Gold Battle KoTH

the bug is fixed

I was using an old version of the controller

@dzaima I'm implementing the feature for recording all attacks toward a bot. Are there any other bits of information you want saved between rounds?

@RedwolfPrograms i've already got access to the moves, what i need is to parse the data with accuracy

@dzaima please elaborate

@RedwolfPrograms elaborate what? my controller already has the moves taken available in the table (and of course in JS), all i need to do is save all of it in some big arrays and manipulate it until i get the exact results i want

12:46 AM

"should probably make a thing that calculates which bot started attacking mine to know what things i should defend against"
and
"and doesn't report bots which just hit mine once at a random point, and also ones that joined in late to the kill party"

So do you not need these anymore?

@RedwolfPrograms getting the attacking data is easy, what's hard is analyzing what's the cause and what's effect, plus probably a couple more bot-specific metrics

I'm storing attacking data for a single bot as an array of bots, each containing an object with an array of attack turn and attack damage

Then, the order of attacks could be used

What's the end goal for the data to be used for?

@dzaima

@RedwolfPrograms make bot lose less, obviously

Yes, but would the goal be to find the first bot that attacks, the point at which other bots start to "jump on the bandwagon", the turns until the first attack, the average damage dealt by attackers as a function of the turn number, etc.

Each one of those would be approached differently, but easy enough to compute

@RedwolfPrograms i don't know, all i know is i want data but i haven't gotten around to parsing it

aka don't worry about my data gatherings, i'll be fine

12:57 AM

Oh, okay. I guess I misunderstood. Sorry!

1:08 AM

pastebin.com/0PXET7jg

1000 round tournament

Don't worry, my bot Tweedledoom is going to be here soon to wreck the teamers

I'm trying to decide if targeting two teaming bots counts as "targeting a bot" or "targeting a common strategy"

if it applies to any 2 teaming bots, its targeting a common strategy

if it only works on just 1 team, its targeting a bot(s)

^my interpretation

My Tweedledoom bot successfully stops the strategy

1. Kill Assist: 39642.67
2. TeamBot - Attack: 34922.799
3. The Accountant: 33600.836
4. Switzerland: 30061.553
5. Kill Stealer: 27101.514
6. The Bot That Farms, Attacks, Shields, And Even Heals But Never Stuns: 22786.407
7. Unkillable: 21911.927
8. ScaredBot: 21767.274
9. CampBot: 13131.122
10. Captain Farmer: 12026.521

(1000 rounds)

Kill Assist will still be in the lead if I post Tweedledoom, though it is very tailored to the TeamBot twins

So I may have to make it more fair

pixma140

7:27 AM

Don't know if this was already asked here, but could this challenge be modified into some last bot standing challenge? Or was this already a challenge here on this platform? I really like the setting and the methods the bots have here. Would be interesting if the winning criteria would change -> for a new entry I mean.

12:52 PM

still seem to win with the new bots
https://tio.run/##vVhLbhzJFdz7FHkBF/J98rcUbcsDGMIIpnyAEtkzak@zSTebY4grw2svvNDCh/DSJ9JFNPFeZn@obskEBIxAJQrdVZVREfHiveqref3zfP/p08cP//r44R/f9vffZ3/7nONn3vZbTv71//79m48f/hme/rt5f3G7ZXyuUbNMrWUcSy5YaxWsOfonpKL2IVbVkhOrXW33y8whCAs@SElyoEgxBSpJMz5mbqHkljXUqtRCq9ISLnsK5U/L1Sq8uL9f3m99@1R5YrXzmBRrS8kBRayUMku1g5ZUSaK0AaWySKBcOAfmUjlwiTkFEU1YW6w5aMJzhEQE0FlzK0@hvHm3CC@urm4f1tt5bVA4K9GU2G5PDqg5DcVXiipJ/CCRllQoDyg5UgmhFimBcB0FfN0AiEEtWLFjKUoaVFtsIYEnOWHl8u/L7eNis5rX1wYFz8GTJhMrNHZWTJpcyRCQFFctVK0gJbc0qUNJiueGBpRCaIVrIKaKNQuOOQtRkCytBI3VGIJM@QQKwFzNm8U1zGKbgTidqPh@wbT…

are you running the latest version of the controller?

@dzaima

@SuperStormer is there some major change to how it functions?

all i see are logging changes, but i'll update anyway

i ran a 1000 yesterday that had top 3 scores >40000

@SuperStormer well my bot kills everyone before then

I know the 1st 2 columns are avg score and total wins; what are the rest?

