Automated dynamic custom game content generation

1. A method comprising, at a games server system that serves a computer-implemented game played by multiple players via respective client devices:

accessing historic player model values for a target player group of players forming a subset of the multiple players of the game, the historic player model values for the target player group comprising multiple respective values for a multi-dimensional player model, the player model values having been generated based on previous in-game behavior of the target player group;

in an automated operation using the historic player model values for the target player group as seed information, generating a candidate custom content set comprising a plurality of game content items for a gameplay unit of the game, the candidate custom content set being customized specifically for the target player group based on the associated historic player model values;

forecasting, using the historic player model values of the target player group, a predicted sequence of user actions during gameplay of the gameplay unit having the candidate custom content set;

using one or more computer process devices configured therefor, computing, based on the forecasted sequence of user actions modeled gameplay performance of the unit of gameplay;

comparing the modeled gameplay performance to predefined implementation criteria; and

conditional on the modeled gameplay performance satisfying the implementation criteria, customizing the game for one or more of the target player group by inclusion thereof of a custom variant of unit of gameplay, which custom variant includes the candidate custom content set.

2. The method of claim 1 , wherein the computing of the modeled gameplay performance comprises computing, based at least in part on software-defined outcomes of the forecasted sequence of user actions in the game, a set of metrics associated with playing of the gameplay unit consistent with the historic player model values associated with the target player group.

3. The method of claim 2 , wherein the comparing of the modeled gameplay performance to the predefined implementation criteria comprises:

determining whether the set of metrics falls within a predefined range, the customizing of the game being conditional upon determining that the set of metrics corresponds to the predefined range.

4. The method of claim 3 , further comprising:

upon determining that the set of metrics does not correspond to the predefined range:

adjusting the candidate custom content set, thereby producing a revised candidate content set;

re-forecasting the sequence of user actions with the revised candidate content set; and

re-computing the set of metrics based on the re-forecasted sequence of user actions.

5. The method of claim 4 , wherein adjusting the custom content set leverages a hyperparameter search based on the predefined range and the custom content set.

6. The method of claim 4 , further comprising:

repeating the adjusting, the re-forecasting, and the re-computing operations until satisfaction of at least one implementation criterion from a predefined group of implementation criteria consisting of:

that the set of metrics corresponds to the predefined range;

that at least n adjustments have occurred where the set of metrics has not gotten closer to the predefined range, wherein n is a positive integer greater than or equal to two; and

that the adjusting, the re-forecasting, and the re-computing operations have been repeated at least a predefined number of times or over a predefined time period.

7. The method of claim 2 , wherein the set of metrics comprises a user engagement metric.

8. The method of claim 7 , wherein the user engagement metric comprises a repeat gameplay metric.

9. The method of claim 7 , wherein the user engagement metric comprises a gameplay duration.

10. The method of claim 2 , wherein the set of metrics comprises a gameplay metric.

11. The method of claim 10 , wherein the gameplay metric comprises a minimum win rate or a maximum win rate.

12. The method of claim 2 , wherein the set of metrics comprises a revenue metric.

13. The method of claim 1 , wherein the gameplay unit is a game level, the customizing of the game comprising provision of a custom level variant to a targeted sub-population of players based on the historic player model values of the target player group.

14. The method of claim 13 , wherein the custom content set comprises interactive content items prompting responsive user gameplay action during play of the game level.

15. The method of claim 13 , wherein the custom content set for the custom level variant is generated based on at least the player model and previously existing game content items for the game level.

16. A non-transitory machine-readable medium storing instructions which, when executed at a game server, cause the game server to perform operations comprising:

accessing historic player model values for a target player group of players forming a subset of the multiple players of the game, the historic player model values for the target player group comprising multiple respective values for a multi-dimensional player model, the player model values having been generated based on previous in-game behavior of the target player group;

in an automated operation using the historic player model values for the target player group as seed information, generating a candidate custom content set comprising a plurality of game content items for a gameplay unit of the game, the candidate custom content set being customized specifically for the target player group based on the associated historic player model values;

forecasting, using the historic player model values of the target player group, a predicted sequence of user actions during gameplay of the gameplay unit having the candidate custom content set;

using one or more computer process devices configured therefor, computing, based on the forecasted sequence of user actions modeled gameplay performance of the unit of gameplay;

comparing the modeled gameplay performance to predefined implementation criteria; and

conditional on the modeled gameplay performance satisfying the implementation criteria, customizing the game for one or more of the target player group by inclusion thereof of a custom variant of unit of gameplay, which custom variant includes the candidate custom content set.

17. The machine-readable medium of claim 16 , wherein the computing of the modeled gameplay performance comprises computing, based at least in part on software-defined outcomes of the forecasted sequence of user actions in the game, a set of metrics associated with playing of the gameplay unit consistent with the historic player model values associated with the target player group.

18. The machine-readable medium of claim 17 , wherein the customizing of the game is based at least in part on the set of metrics and comprises:

determining whether the set of metrics corresponds to a predefined range; and

upon determining that the set of metrics corresponds to the predefined range:

storing the custom content set for transmission to a client device associated with the player model.

19. The machine-readable medium of claim 18 , the operations further comprising:

repeating the adjusting, the re-forecasting, and the re-computing operations until satisfaction of at least one implementation criterion from a predefined group of implementation criteria consisting of:

that the set of metrics corresponds to the predefined range;

that at least n adjustments have occurred where the set of metrics has not gotten closer to the predefined range, wherein n is a positive integer greater than or equal to two; and

that the adjusting, the re-forecasting, and the re-computing operations have been repeated at least a predefined number of times or over a predefined time period.

20. A game server comprising:

processing circuitry; and

a memory storing instructions which, when executed by the processing circuitry, cause the processing circuitry to perform operations comprising:

accessing historic player model values for a target player group of players forming a subset of the historic player model values for a target player group of players forming a subset of the multiple players of the game, the historic player model values for the target player group comprising multiple respective values for a multi-dimensional player model, the player model values having been generated based on previous in-game behavior of the target player group;

in an automated operation using the historic player model values for the target player group as seed information, generating a candidate custom content set comprising a plurality of game content items for a gameplay unit of the game, the candidate custom content set being customized specifically for the target player group based on the associated historic player model values;

forecasting, using the historic player model values of the target player group, a predicted sequence of user actions during gameplay of the gameplay unit having the candidate custom content set;

using one or more computer process devices configured therefor, computing, based on the forecasted sequence of user actions modeled gameplay performance of the unit of gameplay;

comparing the modeled gameplay performance to predefined implementation criteria; and

conditional on the modeled gameplay performance satisfying the implementation criteria, customizing the game for one or more of the target player group by inclusion thereof of a custom variant of unit of gameplay, which custom variant includes the candidate custom content set.