Some weeks ago with nullpointer on the telegram chat we had a discussion (at the end we disagreed) about the following point: I was insisting that a stable top3/5 player, playing after a week (or more) of pause and without changes, would lose most of the points falling in top10 or top20. This under the condition that the game mechanics would not change much. Nullpointer did not see it in this way.
Now in this period I did not play much, neither I modified my AIs during pauses, still I was impressed that when I played from time to time, meeting top20 players, I was able to hold the ground instead of just losing. Checking other players seems to confirm that, for example Tralalo did not play much between February and March and still holds his spot every time he plays again.
Of course after a while one changes little stuff, but the point is that "the first impact after a long pause is not devastating".
So, it seems that nullpointer was more near to the truth for the observations made.
Still I have the below ideas about evolution of player strength.
Given a range for scores (1), say 1000 as the first score and 2000 as the maximum score, and the current tiered leagues (there are 7) with XP levels (there are 7) to unlock bot classes and AI controls. Assuming that the top 2 players are around 2000 points, top 5 around 1900, top 10 around 1800, top 20 around 1700. I would say the following.
- from match 1 to 150 one is still discovering the basic qualities and weaknesses of bot classes and basic functions of AI controls.
- after match 150
-- up to 1400 score points: the player has still to work out the basic structure of his AI (whatever is a generalized AI, a generalized AI per bot class or specialized per map and bot class). This means: the player did not yet define in a stable way how his bots should behave in attack, retreat, resource capture, and so on. From time to time the player makes quite big changes for certain behaviors.
-- 1400-1600 points: the player has a basic structure, but now it starts to be complicated because games against stronger or equal players are now decided by smaller differences, or optimizations. This means that the general behavior from the AI when the player was 1400 is not so different, but some small changes in nodes, filters and priorities helps a lot in terms of results.
-- 1600-1700 points: the player starts to have the first useful subtle optimizations to hold his ground. So, once again, the basic behavior of the bot is there since the player left 1400 points, but now it is more refined to handle specific situations instead of the generic case.
-- 1700-1800 points: even more optimizations and this is the level, in my idea, that can consistently snatch points from the top10.
-- 1800-2000 points: way more optimizations than before, but, and this is my idea that so far did not hold the observations, I thought that the continuous competition between players 1700+ would produce more and more optimizations for particular cases and/or increasing efficiency in scoring balls or pushing or damaging the enemy. This means, in my idea, that a top 10 player not playing for a week, would be at least 300 games of experience behind the current optimizations. Therefore losing a lot of ground before catching up and going back to his previous spot.
End of the exposition of my ideas about evolution of player (if it is not clear, ask!).
Anyway, this seems not valid. At least for the check that I did, through the stats page or personal experience, it seems that top 10 players not playing for more than a week (and likely not editing their AI during the pause) just hold the ground without losing a lot of points. Surely they lose a bit more than before, but still not horribly, before catching up.
Why is it so? I do not even have partial answers at the moment. What do you think about it?
(1) If one does not set a range, scores are meaningless because while the starting score is still the same as in November 2016 (1000), the top scores are increasing.