1:07 PM

same as before, just forgot to copy-paste
https://tio.run/##vVjLbhvLEd3nK3qZLNLoqq5@bQJYSZwLBMY1ImcbYEzxWowpSqEoO9IqyDqLu/AiH5Flvuj@iHOquvmQSTkCDFxDbgw4w@kz55w6VcPZtPow3X7@7J78N314h/XjYnXrfkPv8T@8d@5q@ns/9cuP1@vN5V/4V85R2Fw6x7ZGW8XWZGu2tdhabW263szXs/lqs1jOf/HTp3/99Okf3/b3n2effc7xM2/7LRf//H8/guh/fqHx1f3Z9YbxuQTJ0beWcRxzwVprxJqDfUISRT/EKlJyYtFv6/0yM1TniA9SihlmoJAclSQZHzM3V3LL4moVaq7V2BK@9hjKHxfLpXtxe7u43dj2qbJn0euYBGtLyQAFrJQyx6oHLYlQDLENKJVjdJQLZ8dcKjsuIScXoySsLdTsJOE5XCIC6Cy5lcdQ3lzO3YvZ7PputZlWCoWzEPnEensyQM1oKLZSkJiiHSSSkgrlASUHKnB7icURvkcOpxsAMagFK3oci5A4kRaaS@ApHrFy/nGxeZivl9PqQqHgOdhLU…

my bot pair has an average chance of 97% of identifying the other bot now, but i really want to make a smarter algorithm for the starting communication :|

i guess more important would be to get a way to switch the target to another bot if my other bot is stunned

@pixma140 I think the challenges would be too similar like that, but I am thinking about a similar contest

i figured out why my bot is doing bad

because @Draco18s modified their bot to do the same as my bot to mess up ID

1:23 PM

for fun, i wanted to see what would happen to the scores if every dead bots worth at the end of the round was divided by 2
https://tio.run/##zVlNbxvJEb3nV/QxOWTQ1d99CWAlcRYIjDUi5xpgTHEtxhSlkJQd6xTknMMefMiPyDG/aP@I87qqejgSyUDG7mEJeZocDrvfvKr3qnq8GDcfxt2XL@b8a/zwDsePq83O/Ibe4599b8zN@PfD17/8eLvdX//F/ap9Jru/Nq4d8PI6Bh2jjknHrGPR0dT25m65XSw3@9V6@YsfPv/rh8//@HF///maC57z/mvn/0l@8vP/@x7R@ueT5Ln5dHG7dzgfHIUwOKoRHzwfiyMcY0ntDGUbvTHtbS411VB8lSnapI7cEA35ECWbbJZkCbZlXI45tNPR1nY@1@qpJRVexBlpYw080WOEf1yt1@bFbrfa7XHeVZfy4Bxjc5RxrMHimFNDSCVjAZ4Fd@J8cTEdEAabAxAW3AUDtYlXziEQPrtgS8LonbMNsa/EyEKOpeC8iZV/8Bjhm@ul…

I'm working on a less-cheaty Tweedledoom...get ready, TeamBots

@SuperStormer could that be called targeting a bot? though it is only 4 moves

pixma140

@RedwolfPrograms is there a possibility to get a ping on that if it happens that this challenge will come up? :D

@pixma140 Sure!

It'll probably be a few weeks from now, so there'll be enough time to review the controller, make the rules better (but no too different), etc.

Do you think getting gold automatically on every ~5 turns would be good?

That way, bots could upgrade their auto-gold so they don't have to worry about farming so much

would that gold count towards worth?

1:31 PM

No

Or maybe if it were upgraded, the amount added would

So if it were, say 5 gold every 5 turns, and you upgraded it to 6, only the one extra per 5 turns would count toward worth

1:44 PM

Currently, Tweedledoom finds an average of 6.4 bots per round it considers "likely teams"

It also captures a few of the meaner bots in that category, like BullyBot

@RedwolfPrograms does TeacherBot messing up TeamBots' ID count as targeting a bot?

I'm honestly not sure. I think we need to decide on a few more specific rules if this challenge is going to end up like this

I'd say it counts if it could be reasonably considered an attempt at targeting a bot.

@RedwolfPrograms good, then my strategy is definitely safe

What's your strategy?

