Implemented Remove Bots from Ladder during Suspect Tests

Status
Not open for further replies.
quziel

quziel

I am the Scientist now
is a Site Content Manageris a Forum Moderatoris a Community Contributoris a Smogon Discord Contributoris a Metagame Resource Contributoris a Top CAP Contributoris a Contributor to Smogonis a Battle Simulator Staff Alumnus
Don't allow bots on suspect ladders.

Unsure what else to say, but including bots on suspect ladders, especially lower activity suspect ladders, really degrades game quality, and can allow folks to "cheese reqs" by trying to fish for the bot during low activity times. While this won't really get them to reqs itself, laddering at like 4 am America time to try to repeatedly get the bot, and thus get to a fairly safe 60-70% GXE isn't something we should really be allowing/incentivizing.

I'd say ban bots totally, but given how easy it is to create a new account, idt this is really feasible, and they do serve a purpose by populating low ladder. Just keep them off ladder during the actual suspect tests.

Thanks.

Edit:

Hypersonic-Alt had played 17,707 games as of May 2nd, 2022, and has played 43,843 games as of August 27th, 2022. This works out to an average of 223 games a day, or 6,924 games in a 31 day month.
 
Last edited:
I believe this would definitely help our issue of people not facing actual up-to-date and relevant teams because you can't tell me people who are testing our tier can get a 100% accurate POV in our suspect processes by facing this bot who's using very poor Pokemon at times (Arcanine+Doublade) + taking up the majority of the suspect experience lol.

edit: testing not casually playing
 
Bots gain ladder rating same way as players do. There is no real difference between playing a person who is just bad at mons and a bot who is just bad at mons. Bot games are not inherently easier regardless of rating. A game against a person at 1100 will tell you about as much about the metagame and challenge you as much as a game against a bot of the same rating. Or do you think you would get quality informative games on low ladder, just if the bot wasn't there? If you are worried about the reqs system itself (i am), think about the reqs system.

The only concern with bots is the influence on usage stats which is not what this thread is about.
 
Wingless said:
Bots gain ladder rating same way as players do. There is no real difference between playing a person who is just bad at mons and a bot who is just bad at mons. Bot games are not inherently easier regardless of rating. A game against a person at 1100 will tell you about as much about the metagame and challenge you as much as a game against a bot of the same rating. Or do you think you would get quality informative games on low ladder, just if the bot wasn't there? If you are worried about the reqs system itself (i am), think about the reqs system.

The only concern with bots is the influence on usage stats which is not what this thread is about.
Click to expand...
the guy I beat six times in one session stops laddering after an hour

the bot I beat six times in one session plays 200 other games that day

they are not the same
 
and what is the difference concerning you, your battles, your reqs? I don't get the point.

Is the ladder guy assumed to be underrated because they play less games? That's.. not very likely or.. there's no reason for us to assume that, right? They are at their rating for a reason unless it's someone on a new alt.
 
Wingless said:
and what is the difference concerning you, your battles, your reqs? I don't get the point.

Is the ladder guy assumed to be underrated because they play less games? That's.. not very likely or.. there's no reason for us to assume that, right? They are at their rating for a reason unless it's someone on a new alt.
Click to expand...
an exploitable AI that has no chance of winning just handing out points during suspect tests completely delegitimatizes the integrity of them. Besides suspect tests, by doing some basic math this bot accounts for about 15% of NU games played each month, that is an insane amount for one account and definitely not comparable to a real player.
 
Wingless said:
Bots gain ladder rating same way as players do. There is no real difference between playing a person who is just bad at mons and a bot who is just bad at mons. Bot games are not inherently easier regardless of rating. A game against a person at 1100 will tell you about as much about the metagame and challenge you as much as a game against a bot of the same rating. Or do you think you would get quality informative games on low ladder, just if the bot wasn't there? If you are worried about the reqs system itself (i am), think about the reqs system.

The only concern with bots is the influence on usage stats which is not what this thread is about.
Click to expand...
replying to this now because you make some good points, Roxiee's post derails quite a bit but the main issue is the bot is quite easy to queue into multiple times during a suspect run, as its always playing. I'd say maybe my past 2 or 3 suspect tests (Dragalge, Rotom-C, Machamp) were all fed around 5, but potentially more, wins out of about 30 total games, making a solid amount of my needed GXE and games required far easier than desired given there is a strict GXE and games played ratio requirement established to ensure players of higher skill levels are determining the results of these votes, as all other suspect tests do.
 
So the reason is that bots are bad at the game. Well yes, that's true.

But that doesn't matter, what does matter is how they are relative to the actual players at the same ranking which, well, i don't think the bot is inherently worse there? We are not talking about some auto-forfeiting script, but about something that also *gains* the rating from actual players. The terrible NU bot has 50,1% GXE, a better PU bot has 57,7% GXE, that's not just something that would freely give out wins, it's just, well, a 1100 elo game.

