Techstrong Gang – April 16, 2025

Hey everyone. It's Wednesday. Do you know who wrote your code?

You're watching Text Drug Gang. Hey, it's Alan Hummel Techstrong here for Techstrong Gang. Happy Wednesday to you everyone.

We've got a great show with some great topics to talk about. As usual, we'll give you plenty of ai. We'll give you a little bit about what, how, what's changing in the US government, but let's end it up with something as American as Apple Pie Baseball.

Um, who could ask for anything more? Let me introduce you to our gang for today. It's a bunch of regulars.

They really don't need much of an introduction, but I think we'll start off down in Texas. And if we're going to Texas, we're going to go to Amanda Ani, our managing editor down in San Angelo, Texas, big country. Hey, Amanda, how are you?

Doing well, Welcome. It's great to have you on. Moving from San Angelo.

We'll move over to, I guess he's, looks like he's in Austin. I, yeah. He is a, uh, principal, analyst and CTO with VI as well as an analyst with Futurum, our and our friend from Dub Bronx.

Guy Courier. Hey guy. How are you?

Good to be here. It's good. From the Bronx to the hill country To the hill country.

The Hill country of Texas. Yeah. Very cool.

Good to be here. All right. Moving from the Hill country to the mountains.

It's our guitar man, uh, FU vp DevOps analyst, Mitch Ashley. Hey, Mitchell. How are you?

Great to be here. Great. Just, uh, holding up the high end of the organization, and I do mean high.

Yeah, well, in Colorado when they mean high, they mean it. Um, and then we'll, we'll move from there. We'll, we will go to the, the dean of of tech strong content in Harrison, New York.

Not named for any president. I'm still in Dublin, Actually. Oh, you're still in Dublin.

I should have recognized yeah. From the, from the sweater. But, um, I'm heading out this afternoon, so we'll be flying home and see you guys on the other side of the pond.

I Know if I rush home, if I were you, there's a lot of freedom over there in Dublin. Um, I'm Wondering if you I'm, wait, I can't wait to see if they actually let me back in. I, I wouldn't mention that dual citizenship.

I, um, I, yeah, I was fully expected to be questioned when I came home. Mm-hmm. But, uh, anyway, that I digress.

That's for tomorrow show. Perhaps. Let's, let's, let's try to stay on topic.

Um, our first topic for today is, uh, you know, in spite of the naysayers, this Dawn AI keeps getting better at coding True. Keeps getting better at coding. Mike, what do we got?

ai and a cloud service for these AI type platforms. And it seems to perform as well or better than some of these major proprietary models that everybody is talking about. And this one's open source.

Mitch, when you look at this, I can't help but wonder if quietly the open source community is not only catching up to the proprietary folks, but maybe surpassing them, and will the cost of these tools and platforms kind of move to zero? Well, open source is certainly catching up of not changing what the proprietary MO proprietary models are doing. And with, uh, this particular one called Deep Coder 14 B, um, it sounds like a, a covid strain, but no, that's something else.

Uh, his, uh, did really well at some of the benchmarks, like the live code benchmarks. They had a 60% pass rate plus, um, basically matching what open AI's oh three mini 2025 model. I mean, a lot, a lot of jargon.

There's, there's reasons why this is notable. One is, first of all, they open source the model, and they open source the training, da the training data set, the logs, all the system op optimizations that they did. So essentially, they kind of turned the whole thing over if you wanted to go recreate that.

But more importantly, you can learn from what they did, and they used a lot of reinforced learning in training this model and even, uh, kind of customize or created a, a, a variant of one of the libraries that you use to do this with. So it's, it's interesting and we'll see where is this gonna show up in developer queue and or queue developer and in, in, you know, copilot and all these other places is an another model to use. We'll kind of see how people are adopting it.

I think the most important thing is that we're seeing models now come, come to fruition. And we've talked about this, predicted this, that we would see LLMs that are specialized in software development, specialized in secure codes, specialized in code quality, things like that. I think we'll even see narrower specialization than that.

