Caleb Williams, QB, USC selected #1 overall

[adam]

13 hours ago, AZ54 said:

Pix, 

My condolences.  I just lost my mother about 10 days ago so I haven't been online much recently.  I cannot imagine what you have endured.  Your message rings true.  Our families are a gift, Bear Down and make the time for them. 

 

AZ, I am sorry for your loss. Getting old sucks. I lost my dad 10 years ago and my mom is hanging in there but getting close. 

[dawhizz]

https://www.si.com/nfl/2024/05/06/inside-bears-scouting-process-caleb-williams-draft-mmqb

[Pixote]

39 minutes ago, dawhizz said:

https://www.si.com/nfl/2024/05/06/inside-bears-scouting-process-caleb-williams-draft-mmqb

Excellent! Great article.

A very detailed account of the process used in evaluating Williams. 

[BearFan PHX]

Im just posting this again. Instead of reading articles (although that one above is excellent!) I just watch 5 min of this at a time. Sometimes 20 minutes LOL

It's literally every throw from 2023. The more I watch it the more I think he is even better than I thought he was.

It includes all the bad plays too. But you can watch (some of) it and get a real sense of him as a player. It transcends stats and narratives, and you can just see what kind of QB he is. Spolier alert: he's very very good.
 

 

[AZ54]

23 hours ago, BearFan PHX said:

Im just posting this again. Instead of reading articles (although that one above is excellent!) I just watch 5 min of this at a time. Sometimes 20 minutes LOL

It's literally every throw from 2023. The more I watch it the more I think he is even better than I thought he was.

It includes all the bad plays too. But you can watch (some of) it and get a real sense of him as a player. It transcends stats and narratives, and you can just see what kind of QB he is. Spolier alert: he's very very good.
 

 

The past doesn’t matter.  Now that it’s done he must be the one.  

[BearFan PHX]

38 minutes ago, AZ54 said:

The past doesn’t matter.  Now that it’s done he must be the one.  

Im just scouting his abilities, and what his game looks like. Im not still at the decision to take him, Im interested in his style, and what he does well.

[AZ54]

2 hours ago, BearFan PHX said:

Im just scouting his abilities, and what his game looks like. Im not still at the decision to take him, Im interested in his style, and what he does well.

I’m just joking around a little bit although I am at the point where I just want to see him throwing too NFL caliber receivers.   Fortunately we actually have some of them on our roster now.  Smile.  

[BearFan PHX]

6 minutes ago, AZ54 said:

I’m just joking around a little bit although I am at the point where I just want to see him throwing too NFL caliber receivers.   Fortunately we actually have some of them on our roster now.  Smile.  

for sure, and for me, against NFL DBs.

I think our defense will do a good job of giving Caleb looks before he gets into NFL games. I remember for many years Bears QBs looking shell shocked when we played other teams, because our defense in practice wasnt very good.

[AZ54]

6 hours ago, BearFan PHX said:

for sure, and for me, against NFL DBs.

I think our defense will do a good job of giving Caleb looks before he gets into NFL games. I remember for many years Bears QBs looking shell shocked when we played other teams, because our defense in practice wasnt very good.

I watched the Kay Adam’s interviews from Halas Hall today.   It’s safe to say it’ll be game on in practice. 

[BearFan PHX]

1 hour ago, AZ54 said:

I watched the Kay Adam’s interviews from Halas Hall today.   It’s safe to say it’ll be game on in practice. 

[adam]

6 hours ago, AZ54 said:

I watched the Kay Adam’s interviews from Halas Hall today.   It’s safe to say it’ll be game on in practice. 

So traditionally our defense has played against a bad offense in practice. Now they are going to get some real competition, which should make them even better. Another added benefit of this offseason's acquisitions.

[Stinger226]

1 hour ago, adam said:

So traditionally our defense has played against a bad offense in practice. Now they are going to get some real competition, which should make them even better. Another added benefit of this offseason's acquisitions.

Last year we led the league in interceptions (tied with SF) with a poor pass rush for half the season. Now we have Sweat for the whole year and a baller at FS with 28 int. in 8 seasons. Caleb will be under fire every time we practice, that will make him better. Also 3 #1 WRs will make the DBs that much better. 

[adam]

So a lot of people are now using AI for season projections. Here the 33rd Team has all 3 of the Bears top WRs getting over 1K yards. If you add Kmet, Everett, Swift, Herbert, Johnson, and Scott, Williams would have close to 5K passing yards.

Also, as a disclaimer, AI is very accurate, so at least 50% of the time, these predictions will be 100% accurate.

 

[BearFan PHX]

Ha! Love this! Has anyone ever seen anything like this before? It's rare enough for the guy being tackled to try to lateral or something, but the RB has no idea what Williams is doing!
 

 

[AZ54]

4 hours ago, adam said:

So a lot of people are now using AI for season projections. Here the 33rd Team has all 3 of the Bears top WRs getting over 1K yards. If you add Kmet, Everett, Swift, Herbert, Johnson, and Scott, Williams would have close to 5K passing yards.

