Yep. Indeed.

Leo [00:17:13]:
So, gosh, we haven't talked about AI for a while. Let's talk about AI some more. One of the things that came up yesterday, I think I ran across it in a news feed, is that Amazon is so suing Perplexity. And the reason they're suing Perplexity is not because of the common reason that you sue AI or AI companies, plagiarism or unauthorized content access. Amazon is suing them because they are using AI agents to allow people to purchase things from Amazon. In other words, people can apparently configure agents with whatever criteria an agent might have to then run out and say, okay, I can find this on Amazon, I will go buy this on Amazon for you. So for example, I might say, you know, hey, Chat GPT, I need a new coffee maker. Please get me one that meets these criteria and has a good price under this amount.

Leo [00:18:19]:
And it just goes off and does things. And one of the things that it might do is push a Buy now button on Amazon. I don't have a dog in the Amazon vs. Perplexity lawsuit and you have.

Gary [00:18:36]:
Nine of them, so it's surprising.

Leo [00:18:39]:
Indeed, I do have nine of them, but none of them are involved.

Gary [00:18:42]:
None of them are involved. Wow.

Leo [00:18:45]:
The thing is, I mean, yes, it's an open issue. I can see that it's complicated. Amazon certainly has their terms of services. There's pros and cons to letting AI agents do this kind of stuff. My initial reaction is much more visceral. I don't trust AI agents. I don't trust them at all. Certainly not right now, certainly not yet, and most certainly not with my credit card.

Leo [00:19:14]:
Because in order for this scenario that Amazon is fighting against to work, of course they need to have your login credentials to Amazon, which means, of course they have the ability to push Buy now and use your credit card on file. I just, I mean, agents have potential. I get that. Rather than just answering questions or processing data on your screen, actually having AI, quote, unquote, do something, which is the short version of what an agent is, seems really, really lucrative. But man, are we early on that bell curve. It is very scary to think of turning these. I mean, AI in general is known for presenting you with misinformation. Take that same.

Leo [00:20:08]:
You know, I got it. I get it wrong. Maybe 1 or 10% of the time or 1% of the time. Now apply that to your purchases. That's not acceptable. Right. That's just not something that is acceptable. So I don't know.

Leo [00:20:21]:
I have not played with agents yet, but there's a reason I just don't trust them. I don't know if you've had any experience with them yet or not, but.

Gary [00:20:28]:
No, I. I don't, but. And I think, yeah, it's important to define trust here. And strangely, I talk a lot about trust. I talk often about trust because I often find it's a word that really means two separate things. Like when I used to have employees, the idea that you need to trust your employees comes up a lot. But I found that there were two totally separate things when it came to trust. One is, the thing that everybody just jumps to immediately is like, are they going to steal stuff? You know, are they going to do something immoral and unethical? Are they going to, like, you know, whatever.

Gary [00:21:06]:
There's that level trust. Everybody already thinks that. It's like, why would you. If you even question whether or not you trust your employees, why are they your employees? Like, yeah, but that's not the trust I'm worried about. Because you're right. If there's any question of that, they're out. The other type of trust is like, if they say they'll get something done, will they get it done? Will they get it done? Right? Well, yeah, Competency trust and that, you know, are they responsible people? That is like a thing. And it applies here to AI agents as well.

Leo [00:21:35]:
Absolutely.

Gary [00:21:36]:
A lot of people, I think, immediately jump to when you said trust the whole thing. It's like, oh, they're going to go rogue, like in science fiction novels.

Leo [00:21:44]:
Yeah, yeah.

Gary [00:21:45]:
They're going to take my information and buy stuff and give it to other people or, you know, whatever. I mean, they're going to, like, do stuff. Right. That bad because they're untrustworthy. That's not the type of trust I'm worried about. And I don't think it's what you're worried about either. It's like them making mistakes.

Leo [00:22:02]:
Are they going to do the right thing? Are they going to do what I actually want?

Gary [00:22:05]:
Right. Do they order, like a $10,000 thing from Amazon that has a similar name to the $5 thing that you actually wanted.

Leo [00:22:13]:
Exactly.

Gary [00:22:14]:
That kind of thing.

Leo [00:22:16]:
There's actually another level of trust that kind of sort of combines the two concepts of maliciousness and competency. And that is, for lack of a better word, malware. Right. We have run into some scenarios where AI has been tricked into doing things. In fact, if I'm not mistaken, wasn't it like last week that Claude was being used to perform a DDoS attack on, or to describe or implement a DDoS attack on, maybe even Cloudflare, unrelated to what happened?

