"Reduce the scouting error."

Ben E Lou · 10-28-2007, 08:27 AM

Does anyone feel like they have a handle on what this statement means yet? I've got some facts, as well as some speculation, but I'm pretty sure there's more to learn here. Here's what I've observed after 11 seasons with 6.1...

17 of the 51 players I've drafted had their future potential change 15 points or more. (Just a little under 35% of my 6.0e draftees changed by that much. This difference appears to be minimal.
Of the guys I've drafted, interview results so far are a bit more accurate, even though so far my scouts are worse than those I used in 6.0e, due to my having a stadium build in progress. They were around 82% accurate in 6.0e, so far they've been 91% accurate.
Anecdotally, the AI is doing a *much* better job of drafting guys that I've put on my target list. Not only do they appear to be going earlier, but they're going before other players at their same positions who are of lesser true value.
Another impression so afr is that it appears that there are fewer players whose combines don't match their bars, and the AI is drafting better, so perhaps the reason that *my* draftees are changing at a similar rate to 6.0e draftees is that I'm drafting late in the first round, and am therefore forced to go after fewer guys who are going to stay the same.

So, I have three points to put out there so far:

1. SEMI-EDUCATED GUESS: The potential magnitude of scout error is either unchanged, or changed in a very minor way.
2. SEMI-EDUCATED GUESS: The reduced scout error is most heavily noticed in the accuracy of interviews, and in the AI's ability to draft.
3. STRONGLY-EDUCATED GUESS: The unmasking process is accelerated in this version.

Any thoughts on how you have seen the reduced scout error play itself out???

Ben E Lou · 10-31-2007, 03:23 PM

So, uh, any thoughts on this?

Surtt · 10-31-2007, 05:19 PM

Have you looked at volatility?

Maybe a highly volatile player will cause the scouts error to be higher.
Jut a thought.

Ben E Lou · 10-31-2007, 05:27 PM

Quote:

Originally Posted by Surtt

Have you looked at volatility?

Maybe a highly volatile player will cause the scouts error to be higher.
Jut a thought.

I've found zero correlation between the amount of unmasking and volatility.

Yoda · 11-10-2007, 07:15 PM

I can tell you one thing about scouting error, that I found really interesting.

My scout gives me different reports on the same players, between my laptop and my PC.

QuikSand · 11-10-2007, 07:22 PM

That is pretty interesting. I know that I re-ran the same series of interviews once or twice to test whether the outcome seemed to be variable, and they were in lock-step agreement... indicating the interview result was either set firmly or else set earlier by some variable process.

Maybe the results are pinned to (seeded by?) some base that's static within a given computer, but may vary from one to another.

Yoda · 11-10-2007, 07:37 PM

That's my thinking.

highfiveoh · 11-10-2007, 07:53 PM

You're certain that both installations have the latest 6.1 patch?

Yoda · 11-10-2007, 08:09 PM

Sgran · 11-11-2007, 06:59 AM

Quote:

Originally Posted by Yoda

I can tell you one thing about scouting error, that I found really interesting.

My scout gives me different reports on the same players, between my laptop and my PC.

what was your jump-off point?

MIJB#19 · 11-12-2007, 08:36 AM

Anyone have any suggestions how that would technically work? I'm having a very hard time believing Jim would use some kind of computer based number to alter the scouting error, giving people who have two copies of FOF running on different computers a (dis?)advantage.

Edit: Doing some additional thinking, maybe it has something to do with random numbers generators?

Ben E Lou · 11-12-2007, 08:38 AM

Quote:

Originally Posted by MIJB#19

Anyone have any suggestions how that would technically work? I'm having a very hard time believing Jim would use some kind of computer based number to alter the scouting error, giving people who have two copies of FOF running on different computers a (dis?)advantage.

Based on the kickoff return for TD issue that time in IHOF, we know that he's used a clock-based random number seed for some things. It's not unreasonable, I suppose, to think that the same is true for this.

gstelmack · 11-12-2007, 09:35 AM

Are the machines in question an Intel and an AMD? I could see some minor floating point differences between them if floating point were used.

Floating point error sucks.

Yoda · 11-14-2007, 09:01 PM

My laptop is a Athlon Turion64x2 and my desktop is an Athlon as well.

cuervo72 · 11-15-2007, 10:59 AM

We could do some serious testing if any of us had 50 computers at our disposal.

10-31-2007, 03:23 PM	#2
Ben E Lou Morgado's Favorite Forum Fascist Join Date: Oct 2000 Location: Greensboro, NC	So, uh, any thoughts on this? __________________ The media don't understand the kinds of problems and pressures 54 million come wit'!

10-31-2007, 05:19 PM	#3
Surtt College Benchwarmer Join Date: Nov 2003	Have you looked at volatility? Maybe a highly volatile player will cause the scouts error to be higher. Jut a thought. __________________ “The greatest dangers to liberty lurk in insidious encroachment by men of zeal, well-meaning but without understanding.” United States Supreme Court Justice Louis D. Brandeis

11-10-2007, 08:09 PM	#9
Yoda High School Varsity Join Date: Mar 2006 Location: Woodstock, GA	Yes.

11-12-2007, 08:36 AM	#11
MIJB#19 Coordinator Join Date: Oct 2000 Location: Maassluis, Zuid-Holland, Netherlands	Anyone have any suggestions how that would technically work? I'm having a very hard time believing Jim would use some kind of computer based number to alter the scouting error, giving people who have two copies of FOF running on different computers a (dis?)advantage. Edit: Doing some additional thinking, maybe it has something to do with random numbers generators? __________________ * 2005 Golden Scribe winner for best FOF Dynasty about IHOF's Maassluis Merchantmen * Former GM of GEFL's Houston Oilers and WOOF's Curacao Cocktail Last edited by MIJB#19 : 11-12-2007 at 08:38 AM.

11-12-2007, 09:35 AM	#13
gstelmack Pro Starter Join Date: Oct 2000 Location: Cary, NC	Are the machines in question an Intel and an AMD? I could see some minor floating point differences between them if floating point were used. Floating point error sucks. __________________ -- Greg -- Author of various FOF utilities

11-10-2007, 07:15 PM	#5
Yoda High School Varsity Join Date: Mar 2006 Location: Woodstock, GA	I can tell you one thing about scouting error, that I found really interesting. My scout gives me different reports on the same players, between my laptop and my PC.

11-10-2007, 07:22 PM	#6
QuikSand lolzcat Join Date: Oct 2000 Location: Annapolis, Md	That is pretty interesting. I know that I re-ran the same series of interviews once or twice to test whether the outcome seemed to be variable, and they were in lock-step agreement... indicating the interview result was either set firmly or else set earlier by some variable process. Maybe the results are pinned to (seeded by?) some base that's static within a given computer, but may vary from one to another.

11-10-2007, 07:37 PM	#7
Yoda High School Varsity Join Date: Mar 2006 Location: Woodstock, GA	That's my thinking.

11-10-2007, 07:53 PM	#8
highfiveoh Banned Join Date: Oct 2006	You're certain that both installations have the latest 6.1 patch?

11-14-2007, 09:01 PM	#14
Yoda High School Varsity Join Date: Mar 2006 Location: Woodstock, GA	My laptop is a Athlon Turion64x2 and my desktop is an Athlon as well.

11-15-2007, 10:59 AM	#15
cuervo72 Head Coach Join Date: Dec 2002 Location: Maryland	We could do some serious testing if any of us had 50 computers at our disposal. __________________ null

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)