Are AI teams affected by scouting fog of war?

[injury log]

Originally Posted by Biggio509

The evidence is in the manual. There is ZERO evidence to support the AI uses real ratings at ANY point. Is there evidence the AI might have scouting cheats or some built in cheats to even the playing field? Yes. However there is ZERO evidence that says the AI must use real ratings.

I was the beta team lead for Scouting when the model was overhauled a couple of years ago, so I have some idea of how the system works. The manual won't be right about some of the details, and you should stop reading it like it's scripture - the people who wrote the manual are often not the same people who designed and tested the game.

Originally Posted by Biggio509

Therefore if the OSA scout "assigned" to prospects evaluates the same guys daily the information will eventually be very accurate. So accurate that there is no difference than if they saw the true ratings. 100% scouting accuracy means they are seeing the true ratings. That does not mean there is 0 blur in the model just that continual repeated scouting makes that blur close to 0. So after so many reports on the same guys the scouts become something like 99.99999% accurate.

When someone demonstrates the AI is instantly sensitive to one-point changes in ratings (independent of whether the player has been repeatedly scouted), sure, you can say that the AI has instantaneous and 99.9999% accurate scouting. That is just another way of saying that the AI can see a player's true ratings; if we're just discussing logically absurd semantic distinctions, this whole discussion is a waste of time. But there's a ton of evidence that the AI knows immediately when underlying ratings change; you just need to play around with the player Editor to prove that. If your counter-argument is "how do you know the AI isn't off by 0.00001%", I'd reply "if no one can tell the difference, what difference is there?" And if the AI were using scouting reports that were that accurate, I'd hope Markus would do away with AI scouting altogether since it would clearly just be a waste of system resources.

Originally Posted by HH20xx convert

The 'facts' you state in your post are your conclusions based upon your observations, not facts.

No, they are based on facts. I wrote a long post two or three years ago that summarized all of the variables which influence a player's ranking on the BA prospect lists. It's easy to see that those rankings are based on underlying ratings; just go into the player Editor, change a guy's ratings, and watch how he rises or falls on the prospect lists (be sure the top 100 is set to dynamically generate). Just because you find some of the rankings 'questionable' doesn't mean that scouted ratings are being used, it just means you disagree with how the AI ranks players.

At least that's what was true in several versions of OOTP before OOTP-12, and I assume is still true. I suppose there's some small chance Markus changed the coding of the top prospect lists, but I can't see why he would.

[Dr. dru]

Originally Posted by HH20xx convert

The 'facts' you state in your post are your conclusions based upon your observations, not facts. For example, at the bottom of the post you state the 'fact' that the BA prospect reports are based upon true ratings. Would you care to share your proof? The prospect ranking in past iterations of OOTP was based upon the ratings of a 'league average scout' according to the description of the report within the program. That explanation was dropped when the affiliation with Baseball America started. That does not necessarily mean the program was rewritten, but perhaps it was. My observations using dynamic scouting convince me that these are not true ratings, but I submit that as an opinion, not a fact. I do find it unlikely that a true rating exploit has been available for years and no one has noticed until now. The only fact I offer to the discussion is this: the only definitive description of the top prospect report was its description as that of a 'league average scout', several years ago. Please share your proof this has changed in OOTP 12.

This has been known since OOTP 11. Check the OOTP 11 thread it was pointed out there at least twice by PSUColonel. It has been known for awhile, you just missed it.

[Dr. dru]

Originally Posted by Biggio509

Again no argument from me. It is Dr. Dru and a couple of others, although I think Dr. Dru and PSU colonel are the same person, that are arguing the AI's perception of ratings are not blurred in trades or any other time. We are in agreement here.

I can guarentee you I am not PSUConlonel.

[Dr. dru]

Originally Posted by injury log

I was the beta team lead for Scouting when the model was overhauled a couple of years ago, so I have some idea of how the system works. The manual won't be right about some of the details, and you should stop reading it like it's scripture - the people who wrote the manual are often not the same people who designed and tested the game.




When someone demonstrates the AI is instantly sensitive to one-point changes in ratings (independent of whether the player has been repeatedly scouted), sure, you can say that the AI has instantaneous and 99.9999% accurate scouting. That is just another way of saying that the AI can see a player's true ratings; if we're just discussing logically absurd semantic distinctions, this whole discussion is a waste of time. But there's a ton of evidence that the AI knows immediately when underlying ratings change; you just need to play around with the player Editor to prove that. If your counter-argument is "how do you know the AI isn't off by 0.00001%", I'd reply "if no one can tell the difference, what difference is there?" And if the AI were using scouting reports that were that accurate, I'd hope Markus would do away with AI scouting altogether since it would clearly just be a waste of system resources.