Gary [00:22:51]:
Maybe. I mean, I think so.

Leo [00:22:53]:
That worries me too, right? I mean, if an agent has the ability to do things on my behalf that involve my credit card number, you have to trust that the agent is going to do what you want, but you also have to trust that the agent is itself secure and not liable to, or not at risk of being controlled by outside malicious influences.

Gary [00:23:18]:
And you could, you know, you could mitigate that very easily. A lot of work's already been done with people that have, like, over the years, you know, rich people that have personal assistants and stuff, have things where, like, for instance, you could have an AI agent that can. You could say, I need a new hard drive, I need a new 4 terabyte SSD drive. And it could go and say, instead of saying, purchase, it'll arrive tomorrow, just have it say, I picked out what I think is the best one for you. It's in your Amazon cart, and I have to go. And then even give me a link, would you like me to open your Amazon cart for you? Now I have to go, I see what's in the cart and I press buy. So if you design the tools, not necessarily the AI, but the tools that the AI and you are sharing in a smart way, I think you could do it. And I think we're going to see the gradual rollout of stuff that.

Gary [00:24:15]:
The same way that we have the gradual rollout of smart car stuff. You get the beeping when you're backing up and then the camera, and then it auto parks and then it goes on and on. And you're going to get the same thing where they're going to go and say, look, we're just not going to say it can do anything now. We're going to say it can very specifically add an item to your Amazon cart, and we'll mark those items as ones it added if there's more than one item. And this is the very specific thing we're allowing, we're going to see how it goes. And then After a few months it says, oh, based on the feedback from that unexpected stuff happened, we'll change it to be like this. I think you could gradually add stuff like that and give agency to AI in the right places in the right ways and then have it customizable. I mean, there might be some people that are like, I wanted to just add an item to my Amazon cart, tell me what it is, and then let me say, yeah, go ahead and order it.

Gary [00:25:16]:
And then maybe it doesn't have the payment information, but Amazon does. Right, right. I don't need to give Amazon my payment information. I could just, you know, so, and, but somebody else might feel uncomfortable and say, no, never go beyond just adding it to my cart or maybe have a separate cart that is like, I have to, like, it's a step before items to be added to your cart that the AI has picked out or something. There's a lot of good ways to do it and I think we're going to see all sorts of people doing it in all sorts of different ways.

Leo [00:25:46]:
I absolutely agree. It's honestly, it's a trust building exercise. Right. It's, it's one of those things where you, you give it a little bit, see how it does, and if it does well, you give a little bit more. The problem that I'm concerned about, of course, is, you know, we're talking about the shallow end of the pool and there's people taking deep dives on the other end. You know, there are people that are just saying, okay, let's turn this sucker loose and see what happens.

Gary [00:26:09]:
The same ones that sold, bought and sold all those NFT art pieces. I mean, you know, let them, let them go and do it. They're pumping the money into the economy by doing that. It's the same thing here. Let them buy a $10,000 thing at Amazon that they didn't want because they thought it was really cool to just have the AI agent just do everything for you.

Leo [00:26:28]:
Yep.

Gary [00:26:28]:
So cool. Yeah. So just getting moving on to some other topics here.

Leo [00:26:36]:
Yeah.

Gary [00:26:37]:
This is what I've been meaning to talk about for a while, but I think we have talked about it in the past and I wonder if you hear this a lot. A lot of the people, a lot of the requests I get for how to customize things, change things on Macs and iPhones and such really come down to what I would call cosmetic. It's just cosmetic doesn't change out works, it doesn't prevent you from doing things, getting things done, whatever. But people don't like how things work. They not how things look. It's not always looks, though. It's, you know, experience would be the correct thing because sometimes it is how things work, whether you right click or left click or you drag or whatever. But it's part of the experience.

Gary [00:27:19]:
And they're very small things that when I look at them, I say, what. What difference does it make? Right. And. But other people look at them and it is their primary task today. The thing that pushed them to ask another person for advice was the fact they didn't like how rounded the corners are in macOS now. It's a very common one. It's like, what difference does that make to getting stuff done? And a lot of people come throw around the term ocd. Obsessive Compulsive Disorder.

