June 2018
The algorithm theory (now proven?) does make some sense.
I've noticed odd behaviour. Such as my opponent starting with a score 10-20K when I hadn't even got a chance to see/kill the first unit on the board yet.
Something doesn't make sense there...
In fact, it CAN'T make sense...
June 2018
June 2018
"I suppose it would be a good experiment to write down the scores and compare them over a long progress,"
Okay now! Tonight I decided to put this down to a test... I decided in the art of science, to burn off a bunch of gauntlets and check the results. I was in the middle of a decent streak when I started this. Below are the results.:
Opponent Left, Me on the Right (yeah, I know this is reverse from how we normally look at it. Sorry)
1,061,300 * 1,602,200
1,113,800 * 1,755,400
1,915,650 * 1,032,000 Lose Streak
---
299,750 * 597,550
331,800 * 1,039,000
714,200 * 2,008,900
666,000 * 1,198,500
795,650 * 1,778,450
950,300 * 2,116,750
1,118,600 * 2,323,450
1,214,800 * 1,217,100
[Forgot to record one (I won though)]
2,011,450 * 1,866,750 Lose Streak
---
347,750 * 319,500 Lose Streak
---
176,350 * 36,000 (I won with less points somehow? But game crashed! lol) Bugged game!
---------
I stand behind my original theory. And it should be rather obvious from the results, the deeper the streak you are, the tuffer the opposition. The points you have to beat creep up fast, and when you finally restart that streak, you can expect up to a factor of 10 times the easiness to return, until you climb back up.
Note: I purposely played a lot of weird and horrible strategies for variance in the results (and so I could lose a few times), like boards full of lightning reeds and other JUNK. So you all had better appreciate what I did for you. lol
June 2018 - last edited June 2018
@BBnot8 wrote:According to this post
https://www.reddit.com/r/PlantsVSZombies/comments/8jbpjs/unofficial_battlez_mechanics_faq/"Bot is an algorithm, not a replay."
Well, I was thinking...
Actually we don't need internet connection during Battlez games. We just need it at the beginning to find an opponent and at the end, but not during the game itself.
If it was a replay, it would make sense, just like when you are playing against a ghost replay on another game.The replay data would be transmit to our device at the beginning, and then you could pause, go faster, would not be a problem, etc.
But if it is just an algorithm, then that mean that in the ~5/10 seconds used to find an opponent, the algorithm has the time to do all the plants setup, all the Battlez game, and transmit all of this this to our device ? I'm not an expert but it seems less credible somehow...
Edit: Or maybe the matches are already done and transmit before you hit the Play button, would make more sense.
June 2018 - last edited June 2018
@D0CT0R_Z0MB0SS wrote:"I suppose it would be a good experiment to write down the scores and compare them over a long progress,"
Okay now! Tonight I decided to put this down to a test... I decided in the art of science, to burn off a bunch of gauntlets and check the results. I was in the middle of a decent streak when I started this. Below are the results.:
Opponent Left, Me on the Right (yeah, I know this is reverse from how we normally look at it. Sorry)
1,061,300 * 1,602,200
1,113,800 * 1,755,400
1,915,650 * 1,032,000 Lose Streak
---
299,750 * 597,550
331,800 * 1,039,000
714,200 * 2,008,900
666,000 * 1,198,500
795,650 * 1,778,450
950,300 * 2,116,750
1,118,600 * 2,323,450
1,214,800 * 1,217,100
[Forgot to record one (I won though)]
2,011,450 * 1,866,750 Lose Streak
---
347,750 * 319,500 Lose Streak
---
176,350 * 36,000 (I won with less points somehow? But game crashed! lol) Bugged game!
__________________________________________________________________________________________________________________
I wanted to clear out my backlog of recordings. While I was doing it I took note of the scores.
I had tried 3 or 4 different set-ups. Apart from the mandatory Strawburst the common selected plants were - Primal Peashooter, Cold Snapdragon and Blover. I started off with Explodo-Nuts at the front, backed up with Grapeshot. I then backed them up with Stallia, and then Bombegranet. Neither method worked that well, so dropped the Explodo-Nut and went instead to Chard Guard (Imitated.)
That's the load-out I'm now using and have been since having my run in with a cheater/EA Killer Bot.
ME THEM
1,944,150 1,862,900
2,423,100 1,389,000
2,528,000 377,200
2,188,000 446,200
2,374,150 818,450
2,279,550 262,200
2,499,350 1,671,250
2,485,000 1,321,850
Forgot to record, but won the match
Forgot to record, but won the match
2,324,300 915,700
Recording file corrupted. Won the match with a score lower than opponent.
3,134,300 5,791,600 - Opponent had 3 disallowed plants in load-out. (Cheater or EA Killer Bot?)
3,397,900 2,295,950
3,211,850 1,574,200
3,093,150 1,981,750
3,880,000 867,900
2,722,550 1,439,300
3.071,700 1,511,700
2,946,700 2,718,100
2,809,200 481,800
3,167,400 703,600
3,301,250 1,751,150
2,974,300 662,850
3,008,450 1,268,800
3,059,900 2,093,000
3,130,300 699,050
2,991,050 1,625,300
2,589,800 592,850
2,704,750 637,200
2,997,000 2,190,750
2,805,500 399,950
2,923,250 978,700
3,175,350 2,628,500
2,940,850 2,174,850
July 2018
Here is a sequence of screen shots from last week (when I was unceremoniously dumped to soil league)
the opposition scores do not ‘creep’, they are low and then ‘bam’
July 2018
Next 3 in sequence
July 2018
Next 3
July 2018
BAM
July 2018 - last edited July 2018
Geeze @Humplemann, just when I thought we had it finally figured out, you had to post those results of yours. I guess I should have run a much longer series of progressions to be sure.
LOL at the 5+ mil opponent with 3 illegal plants.
I am starting to think now, when EA says "Please report hackers here-" what they are really meaning in-not-so-many-words is:
"Was one of our bots a little too obvious? Please report it privately here, and details involved in that game so we can have our main coder attempt to cover its tracks better in future gameplay, so that it can not be spotted as easily."
LOL!