And we're on that journey. We're on that path. And this, uh, you know, push sticks, the, uh, that pokes the bear of the big proprietary models, and we'll see how they respond.

There's so much, uh, that's, that's, uh, really interesting about on this news, Mitch. Um, I, I, I don't even know where to start, but there are three things that pop out at me. Um, one is, uh, this, like you say, sort of, you know, maybe let's say better tuned, better focused training for coders.

Um, so we talk a lot on, on this show about, uh, about, uh, uh, AI for coding and how important it is for the developer to pay attention. You know, also the fact that a lot of the AI help, it's in the, the more mundane, let's say, parts of, of coding, like, like annotations and that sort of thing. That's the first thing.

Um, the second thing that pops out to me, um, is, uh, I mean, I like everybody and their mother can and fodder can, can create, uh, uh, AI right now can train ai. I am not saying that AgTech or those guys are just anybody, not at all. Um, but the number of models becoming available, um, that are viable in one setting or another is just exploding.

Um, but the, the third and, and I think most interesting to me is the secret of their success in making this model so small. 'cause 14 billion parameters is pretty small. Um, is data curation something we've also talked about on this show, uh, that, uh, and, and what I thought was deep seeks like secret sauce, uh, that they just didn't want to talk about was their ability to curate the data better dataset, um, better selection of, of codes, code, functions, libraries, and so forth.

Um, and also the training mechanism that they used, uh, eliminated a lot of, just to keep the jargon out of it, what you might call fo pulse positive sort of thing to learn from. So they paid attention to that data preparation and curation stage almost as much as the training itself. And I think that made a really big difference.

And it's almost like, like we're learning, still learning how to train ai. And this is a great example. That was one favorite thing outta this news is the, is the attention to the data.

Yeah, they used, uh, 24,000 coding examples or coding problems in their dataset. And also some of that data was synthetically generated too. Um, but your, your point's spot on of they were very intentional about what data to train with it versus turn something loose and just look at a code and we'll see what we get.

Um, they also, in the modification of their reinforced learning, um, library, something called RLHF, I believe they created a new variant or version of it that essentially could take on tackle longer logic strains, if you will, going through to train this model and collapse that to perform it faster. Again, these are all small innovations, but help accelerate reduce the time. So we're continuing to see innovations in just training models, more or less how we use them.

My question was, um, I know it's great because it makes it more accessible as guy mentioned, and so many people can use it. But I'm just wondering, as with any open source, how do you test the code quality and like security V vulnerabilities? Good question.

There, there are some, there is no standardized, you know, benchmark for how well does a model a perform in terms of generating good code or accurate code or secure code. But there are things like, um, prime Intellect has something called synthetic one. There's live, the probably most popular one is live code benchmark that you'll see.

Another one's called taco. I don't remember what it stands for other than it's a great food that I like to eat. Um, so there are some benchmarks out there.

The, the, the challenge with that is, you know, to the average practitioner, what does that mean? It's all kind of, well, do I trust live code benchmark more than I trust I believe Taco, or which one's more representative of the kind of software we're building? It, it's, it's very murky and not really helpful other than just comparing models against each other.

I'm not sure it's how good it is at representing well, which one's the kind of code that we're gonna generate for this problem. Yeah. And on the security side, on, I just wanna mention, and most of the folks on this, you know, show no more about security than I do, but prompt security, there's nothing to address the prompt security vulnerabilities here.

Um, that's just a separate item. Um, and as far as the creation of, of, uh, uh, let's say insecure code goes by this model, whatever the training set is, I feel like that's regardless an issue. I mean, at least the training set is open and it can be inspected.

And, you know, with an army of however many folks out there wanting to inspect and poke holes in it, you know, some of them might find it, that makes it more secure in the ways that we're familiar with in open source, all those eyes on it. The thing I like most about this is the transparency. And there wasn't a whole lot of magical nonsense and handwaving going on from some vendor somewhere.

