The Goal - The goal of the experiment is to explore the effects of running a game on Full Detail. The experiment will create multiple versions of the same save, with a different number of games in the EPL run on Full Detail. The saves will all have a certain amount of results in common with each other, and I will report, and compare and contrast, the differences in the results of each save.
The Set-Up - I'll run this experiment on FM08, 8.02. A database will be created with the English Premier League and the Coca-Cola Championship leagues running. This will be called Database A. I will immediately create a copy of this database, and call it Control.
In both Database A and Control, the Under 18s and Reserves competitions will run without Full Detail. The FA Cup, The Carling Cup, and The Championship will all be run at Full Detail. In Control, the EPL will run at Full Detail, while in Database A, it will not.
The Experiment - Both files will be holidayed 6 games into the EPL season, to the point where as many teams as possible have played 6 games, but no team has played 7. At this point, I will report on the status of Database A only. (I will not report on Control until the very end.) Both databases will be saved, and then I will create a copy of Database A, which will be called Database B. In Database B, the EPL will have Full Detail changed to on. After 6 games, Database A and B will have identical standings in all competitions.
The three databases will all be holidayed another 8 games into the EPL season, to the point where as many teams as possible have played 14 games, but no team has played 15. I will report on Databases A and B, and see what sort of differences have developed with one database having played 8 games with full detail and 6 without, and the other having played all 14 games without. At this point, another copy of Database A will be created. This will be called Database C, and it will also have Full Detail changed to on. The experiment will continue on as above.
At the conclusion of the experiment, there will be six databases, each with a different number of EPL games run at the Full Detail level.
Control - all 38 games played on Full Detail.
A - no games played on Full Detail.
B - 6 games identical to Database A, 32 games played on Full Detail.
C - 14 games identical to Database A, 24 games played on Full Detail.
D - 22 games identical to Database A, 16 games played on Full Detail.
E - 30 games identical to Database A, 8 games played on Full Detail.
At each point where I create a new save, I will report on each save, and some of the key differences that have come up.
I understand that the random nature of the AI will have a significant impact on the results. In fact, I could probably do the exact same experiment never changing the Full Detail option and come up with six completely different results. But like any good experiment, you need to be able to exert control over some element in order to create an environment for different results, and this is a lot easier to manipulate than the habits of the AI.
Thoughts, comments, criticisms welcome. This is my first go at this, so if I've missed something, or there's something you'd like to see included, please let me know.
Challenges / Sign-Ups / Holiday Games ideas thread.
in Challenges, Sign-Ups & Experiments
Posted
Hi everybody,
I'd like to share my idea for an AI Experiment.
The Goal - The goal of the experiment is to explore the effects of running a game on Full Detail. The experiment will create multiple versions of the same save, with a different number of games in the EPL run on Full Detail. The saves will all have a certain amount of results in common with each other, and I will report, and compare and contrast, the differences in the results of each save.
The Set-Up - I'll run this experiment on FM08, 8.02. A database will be created with the English Premier League and the Coca-Cola Championship leagues running. This will be called Database A. I will immediately create a copy of this database, and call it Control.
In both Database A and Control, the Under 18s and Reserves competitions will run without Full Detail. The FA Cup, The Carling Cup, and The Championship will all be run at Full Detail. In Control, the EPL will run at Full Detail, while in Database A, it will not.
The Experiment - Both files will be holidayed 6 games into the EPL season, to the point where as many teams as possible have played 6 games, but no team has played 7. At this point, I will report on the status of Database A only. (I will not report on Control until the very end.) Both databases will be saved, and then I will create a copy of Database A, which will be called Database B. In Database B, the EPL will have Full Detail changed to on. After 6 games, Database A and B will have identical standings in all competitions.
The three databases will all be holidayed another 8 games into the EPL season, to the point where as many teams as possible have played 14 games, but no team has played 15. I will report on Databases A and B, and see what sort of differences have developed with one database having played 8 games with full detail and 6 without, and the other having played all 14 games without. At this point, another copy of Database A will be created. This will be called Database C, and it will also have Full Detail changed to on. The experiment will continue on as above.
At the conclusion of the experiment, there will be six databases, each with a different number of EPL games run at the Full Detail level.
Control - all 38 games played on Full Detail.
A - no games played on Full Detail.
B - 6 games identical to Database A, 32 games played on Full Detail.
C - 14 games identical to Database A, 24 games played on Full Detail.
D - 22 games identical to Database A, 16 games played on Full Detail.
E - 30 games identical to Database A, 8 games played on Full Detail.
At each point where I create a new save, I will report on each save, and some of the key differences that have come up.
I understand that the random nature of the AI will have a significant impact on the results. In fact, I could probably do the exact same experiment never changing the Full Detail option and come up with six completely different results. But like any good experiment, you need to be able to exert control over some element in order to create an environment for different results, and this is a lot easier to manipulate than the habits of the AI.
Thoughts, comments, criticisms welcome. This is my first go at this, so if I've missed something, or there's something you'd like to see included, please let me know.