Also, as a disclaimer, AI is very accurate, so at least 50% of the time, these predictions will be 100% accurate.

 

AI is complete and utter BS.   To borrow a phrase from the 1980s.... garbage in garbage out.  Plus on the top of the "data set" they are using or not using you have absolutely no idea how that AI is programmed.  The most common AI systems in use today are simply plain text readers.   Meaning you can interact with the computer via everyday language.  Which is cool but you still have no clue what the underlying program is designed to do.  Drug company using AI to find new cancer drugs... but does it program it to find drugs that maximize profits?  Is the most economical for consumers?  Most efficacious?    Just using a theoretical example people can relate to.  

[BearFan PHX]

AI is different from the other kinds of computer programming we've known. In regular deterministic systems, you can look at the program and see the algorithm. You can look in the code and see that completion percentage is defined as completions divided by attempts. And you can see errors or bad assumptions such as having that fraction accidentally upside down so all completion percentages would end up over 100% lol or you can see organizations like PFF who try to relate different data points into compound stats, that we all know rarely align with reality. For example, if i took a QBs completion percentage, how often they ate kale, whether their name begins with a letter in the first or second part of the alphabet, and used that as a predictor, I could probably find some statistical trends that would fit all the data so far. I am also pretty sure it would fail to hold true into next season.

AI on the other hand is kind of like a black box. It trains itself on a data set and finds patterns in the data. Anyone who has interacted with ChatGPT knows how eerily familiar and human the text completion engine they have built is. At it's heart it's just a bigger version of the program that suggests the next word youre going to type on your cell phone. But with enough computing power behind it, not only to these pattern completion algorithms appear to think, they exhibit personalities too! And this is only because people who had personalities wrote all the material in the training data. For example, if you used every facebook post ever made as your data set, youd find personality threads behind the completion percentages. people who say "F*CK" often say "YOU" as the next word. That doesnt mean the computer is "mad".

One might be tempted to believe that the AI was thinking or feeling, but it was the people who's output created the data set who were thinking and feeling AI is just pattern matching.

It's kind of a humbling wake up call to realize that we as humans are probably not doing much more than pattern matching either. The we hold our thinking in such high esteem, but it is also probably very basic pattern matching too, since ChatGPT can already emulate it so well.

Anyway, AI creates output data that harmonizes with the existing data set. it does not think. For example, several lawyers have gotten into trouble for using ChatGPT to write briefs for them. It writes really good briefs and papers, but it also often cites fictitious court cases that never existed. The formatting, and kind of case law is copied, but the cases themselves were made up. In other words, AI can find patterns and emulate them, but it does not do any thinking or cognition.

So if AI is looking at the NFL season and making predictions, those predictions will clearly fit the data set. The problem is that bettors have been doing that forever too. Sports is notoriously chaotic and hard to predict. yes Vegas favorites of over 7 points probably win the game a large number of times. But most games are closer than that, and the players are human. They succeed and fail. They lose confidence, and gain experience.

No one would have predicted Brock Purdy, just like no one can be sure that Caleb Williams is going to be league leader.

We can use statistics to guess at who a likely candidate to win will be, but on any given sunday...

All this to say that AI season prediction and stat predictions can only be in line with everything that has happened up through last year. They cant include the crucial data of whats going to happen next of course.

Right now, AI should think that the Chiefs will win every Super Bowl they ever have again until Mahomes retires. And that IS is the most likely outcome for each Super Bowl. But it aint gonna happen in the real world.

I dont think the problem with AI is the garbage in garbage out problem of deterministic programming from the last generation of technology. And it's not a problem with poor algorithms either, because AI does it's own programming. We literally cannot look inside the program and see WHY AI made the choice it did. All the node and strength data is available, but understanding how it caused the AI to come to that answer is hidden to us, impossible to see. Just as we can know where every cell in the brain is, and how each individual cell works with complete knowledge of its inputs and outputs and chemical processes, and still not have even a clue about how consciousness arises from it in total.

And I dont think, at least in this case, that its an issue of the prompts being corrupted for example, "AI, show me a statistical outcome where the Packers win it all" then presented as just an unbiased prediction.

I think it's simply a matter that each new season is so far out of line with all the seasons that preceded it, that using last season to predict this seasons outcome is impossible. There are too many stats that affect too many other stats. Hell, we can't even figure the orbit of three bodies in space because each one affect the other (the three body problem) so how can anyone or anything ever solve something this complex?

At the core it's not a matter of the stage of technology. 10,000 years from now they still won't be able to predict the complete outcome of a season of football. If such a thing WAS calculable by any level of technology, then the future would be predetermined, and we would not even have free will!

Quantum Mechanics in physics proves to us that you can never know enough to predict the future with any accuracy. No matter how advanced your technology. The uncertainty principle literally conserves the future as probabalistic, and in doing so, preserves a space for our free will.