They were basically saying, here's how we did it. Here's how it works. And you could see it.

And when you go to the, if you click through from our article onto their website, they did a nice job of being very detailed about how this thing was built. And, and it didn't feel like I was gonna try to change the world through magic. It felt like this was just good old fashioned computer science.

And maybe we reached a point where, you know, all this gen AI stuff, maybe it's no longer magic. Have we got to the point, Alan, where you think this is kind of just standard operating procedure now? Look, there's two things here, and, and all of you have touched on one aspect or another of it.

Number one, it, it's, you know, obviously, obviously obvious how quickly this thing is maturing, meaning this gen AI and the whole being able to train models and, and the output you're getting from these things. 5, the new one. And, um, I wrote, you're using it to help me write something, is one of the first times where I, I, I did give it a crazy good prompt to tell if I say so myself, but in reading the output, I, I couldn't really edit much.

It was that, it was that good. So is it any wonder that it's getting better at coding and we're getting, we've got smart people who are really focusing in on training and reinforced learning and, and so it's getting better. No, and and here's the good news.

We've just scratched the surface still. No, Mike, yes, it's better than it was and, and everything else. It's not magic like anymore, but you know, it's like, I feel like the engine we're still breaking in this engine.

We haven't opened next baby up yet. Okay. And when we do, all hell can break loose, right?

This, this could be really game changing. Secondly, yes, it's open source. So, you know, we, we, those of us on here of an age get that warm and fuzzies from the open source fact of the matter is those thousand eyes on the code don't always find the vulnerabilities you're looking for.

And, and they're still, I would, I would, I would venture to say that there's still gonna be security concerns with this model. And, and whether it can be poisoned or it, you know, hallucinations, hallucinations we're better at. But whether it could be poisoned or just, you know, how secure is the code and what safeguards have been built in, that'll continue to evolve too.

But, you know, the whole open source model is dominant today for a reason. Transparency, as Mike said, some of the things guy cited, some of the things Mitchell said, yes, it's good way perhaps of doing it, but now let's, you know, now we'll go to the foundation model of this where we'll have, you know, different companies weighing in and, and supporting it, hopefully, and, and broaden that base of contributors and let's see what happens. But this isn't the first or the last, and I think it's only gonna get better and better.

You know, speaking of licensing there, Alan, this, this one is under the MIT license, which is very permissive, right? It's, so we'll see people Yep. Take run with this and do something, you know, the next innovations with it Guy.

Do you think that as these kinds of open source projects progress that, you know, wall Street and analysts will take a deeper look at some of the valuations of some of the proprietary players and conclude that, hey, maybe there isn't gonna be this pot of gold at the end of that proverbial rainbow because it's becoming commonplace open source software. So it says, So yes. Another opportunity for me to say, um, this is maybe the first mail in, uh, open AI's coffin.

Um, I, I, uh, I'll keep predicting that until I turn out to be completely wrong about it. Um, which is a real possibility. But, um, I think that that, uh, the, the nature of, uh, the AI enterprise, which started out in research and with what was supposed to be a public benefit corporation, and it still technically is, which is OpenAI, uh, turned into this gold rush.

Um, and so when we're thinking about trying to, there are no natural monopolies here. I guess what I'm getting at, like Google's a what's called a natural monopoly, which is the more people use Google for search, the more valuable the search is. Um, that's not really the case here.

There is no, there cannot be one AI to rule them all. Um, especially as we're seeing this, this, you know, proliferation of this ability for so many different organizations to do this training. But I will say, Mike, that there's tons of money for Wall Street to, to make here, uh, with the use of it, um, with the use of open, you know, uh, just, just think about, you know, this idea that it's the data that you use, curating the data and all that other sort of thing too.

