[Discussion] High Scores

[jwilde35]

NOTE : ALL GAMES SHOULD HAVE THE SAME SETTINGS THROUGHOUT ~ ErmiyaEskandary

is it okay to make an issue about our high scores?? like boasting about our high scores and what settings we have the bot on?

if so, these are my settings/ highscores from last night, take note of the high score XD

[ermiyaeskandary]

Did you achieve that highscore with those settings throughout ? @clemens-tolboom PR which records the time as well would be useful in this case.

[Fredyy90]

80k is my highest score so far.

[ermiyaeskandary]

@Fredyy90 Did you use the same settings all through out ?

[Fredyy90]

Yes didn't change a thing, just had the window in the background running for ~72-96 hours

[jwilde35]

i left the bot on over the weekend at my school and these were the results

note: this was using the task execution mod

[clemens-tolboom]

@jwilde35 nice so those huge highscores for my #275 but I'm really puzzled in how to understand any scoring data.

Having an average is nice to have. Having a median too but what does it say? Using the data till now

• from 192 game average 9276 and median 5428 can still tell not enough about the bot performance. We do not know how bad the scores below the median are. I've seen lots of 12-40 points in some runs myself. We really need more data #286

@Fredyy90 ypur scoring has no median in it Guess you used an older version right?

[Seple]

Please test settings #290

[clemens-tolboom]

I just ran 4 concurrent (develop) bots for 2:12 hours (and recorded it into a 160GB video make system low on memory)

#	a	m	max
21	3914	2378	14510
10	6989	4867	26334
13	6449	1467	26959
17	5660	3271	18321

[grovtor]

Can scores be saved to the log or a desktop text file?

@jwilde35 how does one install that? Explain it to me as if I were 14

[ermiyaeskandary]

@grovtor If you mean the task execution feature, install this

[clemens-tolboom]

@grovtor I use the HTML inspector then select the scores and paste it into a text file.

[tjorim]

For testing, and easy score saving, I would suggest to use https://github.com/K00sKlust/slither.io-bot-manager .

[clemens-tolboom]

I talked to a statisticus and he suggests it's not about the absolute score but the score made relative to the others. It is of no use to register 1000 as a bad score when you end up as # 1 where # 2 has 300 points.

That why I plead for more score info in #286

[ChadSki]

I'm not sure I agree that being #1 on the server with just 1000 points is very impressive. Ideally the bot simply doesn't die that small, and continues to grow even if other snakes are small.

Since we only consider comparing scores valid if you're running both bots at the same time, they should be facing roughly equal server conditions anyways.

In theory collecting extra information won't harm anything, but I'm concerned that people will be using the wrong information to make judgements.

But, I guess it would be interesting to see. ¯(ツ)/¯

[phrokton]

The SnakeParamsBySize of the Task Execution branch seems like it would be best for unattended use especially as snake grows to the point where turning requires an absurd radius.

edit: oh, https://github.com/ErmiyaEskandary/Slither.io-bot/issues/290

I have not yet convinced myself I can edit that section to create more size brackets because it is not clear if those brackets connect to some other part of the code.

[clemens-tolboom]

@phrokton huh ?!??

[jwilde35]

@phrokton you can... i have but did not save it... works fine