All that said, AI can do a damn good job of giving likely outcome ranges. It doesnt take a computer to know that Trubisky isnt gonna have 4,000 yards next year, and Williams will have more than 1,000. There are things that can be reasonably known. From that point of view the info adam has posted is excellent. It says "you have quantifiably good players. You are very likely to succeed with them" It cannot say that they wont get injured, or how many yards they will have with any real certainty. Or that DJ Moore wont get pissed off that he's not the big name in town anymore and quit on the team. That kind of thing isn't in the data set, so it can't find its way into pattern matching outcomes either.

Even the simple question of how Caleb Williams will handle adversity and coaching cant be in the data set. No matter how well you sift his on the field numbers, you wont find an answer there.

So in this case AI is another tool for statistically quantifying that you have good players and arent crazy to have high expectations. It IS likely we will be much better next year. But to say that Moore will have 850 yards or 1,300 is beyond the scope of that or any technology. The existing data doesn't tell anywhere near enough of the story to use it to predict what's gonna happen. Or even how defenses will cover our WRs. Who gets doubled is gonna affect those numbers a lot.

[BearFan PHX]

5 hours ago, adam said:

Also, as a disclaimer, AI is very accurate, so at least 50% of the time, these predictions will be 100% accurate

LOL exactly.

[adam]

OK, after evaluating the first minicamp practice, my predictions for Williams are as follows:

387-574, 67.4%, 4362 yds, 7.6 Y/A, 37 TD, 7 INT 

Bookmark this and come back here in January. I have a feeling he is going to be a top 5 QB out the door.

Comparing him to Trevor Lawrence, Williams is already months ahead in development at this point. 

 

[BearFan PHX]

22 minutes ago, adam said:

OK, after evaluating the first minicamp practice, my predictions for Williams are as follows:

387-574, 67.4%, 4362 yds, 7.6 Y/A, 37 TD, 7 INT 

Bookmark this and come back here in January. I have a feeling he is going to be a top 5 QB out the door.

Comparing him to Trevor Lawrence, Williams is already months ahead in development at this point. 

 

awesome!

Was the minicamp practice online to see? Is there a link?

[BearFan PHX]

I just went looking, and found this. It's just a couple drills, but you can watch how his footwork and timing is always the same and the ball drops right into the same location on the field every time. He is very accurate and repeatable.

If he can make reads, he can definitely make the throw.

We havent had that kind of technical consistency in the passer in a long time.
 

 

[AZ54]

Someone put part of that on X and was making a big deal about how accurate Williams was on warm up throws to receivers who were jogging.  

[BearFan PHX]

7 minutes ago, AZ54 said:

Someone put part of that on X and was making a big deal about how accurate Williams was on warm up throws to receivers who were jogging.  

hes still dropping the throws right into a small area repeatedly, no matter how fast the WRs are running, he's throwing to a spot.

[Stinger226]

I've seen high school QBs do that before. I'm not sure that drill is special. Its like having a pitcher warm up and throw 75mph balls over the plate. Its just a drill

[BearFan PHX]

1 hour ago, Stinger226 said:

I've seen high school QBs do that before. I'm not sure that drill is special. Its like having a pitcher warm up and throw 75mph balls over the plate. Its just a drill

the same is true of every drill and every throw. A 12 yard out is a 12 yard out in practice, at half speed, in high school and in the super bowl.

Some guys drop the pass right into the bucket every time, but most guys have a wider area the ball falls into across many throws. Caleb was like a machine, it looked like a rerun each throw. Awesome.

[Pixote]

14 hours ago, BearFan PHX said:

the same is true of every drill and every throw. A 12 yard out is a 12 yard out in practice, at half speed, in high school and in the super bowl.

Some guys drop the pass right into the bucket every time, but most guys have a wider area the ball falls into across many throws. Caleb was like a machine, it looked like a rerun each throw. Awesome.

I have been to so many training camps for the Bears, going back to the dog days in Plattville. I can promise you this is nothing special. In fact, I bet every QB on our roster could execute that drill without a miss-step.

Now, what you should be excited about is the blurb I read on several sites, (the quote below from nbcsportschicago.com). It tells me that Caleb and Rome are already developing the "mental connection" in day one of rookie camp that legendary QB/WR teammates had to help them have HOF careers. Manning/Harrison. Montana/Rice. Unitas/Berry. Now Williams/Odunze. Yes, we have a lot to look forward to:

Quote

At one point during the 7-on-7 period, Odunze got jammed getting out of his break, which forced Williams to roll out of the pocket and buy time before hitting him along the sideline for a gain of 18-20 yards. It was the type of off-script playmaking from Williams and ad-libbing from Odunze that the Bears hope to see on Sundays, even if it's not supposed to be featured in 7-on-7 work.

The two Bears rookies briefly chatted after the rep about the route and timing before returning to a solid day of consistently accurate work for Williams.