Whether it's to make the foundational model better or to tune it better, or, or, uh, uh, for rag applications, um, there's a lot of money to be made just there, because that is something that takes expertise, attention, talent, um, and to anywhere you have that you have the opportunity to build tools, to build platforms, to, to build service offerings and that sort of thing. So it just may not be at that core foundational model training side, but Wall Street will go where the profit is. You know what, red Hat did pretty good with some open source operating system that everyone else was able to publish too, but Tell us a little bit.

Red Hat You never heard then the guy, they got the Red Hats, hence Red Hat. Uh, but right, so I mean, 'cause that's the thing about this foundational era of open source, right? Everybody gets a great base to start on.

And then how you innovate off of that separates winners from losers success from failure. There's nothing to stop an open AI from using an MIT license model like this and then build their own special source on top of it, or, and not just open ai, nothing to stop Google or Anthropic or Meta or, or anyone. Two college students who knows well, or two college students working in a garage in Silicon Valley somewhere.

Who knows. But here, I'm sorry, go ahead, Amanda. Oh, no.

Well, I was just gonna say, I was just sitting here pondering what guys said about O Open ai, and I'm thinking, but they do have the advantage of being first and so many people were using its tools for so long, um, that they're comfortable with it, that they haven't even necessarily tried other tools at this time because it was first and they've gotten very comfortable with it. Um, so I, I don't think they're going anywhere, but That's to, well, no, but here, here's the thing. Open ai, Andro, Gemini Llama, maybe these are what I call general generative ai.

They're great for people like you and me, Amanda, to bang out some essays, do some marketing fashion, some thoughts, right? Together. What we have here with Deep Coder is, is not a general generator of ai.

It's a specific tool for code, just for code. And just like everything else in this world, you could be a jack of all trades, master of none, or you could be a specialist. And I think the world is ripe for specialist ais for specific tests.

I think we're gonna have a very specialist AI for lawyers, for legal people. Well, maybe not for lawyers, maybe for people who replace lawyers, heaven forbid, we're gonna have another specialist AI for medicine. We'll have another specialist, ai, maybe for civil engineering projects, or, you know, you name it.

So you'll, you'll have your generalist, but you're gonna have these specialties and the specialist will always be special, right? By, by definition. And I think that's, that's what you'll use.

I think this is a good case where taking that specialization further, Alan is just within code generation. Here's, you know, here's some models that are specialized at, you know, uh, prompt for basically constructing the architecture of application. Here's models that are really good around data management, data manipulation analysis.

Well Almost subspecialties, the Ones that are whatever they are. And, and I think it sets us up for a world where, where coding is not just about the model, it's about the agents that perform the tasks and talk to whichever models need to be, uh, you know, the ones that are best suited to the job, right? Right.

Think of a Salesforce or an open AI perhaps who has a, an AI marketplace where you pick the specific models you want to use for a given task. I'm thinking I may pick the, the agents that I want to use. Well, Maybe it's the ai, Those models, you know, so the next level processing mm-hmm.

If you'll, This kind of fits into my, my, you know, statement, intentionally provocative statement that there's no such thing as an AI application. There are AI services and the application that you use use those services. So that chat, GPT, for example, is a chat service.

And instead of an army of people behind it, it's, you know, a, an AI service. And, uh, I think here is where that is a helpful, uh, concept. Even if you disagree with me or it's a terminology or semantic question.

The concept is that you're trying, you're still trying to do the same things you're always trying to do, build applications, uh, services, whatever you want to call them, um, that help people do things, help people accomplish things. And, um, some of the engines in the past were analytic or, or, you know, logical or whatever it might be. Now there's this sort of hyper automated way to, uh, to take the place of some of those in terms of the functionality.

Um, I don't wanna understate the importance of what AI does or, or the lead up to a gi uh, artificial general intelligence. But you know, we're still talking about piece parts that get assembled in order to help human beings accomplish something. So here's my take on it at the end of the day, going back to Red Hat.

So Red Hat and other providers of curated open source platforms will start incorporating all this stuff into their existing platforms where they have existing relationships with developers. And all this is just gonna become part of the furniture as far as their standard relationship is concerned. So that's why I'm scratching my head going, I don't know our organization's really gonna add another vendor like open ai.

