Implement Lie Detector to Address Botting

I think most people can agree that botting is a problem. We have been relying on player reports and GM checks to catch botters before, which may not be very exhaustive and adds additional workload.

Lie Detector is a tool used by GMS to automate the botter-catching process:



How it works:

• When your character actively does something (either use skills or move), Lie Detector pops up at random times.
• The player needs to enter the phrase correctly to pass it.
• The frequency can be limited to once per hour or any other amount that avoids causing too much inconvenience to legit players.
• When a player enters a wrong phrase too many times (e.g. 3 times), or fails to respond within a time limit (typically 1 minute), he/she receives some form of penalty:
  • First-time penalty can be warping back to town.
  • Repetitive offences within a short period of time can lead to more serious penalties, such as autoban for 1 day / 3 days / longer

Pros:

• Helps catching botters, especially in unpopular maps, or the ones that use DC hacks / auto-logoff features when someone else comes. Ensures fairness of gameplay and helps to balance the economy.
  • Botters can do lots of damage to the market. We all know about gold tooth botters that kept gold tooth prices artificially low. People also suspected that botters injected lots of mesos to the economy, and contributed to the inflation after new source. One example is this farewell post.
• Reduces the burden on GMs and players to find, prompt, video-record, and report botters.

Cons:

• Legit player may find it inconvenient to answer lie detectors. To alleviate this issue, the following measures can potentially be implemented:
  • Disable lie detectors in bossing.
  • Lengthen the time limit of lie detectors to 2-3 minutes to accommodate summon farmers.
• Complicated image with lots of noise can be hard to implement and hard to get correct.
  • A simpler captcha can be used, such as plain numbers rotated. This is vulnerable to automatic recognition by OCR technology, but I doubt many botters would go through that much trouble to bot in MapleStory.

In the end this is just an idea. We can just skip it if botting is not as serious as I thought, or if there is a better solution, or if this solution doesn't achieve the desired effect.

 

Does this thing just pop up on the top right not intruding on the players view and button actions?

 

The thing is, it should block the player's views and actions but I suggest that there be some percentage of transparency so that the player is still able to see what he's doing. However, to avoid inconvenience as much as possible, I also suggest adding a notification somewhere that says an anti-botting detector will be up soon, say 20 seconds or so

 

Did this effect AFK people btw?

 

@Dann

 

If I remember correctly, old source used to have an auto ban system, I think thats better than a lie detector

 

Photon and I have a similar experience in that we both started on royals and dabbled a little in another server high on the gtop rankings so I can see where he is coming from.

I really like the idea and photon has listed out the considerations in its implementation. I think this is good advice as the lie detector implementation needs to be tested thoroughly before in order to avoid inconveniencing staff if it happens to ban more than it should. Nor would it be good if it didn't ban anyone. (It was rushed into implementation twice on that server and the result was a 10 page long backlash and it's currently disabled.)

 

If the autoban system was smart enough to tell botting actions from legit player's actions, with low false positive and false negative, then I guess it is the case.

I started in new source so I don't know how good the autoban system was, in terms of detecting botters. Designing such a system sounds very hard to me. I was under the impression that it mostly dealt with hacker doing unusually high damage.

Edit: for example, how do you tell this is a botter programmatically? Even for humans ,we are not completely sure until we prompt the player.

 

Only because that server chooses to autoban on the first failure of completion, and is too stubborn to change it, LOL.

But yeah, I agree with you that Lie Detectors need to be carefully tuned, if ever released.

IMO warping back to town is already enough to prevent much botting from happening. It is totally okay if no other automatic actions are taken, and players who fail a lot of lie detectors are just flagged for further GM investigation when they have free time. If botters continue to move at town, it's a lot easier for players to see and report them anyway

 

This is a great idea to reduce botting since majority of botters never check their screen. Just to add, I have a glimpse of memory of lie detector in old KMS but it was purchasable from grocery npcs and could be used on others. If the mechanism can be changed to only allow usage on characters who are "panting" after physical attack or use of skill, it might be used efficiently to get rid of botters even in the absence of staff member.