Gary [00:27:50]:
It's kind of thrown around now as adhd was like 10 years ago with. I think what's going on is a lot of people, People really have adhd, but a lot of people who don't have it just throw the term out there because their personality. Right. Resembles somebody that has it. And I think OCD thing too. Definitely a lot of people out there with ocd, but there's also a lot of people out there that'll say, I don't like the rounded corners. I guess I'm just ocd. When in fact, no, you would not be diagnosed with that if you actually went to OCD or something.

Leo [00:28:24]:
I don't know.

Gary [00:28:24]:
Yeah, yeah. Or OCD personality disorder or something like that. But anyway, it comes down to tons of requests and it's come to a head in the last few months because the new Mac and iPhone operating systems have a very new look to them. The whole liquid glass thing. There are a lot of people that have been complaining about it, and it really comes down to, oh, I guess I'm just ocd, or people not even talking about that. I don't like how it looks. Which I think it's perfectly fine to not like how something looks, but we're talking about going beyond disliking it and proclaiming your dislike and asking for assistance and trying to change it. Right, right.

Gary [00:29:07]:
It's. It's the kind of. It's just a weird. It's like if I bought a car, I don't. I'm not a particular fan of car design, of automobile design. I don't like cars as far as designs are concerned. And I don't. But I don't go around saying, oh, I bought a car, or whatever.

Gary [00:29:24]:
I don't like how it looks. Anybody Know how I could change it? No, it's just. It's the car. It's fine. It does what it's supposed to do. It doesn't look pretty, but it reminds me of the.

Leo [00:29:36]:
The last I. On the last car that I purchased, the last thing that I cared about was the color.

Gary [00:29:42]:
Yeah.

Leo [00:29:43]:
I'm not buying a color. I'm not driving a color. I'm driving a car serves certain functionality. Of course, the number one comment I got on the car is what a pretty color it is.

Gary [00:29:56]:
I've. I've had the same. I mean, I've. I don't think I've ever. I haven't put that much importance in car color. And I have owned a few cars in my life that have been definitely a color.

Leo [00:30:05]:
I would not have gotten credit where credit is due. Right. That my wife chose the color and she's, you know, she's gotten way more compliments on that than any other aspect of the car that I chose.

Gary [00:30:15]:
I mean, the thing I always say is, on the one hand, what difference does it make what color it is? I don't care. On the other hand, I. I'm spending how much money and I'm not getting it in the color that I like. Come on. Right. In the case of user interface design. Yeah. I mean, there's just the OCD thing.

Gary [00:30:37]:
I think it's perfectly fine to dislike it. Where I kind of. I don't like where things are going when I see somebody actually not upgrading because of the design.

Leo [00:30:50]:
Oh, right.

Gary [00:30:51]:
Because I'm like, just. It's, you know, and usually there are people saying, oh, I try to follow your most recent tutorial, and it doesn't work for me, but I'm back one version because I don't like how liquid glass looks like. Well, you have a reason to update now because you want to do this thing I'm showing you. And that's the first thing. The second thing is, I think if you don't like how something looks, by all means don't like it. That's fine. Fine to feel that way. Valid.

Gary [00:31:20]:
Right. Number two, it's fine to express that I don't like how it looks. Say that. But to actually go and kind of make it somebody else's problem. I mean, even if. Even if it's just reaching out and saying that, how can I change this? There's got to be a way to change it. That kind of thing. It's like, ah, come on.

Gary [00:31:40]:
You know, it's just. Say you don't like it and. But I don't know, it seems. It seems to be odd. There are plenty of things I don't like how they look and how they work.

Leo [00:31:50]:
I just.

Gary [00:31:50]:
I feel. I don't. I recognize that, oh, it's cosmetic. And I'm fine with, like, I'm fine with it.

Leo [00:31:58]:
Where I see a lot of these worlds collide, though, is where it's just a little bit more than cosmetic. There are two. Two realms where this comes up a lot for me, actually. One is the Windows desktop. People love to throw icons on the desktop. That's their primary interface for opening documents because they have a shortcut to the document on the desktop. Or, heaven forbid, they actually have the document itself stored in the desktop. Or they've got shortcuts to apps.

Leo [00:32:29]:
And installation programs in Windows love to throw shortcuts to their stuff onto your desktop. There are individuals who have very, very carefully arranged the icons on their desktop. I get it. It's kind of a neat thing to do to have all your programs over here, maybe in alphabetical order, all your documents over here, all your other document, your music down there. Every once in a while, Windows will just sort of roll the dice and rearrange all your desktop icons, which is, of course, a problem. Knowing that this is a possibility, I not only discourage people from relying on it, but I also. My desktop has one icon on it, and that's the Recycle Bin. So like I said, Windows rolls the dice and just randomly rearranges all of your desktop icons, which of course causes some people a lot, a great deal of consternation.