Are they just gonna go with their steady eddies that they already know and these guys are just gonna encompass all these open source tools into their existing services? Seems that's how it's gonna play out. To me, It's just a difference between, lemme just say one thing, it's just the difference between the DIY approach and the single source approach that you're talking about.

Some will continue doing DIY, they will go direct. The other ones will same use the same tools. They continue to use, sorry, me Look, look a coder, look a coder that just went over a million data and million users on a daily basis.

That's it. It's all, that's the kind of front end to IDE. It's using different models on the backend.

So it's the integration point for all those AI tools. You know, you're using something else to drive that process. A different tool like, like Cursor.

Alright, that's the last word on this one. We gotta take a break. We're coming back.

Contract to Contraction. Is it just a federal thing or is it everywhere you're watching? Textron Gang.

Discover Textron Group, the epicenter of tech innovation. We are your go-to for reaching IT leaders and practitioners worldwide. Our secret impactful content that sparks awareness, engagement, and top quality leads with us.

You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more. Join our satisfied clients. Let's revolutionize your tech journey.

Contact us today and tell your story to the world in the most powerful way with Textron Group. Hey folks, we're back and we're talking about contractors and the cancellation of contracts by the federal government, which seems to be aimed at companies like Accenture and Deloitte. And it looks like, at least if you believe all the reports that are out there, um, the federal government asked these contractors to come up with billions of dollars in savings, and they respondent in ways that I think, at least the financial times, uh, describe the response as quote unquote insulting.

And so now the federal government is trying to restructure all these contracts and replacing some things with, you know, much lower cost products and services. Um, on the face of it, this might be a good thing for taxpayers, but it seems to be one more thing that we are dealing with in a rather, uh, shall we say, uh, method that creates a lot of confusion and a lot of agitation. So, guy, are we gonna see more of this though?

Will this be a playbook than other governments states? Everybody else is gonna use the same playbook here, or is this just a one-time thing because of this administration? Uh, well, one time I think, I think this administration will continue to rinse and repeat this particular technique.

Um, I don't know, you know, how much other governments will emulate it. Um, I, I've been sort of waiting and expecting, uh, the state of Texas to sort of follow up suit in this way. It's, you know, Republican run state, um, with a governor, um, with presidential ambitions.

Um, and, you know, a party that, uh, really likes to see this sort of thing, the Republican party. Um, that said, um, I think that as far as saving taxpayers money, it's pretty clear that this doesn't, we're talking about two tenths of a percent or three tenths of a percent of the annual defense budget. Um, and you know, I think in general, uh, a lot of the moves being made around the federal government, uh, have a lot of noise and a lot of heat, but maybe not a whole lot of light.

Um, so the heat makes a difference. I do think that makes a difference in a couple of ways. Um, one is, this is no longer a standard negotiation between, you know, a, a, the, you know, a big federal agency and a set of contractors because it becomes public.

Uh, and there's a fair amount of, uh, you know, what do you call it? You know, bad words or I dunno, calling an offer insulting or whatever. You know, there's, there's a whole lot of going on here.

Bluster very often occurs behind the scenes, but in the end, there's a, tends to be a relationship between contractor and agency. Um, whether the business is there or not, there goes up or down, just like a typical contracting relationship. There's negotiation and everybody tries to walk away feeling, you know, like even if they lost this battle in general, the relationship is preserved.

This is more an issue of turning, you know, every contractor potentially into an enemy of the, the, the agency. And I think that's a lot more the point for these relationships to become hostile and for the world or the country to be warned. You know, you're not gonna, you're not gonna stiff the government on our watch, you know, kind of thing.

Um, the reality being that while there's a certain amount of waste, there's waste everywhere, um, we don't really know how much there is. Generally when you dig into things, it's less than it's claimed. Um, the one, $2 trillion of supposed savings that, uh, you know, it must be cleared, uh, when he started up, uh, heading Doge is I think the biggest example we have right now.