20 constant random moves tracked each turn, but i'd really like it to be upgraded to one where it's continued forever until both bots have both identified the other and know that the other could've identified the other, plus error protection, etc

1:56 PM

Here's the targeting efficiency of my anti-team tweedlebot:

bullyBot: 97.4%
TeamBot - Stun: 94.8%
Undyable Bot (v3): 94.6%
TeacherBot: 94.2%
TeamBot - Attack: 93.4%
Unkillable: 90.8%
ThanosBot: 48.6%
Rando: 25.6%
FarmHeal Bot: 0.8%
JustFarm: 0.2%

(500 rounds)

It finds an average of 6.4 enemies per round

If it stuns one on the turn it analyzes, it has an average of a 1/6.4 chance of taking out a team ATM

"taking out" ≠ "knowing which is the team"

I'm testing out a new, potentially more accurate method

My new method is 1/5.3 instead

Barbarian Man: 90.2%
Kill Assist: 89.8%
TeacherBot: 89.4%
TeamBot - Stun: 89.4%
TeamBot - Attack: 86.6%
The Accountant: 85.8%
Persistent Blue Shell: 0.2%
Rando: 0.2%

I think I'll go with the first. 31% is good enough, I suppose

2:19 PM

This is my tweedlebot:

function tweedledoom(me, others, storage) {
var enemies = others.filter(el => el.worth == 5);
var hp = others.map(el => el.hp);
if (turn() >= 4 && enemies.length)
return attack(enemies[hp.indexOf(Math.max(...hp))].uid);
return shield();
}

@SuperStormer the Teacher punishes bullies

It doesn't target specifically the TeamBots, it goes for the bots which have farmed once in the first three turns, which contains BullyBot, TeacherBot, the TeamBots, and a few more sometimes

Tweedlebot is going to be posted now

Oh wait, it's called Tweedledoom

@RedwolfPrograms oh so it won't work on any teaming bots, just a specific starting strategy..

Making it work on any teaming bots would be impossible to figure out

But by using this strategy, it helps take out some bullies as well

I should name it AssistantPrincipal

In a surprising turn of events, the TeamBots attacked my Tweedledoom, then TeacherBot finished it off...

[87-] Tweedledoom: Stunned by TeamBot - Stun
(index):4271 [88-] Tweedledoom: Stunned by TeamBot - Stun
(index):4271 [89-] Tweedledoom: Stunned by TeamBot - Stun
(index):4271 [90-] Tweedledoom: Stunned by TeamBot - Stun
(index):4271 [91-] Tweedledoom: Stunned by TeamBot - Stun
(index):4271 [92-] Tweedledoom: Stunned by TeamBot - Stun
(index):4271 [93-] Tweedledoom: Stunned by TeamBot - Stun
(index):4326 [93] TeacherBot killed Tweedledoom
(index):4326 [93] TeamBot - Attack killed Tweedledoom

Teach isn't above getting free money. It's woefully underpaid.

2:25 PM

XD

I like Tweedledoom

It's a nice bot

He takes out bullies for the fun, not the money

:P

(;;;)

Teacher is just there to buffalo the buffalo bots. The tweak to spoof as one of the TeamBots isn't even a real targetting of the TeamBots: Teach doesn't make any attempt to identify them.

True, I suppose

I did try to run it as if(turn() < 5) { return attackBot(...); } but dz's pretty controller didn't like that

(the bot code ends up sandboxed)

2:30 PM

@Draco18s if you call sandboxed eval("["+code+"]")[0], then yeah

yesterday, the competition really skyrocketed

@RedwolfPrograms is calling another entry considered cheating?

would've been more resilient to changing passwords, but more...explicitly targetted

@dzaima "put in a place not immediately accessible"

@SuperStormer "Bots may not read, modify, or add any variables outside of their scope", and functions are variables. and the challenge doesn't guarantee that bot functions will exist either

@SuperStormer Yes

2:32 PM

that, too

1. Kill Assist: 48799.226
2. TeamBot - Attack: 34443.005
3. Kill Stealer: 33876.885
4. The Accountant: 32515.124
5. Switzerland: 31657.641
6. The Bot That Farms, Attacks, Shields, And Even Heals But Never Stuns: 28983.968
7. ScaredBot: 24722.732
8. Unkillable: 19323.435
9. CampBot: 19322.236
10. Barbarian Man: 16538.063
11. Friendly bot: 14477.035
12. Captain Farmer: 13157.775
13. TeacherBot: 11958.827
14. Undyable Bot (v3): 10263.965
15. Bandit: 9255.75
16. PatientStrategistBot: 8887.394
17. Indestructible: 7162.564

(1000 rounds with Tweedledoom)

is calling another one of your own entries considered cheating?