What i am saying is that if you substitute the low ladder bot with a low ladder player you will not get much more insight or information about the metagame. "Completely delegitimizes reqs" "allows people the cheese reqs", these claims just have no basis in reality. The bots make the experience slightly worse, one thing is AI quirks, but.. those vs bad plays, idk what's worse. The second thing and imo the only good argument for banning bots is that you will get more games against the same team. That's still.. not really enough? in my opinion..

The reqs process just... itself doesn't give you what you think the bots are taking from you? Bots or not, the process is nothing more than a useless obstacle.. And the skill level "test" comes either at the higher elo games or just the "don't get matchupped or haxed"

And the amount of games they play is just flat out not relevant here at all, dunno why everyone keeps bringing it up. Bring that up in the bots warping usage stats thread.
 
Wingless said:
So the reason is that bots are bad at the game. Well yes, that's true.

But that doesn't matter, what does matter is how they are relative to the actual players at the same ranking which, well, i don't think the bot is inherently worse there? We are not talking about some auto-forfeiting script, but about something that also *gains* the rating from actual players. The terrible NU bot has 50,1% GXE, a better PU bot has 57,7% GXE, that's not just something that would freely give out wins, it's just, well, a 1100 elo game.

What i am saying is that if you substitute the low ladder bot with a low ladder player you will not get much more insight or information about the metagame. "Completely delegitimizes reqs" "allows people the cheese reqs", these claims just have no basis in reality. The bots make the experience slightly worse, one thing is AI quirks, but.. those vs bad plays, idk what's worse. The second thing and imo the only good argument for banning bots is that you will get more games against the same team. That's still.. not really enough? in my opinion..

The reqs process just... itself doesn't give you what you think the bots are taking from you? Bots or not, the process is nothing more than a useless obstacle..

And the amount of games they play is just flat out not relevant here at all, dunno why everyone keeps bringing it up. Bring that up in the bots warping usage stats thread.
Click to expand...
Its true that the bot is basically just another bad player when comparing its ELO to others of similar ELO's but there's a factor here that actually does involve the number of games it plays, and that's activity. I work two jobs and get home at around 3 am some nights; nobody is only the ladder except that bot and sometimes one or two others, but you can very easily rack up multiple wins by simply loading into the bot over and over again and I have proof that this is possible because I've done this multiple times due to my restricted availability. As for a bad AI being comparable to a bad player... if you've played the bot on the NU ladder, you'd know that isn't true. This bot plays by a strict algorithm and can be feedback looped very easily; it often ends up PP stalling itself or preserving low HP Pokemon just to sac them all consecutively. I'd say a "bad" player is unpredictable, while the bot on the ladder is quite literally scripted in what it does and it only takes a game or two to learn how it works.
 
Setting aside suspect reqs legitimacy for moment, why are bots even allowed on the ladder in the first place? Is it just that it’s difficult to enforce? ‘Cuz I don’t know what useful purpose allowing them at all serves and it essentially opens the floor for blatant usage stat (and thus tiering) manipulation if someone were so inclined.
 
Honestly just the fact that they can skew usage stats - which can have an impact on tiers drops/rises - could be an argument to ban, or at least strongly limit bots, on non randomized ladders. By the way, this is also not the first time bots have been an issue - Leppa Berry was banned for a couple days in National Dex because a bot kept flooding the ladder with a team built to make games as long as possible, though eventually said bot was taken care of by the global staff, allowing the tier to unban this item. EDIT: holy, apparently this bot is back LOL

In all cases, the main problem with bots was mostly their ability to flood the ladder by themselves. I don't know if it is possible in practice, but i believe limiting the number of games bots can play at the same time, or over a given period of time, would eliminate the issue completely.
 
Last edited:
If bots were better they would impact usage more. I believe as it stands now, most of them are in elo hell with bad glicko scores so they're less impactful than a million games would seem.

Regardless of their impact on usage stats, and everything else outside the initial scope of the thread, I think that bots should be removed from 'special' ladders such as during suspects and ladder tours. They are generally safe wins for the parties of concern which makes them good opponents for purposes of avoiding decay or improving/maintaining GXE while increasing your games. The relative safety of fighting a bot on the ladder undermines the point of these ladders.

For what it is worth I will be asking hypersonic to avoid using his bot on the LC ladder during our ladder tour.
 
1661708892784.png
the good ending
 
Hi and good morning! Speaking on behalf of the PS Admin team as the ones who are generally in charge of carrying out requests like this one, our official policy stance is as follows:

If a bot is negatively affecting human experience, then we will remove it from the ladder via a permaban.

While it is impossible to completely ban bots due to the difficulty tracking them across the sim, if people feel that a bot is negatively impacting the integrity of a suspect test, then we can and will do the same during the suspect. You are more than welcome to reach out to myself or another admin to do so only if you are a Tier Leader of the tier the suspect is in and we will make arrangements.

Thanks!
 
Status
Not open for further replies.
Back
Top