It was so far the actual savings, they found it's a lot of money to you and me. It's like 50, a hundred billion or something like that, but it's a fraction of Federal. It's not 50 or hundred billion.

I said the actual, their claims are closer to, well, maybe I'm getting my terms mixed up. It doesn't really matter. We do the Math Are part there at all.

Actual, but actual, Yeah. Let, let's rip the bandaid off of this. Let's rip guy.

You were very diplomatic, I appreciate it, but let's rip the bandaid off here. You're gonna take out, first of all, for anyone who knows anything about the DOD and how government works, 70% of the people in any government, uh, deployment are not government employees. They're contractors.

You are gonna take $5 billion worth of contractors off of mission critical DOD systems, like the kinds that can control our nuclear bombs and, and stuff like that. What could go wrong? What could go wrong?

Go, heaven forbid. We actually get attacked and we find out that we fired the contractor who was responsible for something. This is just, this is more fire, aim, shoot or fire aim load, or whatever it is.

It's best wards. It's best wards, right? And you, I worry about are we putting our national defense posture in jeopardy?

But then again, we've already opened the front door. Why not open the back door to Russia and the rest of 'em, right? Because that's what we're really talking about here.

That's what we're really talking about here. This isn't where, you know, it's one thing when we say, Hey, people can live a month without their social security check. Well, when you're a billionaire, yeah, you can't.

It's another thing when you start degrading the efficiency of the United States armed forces, whose mission is to protect us, that we sleep under their blanket to quote Jack Nicholson, right? And a few good men, we sleep under the blanket of their security. And when you strip out the ability for them to do it this way, you're looking for trouble.

You're looking for trouble, and it's gonna find you. So let's call this what it is. This is a degradation of our ability to defend the homeland.

I'm not sure the evidence is there for that in this particular story. A i i, you're talking about something Different, which to me is not So much budget. You're taking, you're taking $5 billion in contract.

What they've done is they, they, this is, you know, they fired everyone. Let's grind everything into a halt and we'll re award things to, you know, to a bunch of Tesla 20 year olds or something. But until that gets done, the Pentagon is a 24 7 operation.

The DOD is a 24 7. We have boys and and women in harm's way around the world 24 7 who rely on the functioning of this department. And when you just say, stop everything until we figure it out, everything stops.

And that's a real problem. That's a real problem. The Problem too, Alan, has been when the, the kind of cutting and stopping everything is a lack of understanding what the connection points.

It's not just the help desk contract for 500 million, it's what are the other things tied to that that that was supposed to accomplish? And maybe that was too big of, you know, Another, I mean, look at what's going on in other agencies. They fired people only to say, Hey, oh my God, what a mistake we made.

We gotta hire them back. Come on back. Well, now I don't wanna come back, or maybe I got another job, But al this, this situation.

So this situation's more insidious than me, okay? Because the claim and the pr and, and the statements made by, uh, uh, secretary, he about this, um, are all about, we're, you know, doing some trimming. We can, our civilian workforce can do the it, you know, the help desk and the cloud consulting stuff and blah, blah, blah, right?

So they're making it all seem very onic, like, oh, we found this waste and we're cutting it. But to the, the, I think the whole purpose of it is to be random, weird, unpredictable, so that everybody seizes up, like you say. And part of that is this sort of like, you know, well we're, we're just gonna find stuff and we're just gonna cut it, and you're just not gonna know what's happening.

And I think that's more important than, Are you trying to say we're shipping Accenture and Deloitte people to an El Salvador prison? No. Only if they object.

But, but you know, historically, there has been this perception at least that the contractors and the agencies are too cozy and that the agencies are too dependent upon the contractors. And the contractors are, uh, been steadily increasing costs along the way. And these companies are fairly profitable.

So I don't know if the, I doubt this is the right approach, but some additional transparency. Don't be an appeaser here. Yeah.