No, they are based on facts. I wrote a long post two or three years ago that summarized all of the variables which influence a player's ranking on the BA prospect lists. It's easy to see that those rankings are based on underlying ratings; just go into the player Editor, change a guy's ratings, and watch how he rises or falls on the prospect lists (be sure the top 100 is set to dynamically generate). Just because you find some of the rankings 'questionable' doesn't mean that scouted ratings are being used, it just means you disagree with how the AI ranks players.

At least that's what was true in several versions of OOTP before OOTP-12, and I assume is still true. I suppose there's some small chance Markus changed the coding of the top prospect lists, but I can't see why he would.

I'm not worthy of such greatness. I think you just finished this topic. I can't see anybody topping this.

[The Wolf]

And the moral of the thread is: turn scouting off.

[Biggio509]

Originally Posted by injury log

I was the beta team lead for Scouting when the model was overhauled a couple of years ago, so I have some idea of how the system works. The manual won't be right about some of the details, and you should stop reading it like it's scripture - the people who wrote the manual are often not the same people who designed and tested the game.

In absence of seeing the code it is the only thing we have to go on.

[Biggio509]

Originally Posted by injury log

When someone demonstrates the AI is instantly sensitive to one-point changes in ratings (independent of whether the player has been repeatedly scouted), sure, you can say that the AI has instantaneous and 99.9999% accurate scouting. That is just another way of saying that the AI can see a player's true ratings; if we're just discussing logically absurd semantic distinctions, this whole discussion is a waste of time. But there's a ton of evidence that the AI knows immediately when underlying ratings change; you just need to play around with the player Editor to prove that.

Lets note something here that I think the 1 point change in ratings is a one point overall change in overall ratings if I read the post correctly. He was never clearly by what he changed by 1. If it is an overall rating that means changing more than one rating by possibly more than 1 point. There is no direct editing of overall rating. You have to play around will ALL the ratings to get this. So to get one point of overall ratings you would have to change more than 1 thing. The post was never clear about what he changed from 150 to 151.

Secondly, the fact they are listing 150 to 151 out of 200, since ratings cap visibly at 200 just means he found the tipping from so to so to good. I highly doubt you get the same result going from 0 to 1 or 199 to 200. There is some point where players will tip from no offers to be attractive to most teams. Most likely this experiment found the tipping point. It does not say the AI will make any significant changes or appear to notice if you change only say contact from 50 to 51.

If it were just rating changing by 1 that is .5% change. Not huge but not miniscule either. If the poster had to change 5 ratings to get 1 point change in potential or overall then it is could be a lot bigger change, although percentages would be the same it is a lot more stuff getting changed to change the potential or overall rating. It is never really been clear what ratings were changed just something, that was at a pretty high level, was changed by 1. Considering the scale is 200, I would not be surprised if that 1 was the tipping point between 3.5 and 4 stars or 4 and 4.5 stars.

The 99.99% accurate comment had nothing to do with the trade experiment. I was referring to the fact if the BA prospects use the same accuracy of info system that your scouts do and they scout every day as the manual says then they are going to be highly accurate. This has nothing to do with one particular team's blur of ratings when you shop someone around. I never claimed any 1 AI whose scout and scouting budget help determine the blur would ever be that accurate on a player shopped around.

I never argued the AI does not see ratings changes immediately. However, I think it more to do with a scouting on demand system, info as needed, as opposed to regular interval scouting reports that only update so often like a player has. There is no evidence here to convince me that the AI does not follow the same rules for how much blur they see other than perhaps the AI gets scouting on demand as opposed to maybe a 2 month old report. That model being scouting director and scouting budget affect how accurate the AI ratings are that they use in evaluations. Maybe how many times the player is scouted affects the AI as well. One must remember that we only say the final outcome not what the AI "sees" so to speak.

Regardless of the decision the AI is making scouting is only a fraction of it. Even if you set AI evaluation you only know part of the formula. The AI looks at more than just the equivalent of your scout's current ratings in making decisions. It factors in positional needs, maybe stats, scout's idea of current ratings, scout's idea of potential and a few other things. The appearance that the AI knows a rating change the first time it has to make a decision only tells us the info is updated quickly not that real ratings are shown nor that the AI does not use any sort of blurred scouting model as Dr. Dru and Sin8 have tried to claim. They don't understand they are only changing one variable in much more complex system. While they have that the AI picks up on ratings changes very quickly, as you point has already been known, the rest of the conclusion, scouting is not used in anyway and the AI sees real ratings, is pure speculation.