Yes

Tweedle doesn't score very high XD

runGame(10000, 0)

1. Kill Assist: 47722.34
2. <--------The Accountant-------->: 34717.727
3. Kill Stealer: 34481.53
4. TeamBot - Attack: 30528.753
5. The Bot That Farms, Attacks, Shields, And Even Heals But Never Stuns: 30110.666
6. Switzerland: 28579.947
7. ScaredBot: 21866.673

back to second

2:33 PM

@Draco18s Hey, he's not one of these corrupt bully hunters, ya know (:

hehe

i just realized that my controller cuts "The Bot That Farms, Attacks, Shields, And Even Heals But Never Stuns" short, and that that's a coincidentally good thing

Hold on, does my Tweedlebot actually help TeamBot?

idk. just updated accountant 3 min ago. Maybe u have old one?

I'm running a game rn

2:37 PM

k

im running updated game with 10k as well

I probably do have the old one, then

1. Kill Assist: 44255.752
2. Kill Stealer: 36271.07
3. The Accountant: 34163.336
4. TeamBot - Attack: 29773.261
5. The Bot That Farms, Attacks, Shields, And Even Heals But Never Stuns: 29496.463
6. Switzerland: 28020.106
7. ScaredBot: 22014.858
8. CampBot: 18227.322
9. Unkillable: 17012.967
10. Barbarian Man: 15048.993
11. Friendly bot: 12791.018
12. Captain Farmer: 11709.987
13. Undyable Bot (v3): 8936.051
14. Bandit: 8562.452
15. TeacherBot: 8270.541
16. PatientStrategistBot: 7514.544
17. Indestructible: 6473.17

(Without tweedledoom)

how many rounds?

Yeah, Tweedledoom seems to be helping the TeamBots

100 rounds

1000 i meant

But that was with updated accountant

Does that affect it, maybe?

huh

ill see what i get with 10k soon

How long does 10k take for you?

2:40 PM

5 min

Desktop, I assume?

figured out google dev tools much faster

no, laptop

visual studios was really destroying preformance

pixma140

@RedwolfPrograms could be a good idea. if everyone gets it, i would count it to the worth because it is the programmers decicion what to use the gold for. investing it on upgrades or saving it. so yeah

Also benefits survivors

I was thinking instead of automatically getting gold every few turns, there should be some sort of purchaseable item for 25 or 50 gold that gives you gold every turn, so that would still benefit survivors, but make it an investment

2:42 PM

i have the i7 8650U

So you could invest your starting gold if you think you'll be alive long enough

@B0RDERS Oh, I guess that beats my Celeron N3500 by a landslide

ya. double the cores, double the clock speed

Hey @RedwolfPrograms, is it allowed to submit 3 bots?

As many unique bots as you want

Nice >:) (evil face)

2:45 PM

I like the face from Gold Collector KoTH, (;;;)

@RedwolfPrograms, it has been 5 min and no results. i might have lied

@B0RDERS The 40 bots are kind of slowing things down, I suspect

@Night2 oh no will i have to upgrade to 3

@dzaima Then I might do 4 (Just kidding, don't have time)

I'll do 8, just to make sure I'm in the lead

2:46 PM

who wrote the teaming bots?

i think kingslayer and squire would have been better names

HealBot, AttackBot, ShieldBot, StunBot, FarmBot, UpgradeBot, SkipBot, and ErrorBot

Those're my bot ideas

@B0RDERS It was dzaima

Is the dead line "Challenge ends at 1700 UTC on Friday, August 9th" for edits as well? Can I post my bots at last minute and be sure no one else can do "anti of my bots"?