Okay. So li listen here, here's the deal. But the point is, this was not done through a Normal, that's the whole point.

It's done as Mike, what you are talking about a standard operating procedure. Hey guys, we need to renegotiate. Hey guys, this is where we've spotted waste.

Here's where we need to cut some things. Let's do that, that dance. Yeah.

And we'll, and you know what, boo Allen, that's not this. We're gonna give you a press release where you can go and say, we helped cut government spending. We took a, we took a bullet.

We're getting a billion dollars, $4 billion less from the federal government. 'cause we're just so great. Yeah.

That we work with them instead. Let me, Lemme give you a little historical chaotic context. You know, when was the last time the federal budget was balanced and actually ran a surplus in the nineties under Clinton 99 Something.

Do you think they came in and, and pulled this kind of nonsense? I remember, I'm old enough. I remember when this happened, right?

Al Gore was appointed, it was his moonshot mission to cut government spending. They took about a year to a year and a half just investigating. They didn't cut anything, digging in seeing where the money was spent, what were the outcomes.

They then made suggested cuts that were approved by something called Congress. Do we still have a Congress? I haven't heard anything.

Congress. Is that related to Red Hat at all? Right?

Maybe it was with Red Hat. Exactly. Yeah.

Okay. It was approved in a contract with America, Newt Gingrich and the old Republican party was involved in this. It was what they called bipartisan.

And they balanced the budget with a $70 billion surplus. That's, that's how government's supposed to work. That is, I'd be all for it.

I'd be all for cutting, trimming some fat off of a 700 or $800 billion defense budget. This, this is just taking a machete to mission critical, uh, services that I think we all agree that's the federal government's job is to, you know, provide defense Ba based On what we rhyme seen Elsewhere, based on what we've seen elsewhere. I can't help but wonder if somebody at Accenture or Deloitte just missed their invitation to go to Mar-a-Lago, because you Would think, well, but may, but, but isn't that the state of the world?

This issue? Go to Mar-a-Lago, right? This is their invitation to go to Mar-a-Lago.

Yeah. That could, this is the circle back. So I can't really speak, I don't have the financial or political, financial background to understand if this is gonna work or not work.

But I will say my experience with all this is that I have several friends who work as contractors in it at our Air Force base here. And they, they're scared about their jobs because they're restructuring. There's been several big announcements this week about they're restructuring.

They're gonna be cutting a lot of jobs. Um, and nobody knows whose job is still gonna exist. And they may have to move, they may be lending with another Air Force base in other states.

Um, so everything's all up in the air here for some of my friends, and I'll be sad if they move. So that's all I can say is that's what I'm seeing from it here. Thank you.

That's exactly what I'm talking about. The method used here is intended to assert dominance. Intimidation.

Mm-hmm. I mean, what, what's going on with Harvard and Columbia and the universities now? They're telling, they're gonna try to tell universities what they can teach and not teach.

Well, it's chaos. Engineering without the engineering. Yeah.

You know, on the, does anyone else recognize the irony that they're, they want universities to enforce a rule that you cannot wear a mask when you're protesting, but yet it's okay for our law enforcement officers to wear a mask when they're drug grabbing people off the street. What's the difference? Good point.

Hadn't thought about that. Let's take a break. com is the leading resource for news analysis and education on challenges facing the cybersecurity industry.

com covers all aspects of cybersecurity, including data security, DevSecOps, cloud security, application security, network security, security threats, and more. com has the largest selection of security content featuring breaking news, blog posts, podcasts, and more. com to learn more.

com. Home of security bloggers network. Hey folks, we're back with something that well, hopefully is a little less controversial, but you never know these days.

The Atlantic League a, a baseball league, I guess the minor league teams when you think about it. But they are now experimenting with putting QR codes on baseballs. The idea here is that if there's a foul ball, a fan grabs the ball and there'll be a QR code on it and they might win prizes or other things that you could possibly get through just scanning the QR code.