In short, they have something here but I believe they are jumping to conclusions and the only reason is still is going on is because Dr. Dru wanted thought an assumption based on little data that has alternative explanations was the gospel. I don't believe the AI gets bonus info unless they have better scouts and spend more than you do. However, the on demand fashion of the data means they always have a current opinion for the scout not say in July you are looking at what the scout thought on opening day because you spent half the league average on scouting.

[Biggio509]

Originally Posted by The Wolf

And the moral of the thread is: turn scouting off.

I disagree that puts you at less of a disadvantage than the AI if you have accurate ratings info. Scouting on or off puts you at somewhat equal to the AI in info but with scouting you can spend minimums and hire crappy scouts so your bad info levels the playing field with the AI. Then again my point of view is why some people go more extreme and just play stats only so you never see a rating. Can't prove but I believe the AI still uses the same system if you are playing stats only.

[Dr. dru]

Originally Posted by Biggio509

I disagree that puts you at less of a disadvantage than the AI if you have accurate ratings info. Scouting on or off puts you at somewhat equal to the AI in info but with scouting you can spend minimums and hire crappy scouts so your bad info levels the playing field with the AI. Then again my point of view is why some people go more extreme and just play stats only so you never see a rating. Can't prove but I believe the AI still uses the same system if you are playing stats only.

Yes the AI uses the same system regardless. It sees what you tell it to see % based on ratings (slightly blurred real ratings) % based on stats.

[Biggio509]

Originally Posted by injury log

No, they are based on facts. I wrote a long post two or three years ago that summarized all of the variables which influence a player's ranking on the BA prospect lists. It's easy to see that those rankings are based on underlying ratings;

Really? <sarcasm off> Of course they are based on underlying ratings. The manual and experiments show stats play no role in ranking potential. They only play a role in AI evaluation of current ratings. This still does not mean they are using real ratings and it does not mean they are using scouted ratings for potential. It just means only ratings go into potential.

As with so much "evidence" on this thread it tells us very little to the question of how the AI views ratings. We are acting like Political Scientist and trying to axe our way through an open door.

The summation of all the verbage can be put into 2 sentences.
1. Potential ratings do not use stats.
2. Current ratings use AI evaluation.
Now slap your head and say DOOOH!

We have found little to show one way or the other how the AI's view of ratings are blurred. Only that the AI's info is updated very quickly when you change ratings in the editor and that the top prospects list puts rankings in the correct order while your scouts do not. That only proves less blur of ratings for top prospects list not no blur and that the AI updates info much more quickly than you do. It says absolutely nothing about how blur the AI has.

[Dr. dru]

Originally Posted by Biggio509

In short, they have something here but I believe they are jumping to conclusions and the only reason is still is going on is because Dr. Dru wanted thought an assumption based on little data that has alternative explanations was the gospel. I don't believe the AI gets bonus info unless they have better scouts and spend more than you do. However, the on demand fashion of the data means they always have a current opinion for the scout not say in July you are looking at what the scout thought on opening day because you spent half the league average on scouting.

I don't think I ever said they don't use slightly blurred real ratings. I just said it has access to the real ratings and they don't use scouts. They may infact use a slightly blurred real rating. That would actually make sense. All I know for certain is the AI does not use their scouting director's ratings. That is if you take over an AI team and look at the ratings through their scouts eye (The AI does not see that). They see something close but perhaps not exactly the real ratings. The more I play the more this rings true. If you won't believe a dude that studied this on a beta team what will you believe.

[Biggio509]

Originally Posted by Dr. dru

Yes the AI uses the same system regardless. It sees what you tell it to see % based on ratings (slightly blurred real ratings) % based on stats.

Glad to see you are now realizing the AI uses some sort of scouting and not real ratings. Slightly blurred real ratings are "scouted" ratings. So what exactly are you arguing? Like I said we have established long ago that stats only play in a role in current ratings through AI evaluation and potential is based on blurred (scouted) ratings. The only exception is your scout if you don't click use AI evaluation then stats play no role in overall. If you click that you see what the AI overall rating (current) for a player is based on the amount of blur your scout has.

We have found 2 things of note that were not really new.
1. The AI updates info more quickly than you. (Sort of cheats to get an advantage.
2. The top prospects list seems to have more accurate info than your scout.
Beyond that almost everything else is pure speculation.

[Biggio509]

Originally Posted by Dr. dru

I don't think I ever said they don't use slightly blurred real ratings. I just said it has access to the real ratings and they don't use scouts. They may infact use a slightly blurred real rating. That would actually make sense. All I know for certain is the AI does not use their scouting director's ratings. That is if you take over an AI team and look at the ratings through their scouts eye (The AI does not see that). They see something close but perhaps not exactly the real ratings. The more I play the more this rings true. If you won't believe a dude that studied this on a beta team what will you believe.