@B0RDERS the teaming bots? many here seem to be writing some

The equalizer, smartFarm, sniperBot, the accountant -->4 bots

@Night2 Technically, but that's not really fair play

2:48 PM

@dzaima, meant tembot - attack and teambot - stun

10k runs, standings as of 9:39:

1. Kill Assist: 47283.2403
2. The Accountant: 34595.3545
3. Kill Stealer: 33895.4126
4. The Bot That Farms, Attacks, Shields, And Even Heals But Never Stuns: 29444.224
5. Switzerland: 28387.6452
6. TeamBot - Attack: 28228.4719
7. ScaredBot: 22062.5997
8. CampBot: 19622.4811
9. Unkillable: 17905.0672
10. Barbarian Man: 15810.8393
11. Friendly bot: 13280.4554
12. Captain Farmer: 12055.377
13. Undyable Bot (v3): 9879.1569
14. TeacherBot: 8966.2118
15. Bandit: 8854.5291
16. PatientStrategistBot: 8012.5468

@RedwolfPrograms The thing is, it doesn't matter how well you strategize your bot(s), someone reading it/their codes can update their bot to be anti it. I can add a few lines to get targeted less by Teacher or Team bots and that will give a large boost to my results.

@B0RDERS that was @SuperStormer

@Night2 I think that was a mistake in design on my part, we're mostly just holding everyone by the "no targeting specific bots" rule

So I was thinking to post at last minute to avoid that thing

@RedwolfPrograms Yeah, defending your bot against other bots won't be count as targeting.

I think the teaming needs to be restricted a bit to keep things fair.

2:53 PM

Also my actual main concern is the team detection method. As you can see, with 5 minutes of time, someone can understand how team bots are detecting and they can drop something random which won't look like targetting in their bot to make team detection fail.

"different enough" and "not targeting" are hard to objectively define or judge

which makes these concerns harder to regulate

I think maybe teaming will be allowed technically, as long as the bots don't perform checks to identify each other

So, two bots could have strategies which work well together, but they can't identify each other

@RedwolfPrograms Exactly, in general all top contenders will arrive to similar strategies to win, just from different ways. So most of top bots will be similar in majority of things they do, like farming, upgrading and doing the last blow(s).

@RedwolfPrograms so the current posted team bots would be banned?

@RedwolfPrograms Then you have to check every ones code, someone can randomly decide to shield a lot around turn 100-200 and their other bot might be designed to not attack bots that shielded a lot without specifically detecting the other bot.

@dzaima Maybe I'll just leave a comment telling SuperStormer to remove the identification code, but he can keep the overall concept

2:58 PM

@RedwolfPrograms If that will be the rule, then my 2nd and 3rd bot ideas will be cancelled :P I will have to think more now

@RedwolfPrograms i doubt it'll work at all without identification

@Night2 It's mostly just the honor system at that point, we're all here to have fun, and I doubt anyone's going to try to cheat beyond little loopholes

@RedwolfPrograms well this is more like canceling fun

Y'all are right

I'll keep thinking

after explicitly having allowed it

2:59 PM

I would say allow team detection but limit bots per user to 2 or 3

So on last minute, not so many new bots are going to get posted

Or maybe allow only one bot per user to be posted on the last day

And limit bots posted after this point

@RedwolfPrograms That too, but we are already on last day in 2 hours, so many people won't know about it!

I don't think there are too many more people creating submissions, but I'll leave a blockquote at the top of the page stating that

So that means in 2 hours anyone with a bot already posted can only post 1 more bot, right?

Maybe within the last 12 hours

3:02 PM

Edit: anyone with any bots

is IDing teammates legal

We've agreed it is

3 mins ago, by Redwolf Programs

I'll keep thinking

I tried a thing. TeamAssistBot. It joins up with TeamAttack/Stun.

Though we're still debating some of the rules for teaming

3:03 PM

@RedwolfPrograms This could be better, but you should also think that someone can post a new bot tomorrow and also edit their original bot too

after 1000 rounds it landed 8th with TeamAttack in 10th

@Night2 I think changing a bot beyond its original strategy counts as a new bot

It is generally hard to force these rules without putting much time on checking everything

@Draco18s Huh, that's cool

How are you going to decide if "original strategy" is changed or if it is a minor change?

3:05 PM

@Night2 Team ID or something like that would change the strategy

I'm working on my main bot now and I changed 1 single line to constantly get 2nd place.

@RedwolfPrograms I can put the team ID logic in my bot. but just edit a seed number

Will that be allowed?

I guess, as it's assumed trying to imitate your bot's team ID would be cheating

FWIW on the "dive into a dungeon and collect gold artifacts" challenge, teaming didn't work (but was allowed) because it was an elimination match: the lower performing bot would get eliminated when the next grouping was run and the bot that "performed well" no longer had its teammate.

Here everybot is always present at round-start. The more bots on your team the higher you can make one of them score

Maybe run games with only 10 bots at a time, so most of the time a bot won't have its teammates?