Leo [00:33:40]:
It's one of the reasons knowing this can happen. And this is something that I've mentioned to people lots. It feels like relatively common knowledge, but it's one of those things where, yes, I know you like that feature. Don't rely on that feature. Eventually you'll be sorry. It's one of the reasons that my desktop is completely empty. I have one icon on it, and it's the Recycle Bin. I've just learned the hard way.

Leo [00:34:06]:
The other area where this comes up is in spam fighting. People get so annoyed with spam and they keep asking, what can I do? How do I make it stop? You don't.

Gary [00:34:22]:
Yeah, I block them. Every time I block every single one, it's like, yeah, that's not.

Leo [00:34:26]:
Yeah, good luck with that. Blocking doesn't work anymore. Spammers are smarter than that. The only thing thing you can do is mark spam as spam, take a breath and move on. So this kind of sort of feels like it's in the same bucket where, you know, Yes, I get some of these things are incredibly important to you, but in the grand scheme of things, there are ways to make them less important.

Gary [00:34:54]:
Exactly. And, and I think, yeah, I think it's just something people need to. I don't know, they need to think. I wish they would think about more. I wish I had a better answer because I hate if somebody has reached out to me. Maybe after watching me my videos for a long time. The last thing I want to do is say, oh, I don't really want to spend time trying to figure out how to help you here because it's not that important. Like, I don't want to push somebody away.

Gary [00:35:24]:
Right, you don't want to do that at all.

Leo [00:35:25]:
Yeah. It's funny, I. I will not use the. Anything that. That reduces down to. I don't think your problem is important because it's important to them.

Gary [00:35:36]:
Exactly.

Leo [00:35:38]:
But I may try to deflect in some hopefully empathetic way that doesn't involve me spending a lot of time to solve what I would consider to be an incredibly unimportant problem.

Gary [00:35:50]:
Right. Yeah. So, yeah, it's just something we have to deal with.

Leo [00:35:55]:
One more item for today. I think there is one of the concepts that has come up with AI. Again, we haven't talked about AI enough, so we have to talk about it some more is this concept of vibe coding. To define vibe coding is rather than writing software, you describe what you want the result to be and AI writes the software for you. Which is honestly kind of cool. I've used it myself to do some interesting things. However, I have never ever once used it in such a way where I just say, okay, yeah, that looks like code, let's run it and see what.

Gary [00:36:37]:
Happens.

Leo [00:36:39]:
Again, much like I don't trust AI agents with my credit card, I do not trust vibe coding AI with my computer. Right. There's just too many things that can go wrong and especially depending on the problem that I'm trying to solve at that point in time. I want to watch what's going on. I want to understand what's going on. Heck, I'm learning from the code that gets generated, which is actually really cool. There is someone who pointed out that somewhere, someday soon there will be a huge data breach or some other security issue that is will boil down to nothing more than someone's over reliance on vibe coding. Because vibe coding, like the rest of AI, makes mistakes.

Leo [00:37:36]:
And I gotta say that one actually kind of scared me because I completely agree I think that there are individuals and companies that are relying much too heavily on Vibe coding. They haven't learned the lesson yet and they are probably using it in areas of their products and services that they shouldn't.

Gary [00:37:55]:
Yeah, I agree. Most of the Vibe coding that I do is actually, and it only just occurred to me hearing you talk about this, it's actually for one time use like for instance, I have a bunch of data like, oh, I downloaded all my YouTube comments from the last year and I want to analyze like, but, but there it's in a weird format. Sure. So I vibe code to quickly create something that will then, you know, change it and I'm done with it. Like, I don't need that code anymore. That Python code or whatever it wrote. It's like, I don't need it anymore. I'm.

Gary [00:38:29]:
It's just I got, I got what I wanted from it. I do that a lot with Vibe coding. Just one time use that.

Leo [00:38:34]:
I've actually got one tool that I've vibe coded that I'm in the process of refining. So one of the nonprofit organizations that I support uses the tool. Group Me. It's a group chat application. Microsoft owns it. It's used heavily for some of our operations. But there's no obvious or easy way user interface way to archive the chats, which can be useful. But there is an API.

