Legitimate player VS cheater VS bot: a data-science question?

Hi all,

This post is not here to offence in any way, more to raise questions.
Not being from the video game industry, I may also underestimating the complexity of the problem, so my apologies in advance if that’s the case.
Also english is not my native language, please take it into account when reading this.

First of all thanks to the NW team.
We all know that you are working hard on improving the game (there is still a lot to do),
but at least it seems to move in the right direction ([Dev Blog] Update on Current Issues).

Exploiting has always (and will always) be an issue in MMOs.
To cite one dev from a very famous one:

Gold farming and botting, the bane of any MMO economy.

For a majority of us gamers, exploits raise the following questions:

  • Why should I invest time in something, if someone else can cheat and do it in a much less time?
  • How can I get the fruit of my hard labour, if economy is biased?

In New World, this is even more critical since economy plays a key role and user interaction.

Nowadays we are surrounded with big data, everywhere.
ML/AI algorithms gained a huge popularity as tools to analyze this data.
And today I still don’t understand why identifying a true legitimate player from a cheater or bot in this industry is still so problematic.
You seem to have logging in place ([Notice] Temporarily Disabling All Forms of Wealth Transfers - #7 by Lane), so take pro-active actions to counter that!

Take a look a the following suggestions.
I will obviously not go into the details (if needed PM me for more details):

  1. Bots and gold buyers

    Problem: Devaluates the money because (some) users have high income when they should not have. Ruin the market for crafters and gatherers.

    Proposition: Robots should have clear frequency patterns that could be identified through time-series analysis. Gold buyer should be easily detected by tracking the logs.

  2. Dupes (items and gold)

    Problem: It disrupts the market with a high influx of income/objects.

    Proposition: This can also be detected with the logs. A big issue is to be able to trace back the origin if it was introduced in the market, look at blockchain analysis. Using graph-based theory, there are way to trace back the origin of any item. If dupe gold/item is recent (<2 days), alleviate the impact in the economy by reverting all underlying exchanges.

  3. True owner of a mule/alt

    Problem: Cheaters often escape the sentence by using mules. I would not be surprised that many of them are still playing now (even after the ban wave).

    Proposition: Combination of graph analysis (model user interactions) and signals should also help here.

  4. Aim bots and other combat cheaters

    Problem: Huge advantage in combat.

    Proposition*: murder/death and hit/miss ratio, coupled with frequency analysis.

So my question is quite simple, do you have data scientists who works specifically on that subject?
Also I would be really curious to hear what the community thinks about it, and even better if suggestions (also from other data scientists)! :slight_smile:


PS: No deep learning, traditionnal ML or even basic feature processing is more than sufficient here I think.
PS2: I was thinking that MMOs economy reflects well IRL, people escaping sanctions and not following rules. But here we are in a digital world, the entire “context” and rules can be controlled (Big Borther = l o g g i n g).

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.