@Draco18s That is actually a nice idea

3:09 PM

That's interesting, but would change the strategy I think

What do y'all think?

@RedwolfPrograms for the current challenge, imo drastically changing anything this late is a bad idea. for a new one, sure

I'll try running a few games of 10 or 20 bots to see how it changes things

@dzaima I agree with this as well, since you are only 26 hours before deadline, that is an issue

Just destroyed a test:1. <-----------The Accountant----------->: 134277
2. <-----------The Accountant----------->: 130407
3. <-----------The Accountant----------->: 71177
4. <-----------The Accountant----------->: 65602
5. <-----------The Accountant----------->: 38146
6. <-----------The Accountant----------->: 28259
7. <-----------The Accountant----------->: 19136
8. <-----------The Accountant----------->: 10016
9. <-----------The Accountant----------->: 8865
10. <-----------The Accountant----------->: 8593

But I'll do that for the next challenge, which will be similar

3:11 PM

yeah, its way too late because of the bounty to do an overhaul on scoring or how rounds are run. The only reasonable thing is to ban or heavily restrict teaming.

I agree

Maybe not ban, but restrict

just as an example, I ran my TeamAssist bot as 3 clones (TA1 through TA3, was just a cheap way to run a coordinated team) and they all came in 2nd, 3rd, and 4th behind Kill Assist.

I think the key is in the bots being sufficiently different. You could argue either way for the TeamBots, but that's what I'd consider to be the most important factor

The point being that making "distinct bots" that treat each other as allies is predominantly irrelevant to getting at least one of them into the top 3

@Draco18s but does mean that you can't post 5 of the same thing

3:25 PM

I think teaming is bad for the challenge, but it's hard this late in the challenge to prevent it or change the rules to reduce its effectiveness

running a sim of 3 identical bots is just a fast-and-easy way of simulating a team of 3 bots

as a hypothetical system, imagine I write 4 bots: 1 farmer and 3 attackers. The farm bot keeps its HP+shields at a fixed level. If that level ever dips by more than 2 on any given turn, the differential is calculatable. The three attackers can see this and determine the most likely bot that performed the attack on the farmer, then gang up and out-DPR it into the ground.

@Draco18s but you'd have to make the 3 attackers sufficiently different

Each attacker would have to be sufficiently different

I would say limiting to 2 bots per user would be a good step which isn't going to harm, at least that is what I think, I have a 3rd bot idea already, but I think it would better to limit to 2 bots

Maybe not 2, depends on how many bots people already ahve

let me count

3:28 PM

@dzaima which is non-trivial

@RedwolfPrograms i think a 2 teaming bot limit is okay

I think 3

@Draco18s which makes your idea not work

er, distracted. I meant to say its trivial

I think 3, but enforce the different-ness rule more with teams

@Draco18s And not just slight differences; completely varying strategies

3:30 PM

eg:
Attacker 1: upgrades atk and shld while making random attacks at non-allies when the farmer is not threatened
Attacker 2: upgrades atk and shld while farming when the farmer is not threatened
Attacker 3: upgrades atk and shld while making attacks in a bully-like fashion (ie. picks a bot and murders it) when the farmer is not threatened

My ideal picture of teaming bots is something like one farming, one attacking and farming the rest of the time, and one attacking the whole time

I think two varying attackers is defendable, so a three bot limit

For teams

Honestly I think allowing team submissions drastically alters the challenge-as-presented

But it's a bit too late now to take that back

Without drastic changes in rules, or removing at least two bots

@Draco18s well, the challenge-as-presented currently already allows teaming

Not really. Declaring it as not allowed as being against the challenge intent just disqualifies the two team bots

3:32 PM

@SuperStormer What do you think?

@dzaima the team rule was added 21 hours ago, that's not "originally presented"

teaming bots should have each bot have very different roles

The original idea of teams came about due to locating a loophole in the rules.

Yes, Redwolf has said, "I'll allow it" but I'm saying that allowing it has massively shifted the playing field, effectively rendering existing answers obsolete

I'm almost in agreement with Draco

@Draco18s i agree that it should've not been allowed from the start, but imo it's too late to change now

3:36 PM

The only sensible solution at the moment with allowed teams is to score the team based on the team average score. Which the controller can't handle

Actually, that's pretty smart

I like it, and it only affects 2 bots at the minute

but then you have to define what exactly counts as a team

exactly

A team would be any 2 or more bots with a team ID

does TeamAssist count as being on a team, or not?