Leo [00:39:02]:
I don't know the API. I don't want to learn the API. But if I go to, in this case Claude, and I say, hey, please write me a script that downloads and archives group Me conversations. And it did. Yep. Had to tweak it here or there, a couple of changes I still need to make for it, but it's really cool. And yes, that's a script that I'm going to be hanging onto because this is an ongoing need. Right.

Leo [00:39:31]:
Every time we have a new conversation, there'll be a new script, new archive required. But that's a wonderful use of it again when used with the expectation that you have to check what you've got before you put it into production.

Gary [00:39:47]:
Cool.

Leo [00:39:48]:
All right. What's cool?

Gary [00:39:51]:
What's cool? Well, you got the really cool one. Yeah, you. But I will, I will mention something else. So we are both fans of the sci fi author Dennis C. Taylor who wrote the Bob Verse, Bobby Verse, you know, Was it six books now?

Leo [00:40:09]:
I think so, yeah.

Gary [00:40:10]:
But he's written a couple others and I decided to give one of them a try. And have you read Flybot?

Leo [00:40:16]:
I have not.

Gary [00:40:17]:
Yeah, it's very much same style but it is much more down to Earth actually takes place. It takes place just marginally in the future. And it's got a bunch of stuff going on. It's an interesting story that involves a lot of AI, but not too much AI. Like AI is involved, but it's not. There's basically some multiple mysteries going on in some normal kind of people trying to solve it, and he can't resist, but to not throw in some AI and some space stuff in there. So there's some portions where I was like, oh, yeah, so he's back now talking about space travel, even though this is all on Earth, but somehow he worked it in there. But no, it's good.

Gary [00:41:05]:
It's a good little one off book. And the interesting thing is that the book takes place in a. In a near future. Things are pretty much like they are now, right? Just a little bit more futuristic. Like all the. All the cars are driverless cars, for instance, but they're still cars. They're still. They still hail them.

Gary [00:41:21]:
Like you do, like a Lyft or an Uber and stuff. It's. You know. But he does mention several times without going into details, that there was an AI apocalypse, but it was more of an AI apocalypse, kind of like COVID 19. Whereas it happened, but things kind of got back to normal afterwards. So there's basically like an Interpol, but for AI that is active throughout the book. And there's references to things you can and cannot do with computers now because of what happened back then. The idea is like 10 years ago.

Gary [00:41:59]:
So, you know, that's kind of interesting. And halfway through, when he hadn't explained what happened, he just referenced it. I was like, oh, please don't. Like, I don't want, like, I want that to be separate. And he didn't. He. He restrained himself like other authors do not sometimes, and was like, this story isn't about that. It's just.

Gary [00:42:16]:
You just need to know that that happened. Anyway, so, yeah, if you're looking for some light science fiction reading, as you know, since you like the bobiverse books, Flybot is a good one.

Leo [00:42:30]:
Add this one to the list. Yeah, as you mentioned it, it's really popular in a lot of our circles right now.

Gary [00:42:36]:
Yes.

Leo [00:42:36]:
I'm watching Pluribus on Apple tv and oh, my. It is very impressive. I really like how they set it up, how they started. I like where they're going with it. We'll see how it all turns out. For those that haven't even heard of it yet, the best way to describe it is if you've ever seen or read Invasion of the Body Snatchers, it's basically a new take on that scenario.

Gary [00:43:01]:
Yeah, kind of. I hadn't heard it referred to like that until you did it, which is kind of interesting. It is, yeah. It's different than other sci fi shows in a weird respect in that it really sends your brain off thinking about the possibilities of where the story's going and what have we seen that is important. It is. By the way, it's created by Vince Gillian. Breaking Bad. Right.

Gary [00:43:32]:
The crater behind Breaking Bad and Better Call Saul. And it very much has his style on it.

Leo [00:43:39]:
So I've never watched Better Call Saul or Breaking Bad.

Gary [00:43:44]:
Really. Okay.

Leo [00:43:44]:
I really don't have that sense of a vibe for.

Gary [00:43:47]:
Oh yeah, it's a strong vibe and sometimes a lot of the same filming techniques, but it also has the vibe. There are certain lines. Not everything said can be important. That's impossible. Right. But there are so many lines that you're like, huh, why did they say that? Is that going to be really important? And then there's people making TikTok videos all over the place saying, okay, that character said this, here's what it could mean. And just going off and it. And then it's not like, oh, give me a break.