Original idea is definitely better since it'd cover everyone in the server who are hidden in unpopular maps but maybe adding manual method could increase the efficiency of bot detection. It would require limted usage on individual characters / certain time to avoid harassment or spam.

 

Both would be nice. No matter how good server's automatic detection system can develop, it's done by humans afterall and humans always find loopholes around it.

Problem with botting is macro botting and I've seen people trying to be smart by even ccing, climbing up and down the rope between actions at c-1 in magatia to escape autoban surveillance. Such actions will be difficult for server to detect but it'll be quite obvious when seen by players.

For the abuse of the item, can it be limited in usage per target character like only once/ day but not limited on the number of targets it can be used on. If someone is actually botting, that 1 report through item will be enough to alert the system or staff and won't need a 2nd reminder on the same day.

This item can be sold by FM npc at a relatively high price to avoid mass abuse as well. Let's say...2~3m? Not enough to hinder us from buying it but expensive enough to avoid mass purchase for no reason provided usage limit is applied.

 

I totally support this. I have seen this feature in another server where the gm BEFORE coming out of hide will prompt you with a npc maple administrator message where there are 19 NO options and one YES option. If you fail to respond that, the gm will come out of hide and try to talk to you. Perhaps if the prompt were properly designed, GMs won’t need to prompt players anymore

 

What about the little police light at the bottom of the screen? I don't think that has a use, and I think it's used to 'alert' a GM normally. Maybe it can be something to "initiate" the lie detector but only work if it passes some check by the server like Matt said.

 

^ an issue i could see with that is people pressing the button all the time for the fun of it and wasting GM times , but maybe can add a penalty to whoever "summoned" a GM if it was done as a time waster

 

That's why I said a check in the server

The GM's get notifications for all sorts of things, it should be possible for it to also funnel those same checks through this, maybe limit its use to once every few minutes if there's a worry about tying up server resources. Could be used for more than just botting too, if you see a player using a hack it could do the same thing.

For example: if they trip an alert and theres no GM online, if a passer-by notices the odd behavior they can initiate the "check". They won't see the notification that an alert was tripped, but players are pretty good at noticing illicit behavior.

 

Bumping this thread to remind the community and staff.

@Evan @Matt I want to ask if there has been further discussion on this topic.

As we all know, macro botting is the major source of income for RWT sellers. Hacking was and still is a way they use to make income but I believe it was significantly reduced by the server's auto detection system. Vote abusing is also a major problem but it must be way more tedious to re-raise those multiple account characters to 50+ once caught and banned. Therefore, macro botting comes down as the most critical issue in royals.

Whether they were speaking the truth or not, numerous RWT sellers, who were caught, in the past have emphasized the need to terminate botting if we wish to reduce RWT in the server because that's the pipe line for almost all sellers like the case of Chinese gold farming centers.

What are the limitations that may possibly delay the implementation of this system? I believe methods of check-up doesn't have to be a captcha screen but a simple mathematics equation like [10 x 5 =?] and require individuals to simply type in 50 and press enter. Coming up with bank of such equations won't be too time consuming.

We know it's not possible to completely eliminate the cheaters from this game since the door is open for them to return whenever they want. However, we can further reduce them by making this process more tedious for them. Every individual players may experience little inconvenience but I believe security should be prioritized for the sake of a healthier server.

 

if the end result of this a successful reduction of botters and RWT, i'm all for it.
i can care less about inconvenience and few pop ups with simple questions.

hope to see this type of system implemented soon :3

 

I'd like to thread lightly on this one. Let me produce a scenario which I (potentially) can forsee happen:

Step 1. Acquire many of the lie detectors.
Step 2. Find any random mage selling leech
Step 3. Harress the living day out of them by spamming the lie detector.
Step 4. Cause major grief and inconvenience to the player, as one mistake sends him back to town.

Then what

 

For the manual one