3:37 PM

Does it TeamID?

Oh, hold on...

its spoofing a team ID and acts as if TeamAttack and TeamStun are allies (even if TeamAttack and TeamStun don't always recognize it as an ally)

That means I could just make TrashBot and assign it to a random team

(I haven't posted TeamAssist, its just a local hypothetical bot)

the guts are basically TeamAttack except that it stores a list of all possible allies instead of just one

does a bot count as part of team if it doesn't spoof a team ID, it just identifies the team and works with it

I think spoofing a TeamID is basically targeting another bot

@SuperStormer Not sure

3:39 PM

cough TeacherBot cough

I think TeacherBot should have that removed

I think to be part of a team, you must be allowed to be by the person creating the original teaming bot

Do y'all agree?

And therein we have the issue. The original challenge didn't have anything to say about how teams worked, and now, with only a day or two before the bounty, we're having to hash out how teams are allowed and work. Spoofing a Team ID (what the heck is even a team ID in a game with no communication!?) isn't "targeting a specific bot" its "defending against a specific bot"

so what this means is i have to be sure my bot pair is twice as good as everything else

@dzaima No, either one has to be twice as good or they both have to be equally as good

@Draco18s targeting ≠ attacking, there are many forms of targetting

3:44 PM

I think we need to define targeting

53 mins ago, by Night2

@RedwolfPrograms Yeah, defending your bot against other bots won't be count as targeting.

defending based on specific things of a specific bot ≡ targeting

@Draco18s that wasn't said by RedwolfPrograms

But defending by spoofing a TeamID is different

TeamAttack/Stun is bots plural

@RedwolfPrograms Again, what does "team ID" mean in this game?

IMO in any way, shape or form depending on the specifics of a specific bot counts as targeting

3:46 PM

@dzaima that means that TeamAttack and TeamStun are in violation of the targeting rules. Because they "target each other" in order to be buddies

A TeamID is a specific set of intentional moves designed to make other bots recognize it

I think TeamIDing needs its own new rule

@Draco18s imo targeting your own bots is fine

It doesn't count as targeting if the bots are intended by their creator (or mutually by creators) to work together

I assumed targeting is attacking/stunning a certain bot, not a certain strategy

@dzaima "Bots should not be designed to target a specific bot, but can be designed to take advantage of common strategies"

3:48 PM

Common strategies between bots or teams, not bots on teams

@Draco18s it's not targeting if it's intended by both sides

Here's my proposal:

@dzaima "No, I am Sparticus!"

@RedwolfPrograms A bot may not be designed to interfere or attack/stun a specific bot or team, where a team is a group of 2 or more bots which use a system of moves in order to track each other if designed by the creator, or mutually agreed on by the creators.

I object on the grounds that the rules and playing field have been altered dramatically since the posting of the bounty.

If you want to have a team-based KOTH, that's a different challenge

3:52 PM

@RedwolfPrograms why so complicated? just "a bot may not attempt to identify(≈target) another specific bot unless that specific bot wanted to be identified by that bot"

Yeah, that better

at that point you may as well just add a team-id field to the others data array

I partially agree with Draco

It wasn't meant to be a team challenge

How many people have or plan on posting team based bots?

me, if it goes well

I came up with the idea, SuperStormer got to it before I did

as I did not seriously expect it to be legal

3:57 PM

@Draco18s me neither

I think that was my biggest mistake, right after stunning

@RedwolfPrograms stunning turned out pretty cool though

stunning as an action just looked useless, "cooperation" opens up a whole unintended dynamic

Yeah, but Persistent Blue Shell could have been bad if it were any good

I think stunning is cool in singleplayer; with multiple, it goes downhill quick

@RedwolfPrograms that says something about teaming, not stunning

3:59 PM

@RedwolfPrograms I have already a team bot in production and idea for a 3rd one, but I will completely be OK if you forbid it or limit it.

@dzaima, do you think teaming should be allowed?