Gary [00:44:22]:
It's kind of like, oh, that's fascinating. Like that made me think. So, yeah, we could do a whole show. We could just start a whole new podcast on Floribus, as probably many people have done.

Leo [00:44:33]:
I was going to say we probably would not be the first if we went down that path. The one thing I will say is that this, the current storyline, I think what we're three episodes in, maybe four. I'm. It's a it to me, it's a really good sign that I'm vested in the. In the story.

Gary [00:44:52]:
Yeah.

Leo [00:44:52]:
Really, really curious how they're going to resolve it. There's so many different ways that they could take it. And I'm just really interested in understanding, you know, what happens when, when all this is over, so.

Gary [00:45:06]:
I know, I know, it's interesting.

Leo [00:45:07]:
Anyway, Pluribus, Apple tv.

Gary [00:45:10]:
Yeah.

Leo [00:45:11]:
In terms of blatant self promotion. You know, it's funny, the article I'm pointing at is what happens if I don't Update to Windows 11? It's askleo.com 186512 it continues to be an ongoing question with the end of Support of Windows 10 and then the not so end of support by the addition of the extended security updates or esu, the question continues to come up. Right. People are, you know, they've got their older Windows 10 machines, they just want to know, okay, what if I don't? What happens if I don't? Well, we'll take a look at that. So cool.

Gary [00:45:49]:
Point to one called using Apple intelligence and shortcuts to rename files, which actually fits in with what we were talking about earlier in terms of like agency and such. So the idea is that, that I wanted to demonstrate Apple intelligence and a use for it. I built a shortcut, very simple one where basically you say, look at this image, tell me what the file name should be. So it takes a group of files that are like file1jpg, file2jpg and all that and renames them file1 parenthesis elephant in river close parenthesis JPEG file2 waterfall. It basically is a wafer. It's like a demonstration of what's possible. I like it because one of the things I did is I didn't give it too much agency because I said use the original name and then put the new name in parentheses so you don't lose the data of the original information.

Leo [00:46:45]:
Yep, I like it.

Gary [00:46:46]:
And the other thing is that I'm basically saying, hey, you know how AI hallucinates or it tells you things in history that didn't happen or all this. Yeah, it's not going to do any of that. You're just asking it to describe a photo in five words or less.

Leo [00:47:01]:
Right.

Gary [00:47:02]:
So it's like basically it may not be ideal in what it comes back with, but it's going to be better than what you had before, which was.

Leo [00:47:10]:
File1.Jpg I have a challenge for you. Please have it add the that description or that name to the images metadata.

Gary [00:47:25]:
Well, that's a little tougher because I'm using in shortcuts basically. It's like, what power do I have? One of the things I've got is rename file. So I'm using that.

Leo [00:47:35]:
I don't have to run a program.

Gary [00:47:37]:
Yeah. I'd have to install like a terminal program that, like a shell script or something that would rename or that would get into the metadata of a file, allow you to do, you know, so it'd be like a lot of, lot of steps. Not as simple. Like here's three or four lines of a shortcut that uses AI.

Leo [00:47:58]:
The reason that comes to mind is because I have a collection of, you know, stock photos that I've downloaded over the years and the file names are always completely random.

Gary [00:48:07]:
Right.

Leo [00:48:08]:
There's just nothing useful in the file names. But one of the things most of the stock photo agencies do is they do include some metadata, be it keywords.

Gary [00:48:16]:
Well, getting the metadata out and using that.

Leo [00:48:19]:
I want you to put it in.

Gary [00:48:20]:
Right, Put it in, though. I mean, yeah, you could do it if you worked hard enough. It would basically greatly. It would put it up the next level of magnitude in terms of, you know, ability to build that.

Leo [00:48:32]:
Your next Vibe coding project.

Gary [00:48:34]:
Yeah, yeah.

Leo [00:48:36]:
Heck, you could probably. There's probably a Python library for. Exactly.

Gary [00:48:39]:
I mean, it's not really, you know, with photo libraries like the app, you know, Mac and other Apple products, the photos apps can search for objects in the photos. So if I need the picture of the elephant, I don't need to have it in metadata, I could just search for elephant. So, you know, it's. It's something. It's not as critical now.

Leo [00:49:04]:
Anyway. On that note, that does sound kind of fun though. I think that pretty much does us for this week. Thank you as always for listening and we will see you here again very soon. Take care, everyone. Bye. Bye.

Gary [00:49:17]:
Bye.