productive

with Charlie.

And they all said that he threw a soft pass.

So I think that's where that – quote from Gifford comes is, hey, because it was a soft pass, but it was always accurate and on time, I could take a quick peek where the defender was because it wasn't a rocket ball or wasn't a sort of looping long pass all the way through.

Especially Gifford.

Gifford was a you know, caught pass out of the backfield and as a flanker.

So he probably was, you know, doing the swing passes and things like that and needed to know where those defenders were coming.

Tell us about Charlie and that 1958 game.

Yeah.

So the visibility is can I see what it actually did? Predictability is same situation, same behavior, right? [laughs] That's deterministic.

Somewhat of a challenge, right? And then you try to measure, like, how that variance happens in, in identical use cases.

And the last part, of course, is control.

Can we...

If we see something going wrong, can we stop it? Can we adjust it? Um, can we, over time, make it so that control is less and less of an issue? So I, I think that's the way you start building trusted systems is by engineering for visibility, predictability, and control.

[gentle music]

Yeah.

And it seems to me, and I want to get your opinion, that Hurston Waldrop is just kind of another example of where the Braves have been operating out of necessity more than operating out of desire and, and we've kind of run into a point where they probably have somebody in a pitching, uh, in a, in a pitching spot in the rotation that probably needs to be in Triple-A that they really can't afford to, to have him in Triple-A at the moment.

Yeah, I agree with you 100%.

I th- I think in a perfect world, Waldrop would've got more, more, uh, lead time at Triple-A to try to get, hone his, uh, you know, dial in the command and stuff.

And, and I mean, I think you're right.

I, I mean, Braves got off to this great start this season.

I think they exceeded the-- I mean, they were kind of an afterthought, you know.

Coming in, if you looked at all the, uh, predictions and, uh, playoff odds and stuff like that, you know, it seemed like they might be fighting with the Marlins for third place and, and the, and the Mets and the Phillies running away from, uh, with the division.

Do you think, do you think this is a year where they have to, you know, where they have to put up a little more?

What are some of those unexpected contributions from the bats this year that are really helping out from an offensive perspective that you can really point to?

Well, it's been up and down, up and down the lineup.

I think Michael Harris, I know that's a big name.

That's a guy that was a top prospect, came in.

But if you followed that season last year, he was one of the worst hitters in baseball for almost four-fifths of the season.

The last month, he kind of found something.

He's been excellent this year.

What are your thoughts

Right.

Um, you know, linear regression is, is a form of machine learning, which is under an AI category, and, you know, it helps you make predictions.

That's great.

Frankly, until this moment, all that AI has ever done is really classified or made a prediction, right? So that's helping you figure out, okay, you put something in your shopping cart, what should you put next in that shopping cart? Or, you know, even weather forecasts and things like that.

One of the fundamental differences is that these new language modelsStarted to generate.

And even like, you know, the image generation models, these diffusion models, right? Where you, you, you sort of build things out of just random noise.

And I think that is such a paradigm shift, especially when it comes to marketing, right? 'Cause i- in marketing you wanna, you'd be using those tools to predict how well maybe a campaign's going to do, but now you're actually using to create the campaign.

Yeah.