A slightly blurred rating is a scouted rating. There is no way to know if the AI uses its scouting director's ratings. We have no idea what they are. Therefore, there is no way to see if they are being used.

When you take over a team unless you have clicked use AI evaluation your scout sees overall ratings differently then it did when it was a scout for the AI. You may not agree with the AI decision on the team but does not mean the AI is seeing different potential or current ratings than you do when you take over the team with the same scout.

The AI does not base decisions solely on ratings scouted or otherwise. Roster AI is based on how you set AI evaluation. Lineup AI is completely unknown, different from roster AI, and affect by traditional versus sabremetric. So just because you take over a team and say that does not make sense or turn off scouting and it makes more sense does not mean you are seeing different ratings with that same scout than the AI did. The AI decision making is based on much more than ratings. My hunch is that Markus has set the decision making AI to somewhat compensate for bad info.

Ratings, scouted and real, are just info and not all the info used. What the AI does with the info is a completely different story. There is no way to see the end result and know for sure what info the AI is getting and if it is different from yours. We still have no idea the weights the AI puts on certain ratings. Does power gives more points toward a decision than contact? We don't know what goes into lineup AI. We only really know how AI evaluation determines "current ratings" to determine which level a player goes to in the minors or to the majors.

[The Wolf]

Hmm...a great deal of interesting, elevated, and thoughtful discussion there. And you know what? I'm still leaving scouting off.

[Dr. dru]

Originally Posted by Biggio509

Glad to see you are now realizing the AI uses some sort of scouting and not real ratings. Slightly blurred real ratings are "scouted" ratings. So what exactly are you arguing? Like I said we have established long ago that stats only play in a role in current ratings through AI evaluation and potential is based on blurred (scouted) ratings. The only exception is your scout if you don't click use AI evaluation then stats play no role in overall. If you click that you see what the AI overall rating (current) for a player is based on the amount of blur your scout has.

We have found 2 things of note that were not really new.
1. The AI updates info more quickly than you. (Sort of cheats to get an advantage.
2. The top prospects list seems to have more accurate info than your scout.
Beyond that almost everything else is pure speculation.

That stuff is not new. That has been the case since 11.

Also, you said the AI uses the SAME model of scouting the human does which has not been true (since 11). The bottom line the AI hires scouts and has a scouting budget, but it does not get any info from that scout.

[Dr. dru]

Originally Posted by The Wolf

Hmm...a great deal of interesting, elevated, and thoughtful discussion there. And you know what? I'm still leaving scouting off.

Actually leaving scouting off is a good conclusion to draw.

[PSUColonel]

Originally Posted by Biggio509

Again no argument from me. It is Dr. Dru and a couple of others, although I think Dr. Dru and PSU colonel are the same person, that are arguing the AI's perception of ratings are not blurred in trades or any other time. We are in agreement here.

I am not Dr. Dru


very insulting by the way.

[PSUColonel]

Originally Posted by The Wolf

And the moral of the thread is: turn scouting off.

then why is it a feature?

what you are saying for the most part, is that it is a completely broken part of the game, yet it is one that is one of OOTP's biggest selling points. Hmmmm.... where have I heard this argument before???

[PSUColonel]

By the way, I find it impossible that Markus has not seen this thread.

[GrantDawg]

Originally Posted by PSUColonel

then why is it a feature?

what you are saying for the most part, is that it is a completely broken part of the game, yet it is one that is one of OOTP's biggest selling points. Hmmmm.... where have I heard this argument before???

Ummmm......errrrrrrr.....uh.........? How exactly is it "completely broken?" Because it doesn't work how you assumed it did? Or even to be generous, how you were lead to believe?

If the computer AI has an advantage in scouting (which I'll say it probably does), is that "broken?" Not to me. A human player is still going to beat it at its own game 99% of the time. I really don't care as long as it makes the AI a little harder to beat. Challenge=fun. No challenge= playing another game.

If the Baseball America list is completely accurate on the current top prospects, so? I think the RL current top prospect listed are pretty close to accurate on what *should* happen. But I have seen plenty of players on that list never develop into the player they should have been, in RL and the game. It doesn't predict injuries, player laziness, miss-use in the minors, etc. And I've seen plenty of players never listed, or listed about the time they hit the majors. So, there are still "booms and busts." If it were wildly inaccurate, I'd have much more of problem with it.

Turning off scouting doesn't "even the playing field." It just makes it easier to beat the game. If that is what you like, fine. But that is a little like playing poker with yourself. In the end, you are going to always win.