I think this is fun and it's a good idea. Amanda, what's your take? Yeah, absolutely.

It's exceptional marketing. It's getting everyone excited. And I think the first thing, the first thing I thought when I um, started reading this article that you might be wondering is will those QR codes last on those baseballs?

But they've tested them, they've whacked 'em with numerous bats and, um, run 'em through the winger and they are still working on the baseballs. So, um, it's just a really fun thing. I think if you're a kid going to these baseball games or an adult, uh, and you get one of these baseballs, that's really exciting to see what you've gotten for free.

Is it a free hot dog? A t-shirt, um, free ticket. So really exciting marketing strategy, I think, to get people to go watch the games.

Still, I'm a baseball fan. Um, so first of all, Mike, I believe the Atlantic League is what they call an independent league, right? So it's not like a, an affiliate of a major league team per se.

Aa aa, it's more akin to the old barnstorming. You know, my, my youngest son's a sportscaster down near where Amanda lives and there's a team down there called the Bison, and I've gone to see them play and it's in like a, an Atlantic League team. I think the QR codes are an interesting take for two.

One one use is this gamification, right? Scan the ball you got and maybe you get a free hotdog or what have you, or a tat or something. But the other thing, think about it, how much, what, what, who hit the home run?

Someone wanted to pay like $6 million. Was it Aaron Judges 61st or 62nd Home run, or, or Shhe Tony's record breaker. I mean, certain, you know, baseball's game used baseballs, right?

A baseball that, uh, uh, babe Ruth hit out of the park or, or Ted Williams or DiMaggio or Mickey Man or whatever. Um, but you always had the authenticity problem. Is that really the ball?

Now if you scan in the ball every time? So we know what ball it was actually was the one that Aaron Judge hit for that 62nd home run, without a doubt. Sounds like a new betting thing.

Which ball was the home run that Aaron Judge? Yeah, I mean, it, it, it brings a whole, you know, you know, uh, memorabilia collectors, this is a, this is a big boon for them. Sounds like an opportunity for more fan interference too.

So I don't wanna be a downer about this 'cause I love this idea, but I would say that, uh, somebody's gonna start making baseballs and slapping QR codes on 'em and bring people to sites that are full of malware and all kinds of stuff and be Well, This is why we can't have nice things. Exactly. Yeah.

I don't, I don't, I don't understand why exactly. To Mike's point, anybody can't spray paint or whatever, some, a QR code on a, on on a baseball and I don't know if he can spray have a whole a QR code That would rub off, But I, so here's the, I I think the Yeah, no, but the technology to print a QR code, I mean a QR code's pretty, you know, the, the, the definition, you know, the, I don't mean the dictionary definition. Resolution.

The resolution, Mitch, that's the word. You, I think it's gonna take some sophisticated equipment to do that. It's a long way to to, to, you know, hijack some people.

Sounds like a good way to see what the spin is on the ball. You gonna be able to maybe catch that easy? Well, no, what they better than a QR code.

Maybe they embed some sort of chip inside the ball. Oh, yeah. Mm-hmm.

And, and it has to limit. Yeah, you go, that would be pretty cool. Yeah.

Who's a scandal in here? Someone then someone could hack the chip and turn, make it turn, stop, go do all kinds of things. That's a Right.

And the, and the price of a baseball just went up to a hundred dollars a Ball. Not if we make chips in America. Got done it.

Anyway. Hey, that's a good way to end. It's a good way to end a a, uh, contentious tech, strong gang.

So good story there. And by the way, that story's out on Digital CXO as you can see on the bottom of your screen. So do check it out.

Um, have I mentioned Techstrong? It, that's our newest site for it and infrastructure and semis and stuff. Check that one out too.

Mike, Mitch, guy Amanda, thanks for joining us today on Techstrong Gang. We've got a whole bunch of tech strong TV right behind this, so stay tuned for that. But until tomorrow, enjoy your Wednesday.

This is Alan Hummel. We're out.