@RedwolfPrograms I would've said no, but now is a different, much more difficult situation

I think limiting it is the best course of action

A good first step is the rule above

Then, limit bots per team to three

three is a huge number

Maybe two, depends

Three would require very strict differentness requirements

4:03 PM

just two (ignoring Teacher's spoofing) already dominated the existing bots

Because of stunning

Stunning should not be allowed for teams, IMO

and? ah

Or maybe if a team member stuns, it prevents the rest of the team from moving

@RedwolfPrograms i'd rather take banning teaming over overly specific regulations to it

I think teaming could be a fun concept, with a few regulations

4:05 PM

@RedwolfPrograms this requires the controller to know what teams are

14 mins ago, by Draco18s

If you want to have a team-based KOTH, that's a different challenge

I really like the stun skips the team's turn thing, but there would be issues with implementing it

TeamID would have to be recognized by the controller

also also, if a team knows that one member is going to stun, they'll ALL use stun that turn, effectively losing nothing

(and knock out three other bots)

But only one stun would do anything

@RedwolfPrograms why the focus on TeamID? teams is what matters, IDing is bot-specific (and either way, i'm strongly against making the controller worry about teams at all)

@RedwolfPrograms I misunderstood, but that has HUGE major implementation issues

even in a controller designed to handle teams

4:11 PM

@Draco18s Not really

Hmm. maybe not. Wasn't thinking about things the right way

I still think teams should have to ID themselves (no others[i].team or anything), but the controller should recognize when that happens

good freaking luck

2

botData[i].team = "TeamBot"

And then any bots with that team name must store the uid of their teammate(s) to storage.teamID or something within X turns

No, I hate that idea

@RedwolfPrograms again, changing the rules is one thing, but changing the controller is too much

4:14 PM

@RedwolfPrograms I would object and go straight to Meta about massive challenge alteration

I may still go to meta over the allowance of teams

Who agrees with putting it on Meta?

Alion

Even though I'm not participating, I would like to see what Meta thinks about this situation.

I'll put it on Meta, then

I appreciate that.

@Draco18s What exactly should the meta question ask about? Whether teaming should be allowed, or what?

4:20 PM

The question is whether or not specifically allowing for team based submissions is too large of a change for the challenge as originally specified.

Got it

If I were to post I would post with "this change to the challenge has invalidated existing answers."

Okay

Here's what I've got so far:

I posted [this](https://codegolf.stackexchange.com/questions/189110/gold-battle-koth) challenge about a week ago, and it has since become fairly popular with nearly 40 answers. In the [chatroom](https://chat.stackexchange.com/rooms/96920/gold-battle-koth), a user asked whether teaming was allowed. I said it was, if someone could figure out how it could even work.

Two teaming bots were posted shortly after, one of which easily taking first place. Due to the rule that a bot may not target another specifically, it was difficult to top them. After lots of arguing in the chatroom, we're unable …

@RedwolfPrograms markdown doesn't work in multiline messages

Oh, ok

@Draco18s

Would y'all say this is a good representation of the issue to put on Meta?

5:08 PM

and then Persistent Blue Shell died

it's incredibly buggy right now, but still somehow managed to improve the previous bots score 44217 with KA in 2nd with 36243 to 56857 mine over KA's 40734 in 2nd place

full current stats
https://tio.run/##xVhNbxvJEb3nV/QxOWTQVf1R3ZcAVhJngcCIETnXAGOKazGmKIWk7EinIOcc9uBDfkSO@UX7R5xX1U0OJZILAT6sIPfOkkP2m1fvvarWbFx9Gjdfv7qzP@OnD1g/L1Yb9xv6iH/@o3M34z/6e7/8fLveXv@Vf@XIb6@dY1uDrdHWZGu2VWwttrqq/7mbr2fz1XaxnP/ixy///vHLP7/t978vfvcl1y/82m@5@Wf5/QFc/@tZnW8eLm63jNdTLkkG8lGvxWMtUrFmzlhZhLJepMSJay3Z66f1@0QS6s9eUPMk7CinmpyUwN6VRCyoeInF1RJ9gVwqc8HnnmL542K5dK82m8Vmi9ejlxAHINKdfcJasyiiqohIpBZ7K6VMFJlix4KXk6OUqkCQNWDNFTuHSBQgy5qzi1XEu8yRMhTJNcpTLO@u5@7VbHZ7v9qOK8XCoVYeUlBeXFAKqhBWyYqLQi6kWFyRmLlwpiEblpwiYFUuwQFjIEcx1uioFFwDeQgucBWsJWVwF4GrHmJ5Nx9vUBv3a/dqux1nHxULCckgRbEQ6/4h…

I just realized that teambots attack lowest to highest worth

because I forgot how sorting works

5:35 PM

@RedwolfPrograms that's great

Okay, good. Thanks!