Tuesday, January 20, 2015

The Rabbit Hole: Monopolizing Resouces

This week, I'd like to continue discussing the five reasons why real money trading (RMT)  has negative effects on massively multi-player online games as spelled out in the initial post of the series. In his 2006 paper, "A Cost-Benefit Analysis of Real-Money Trade in the Products of Synthetic Economies", Dr Edward Castronova stated that, "RMT induces gold farmers to occupy territory and system resources."  However, I'd like to modernize the statement a little, to state that real money trading results in gold farming activity that monopolizes content and uses disproportionate amount of game server resources.

In order to compete in today's environment, many RMT operations are moving away from human gold farmers and relying on automated software solutions, more commonly referred to as bots.  But with some games vulnerable to such techniques as speed, flying, and teleport hacks, today's gold farmer no longer needs to post an avatar at resource nodes, waiting for the next spawn and discouraging actual players from trying to compete for the valuable item.

In the above half-minute clip, a fed up Wildstar player sat at a resource node spawn location and waited for bots to teleport to the spot and attempt to harvest the node.  The technique is one way for a gold farmer to monopolize the content without physically occupying the zone.  The bots go around in a pattern or a track looking for resources.  If they find them, the take the resources.  If not, they blink away to another location.

An elegant way to attempt to avoid the notice of regular players, although for those who can't get the resources they need to advance in crafting, an extremely frustrating experience.  If the bots are successful enough at escaping detection, players could even begin complaining about the spawn rates, thus causing issues for the customer support staff and taking up time from actual problems with the game.  But at this stage in the industry's history, players know to look for bots.  And in 2014, new releases were plagued with bots.

Perhaps the game that experienced a condition most similar to gold farmers taking over valuable territory in 2014 occurred in Elder Scrolls Online at launch.  In that game, bots literally took over public dungeons to the point players could not complete the content.

In the above video, a player gives a tour of the Wansalen dungeon that is overrun by bots.  While players can kill random NPCs, actually killing bosses looks almost impossible.

But sometimes players can figure out a way to beat the bots.  Below is a video showing the frustration of a player, even though she figures out a way to finally complete her achievement.

The reason for showing the third video is to show how RMT-related activity increases the use of server resources.  The user, in beating the bots in killing the NPC, had to spam commands to the server.  If the bots are also constantly spamming commands that don't do anything, then the servers are definitely getting hit with a lot of load.

But how much can inefficient bots really affect a game's infrastructure?  Back on 22 June 2009, CCP conducted an anti-RMT operation called Unholy Rage that banned 6,200 paying EVE Online accounts in a single day.  In an unusual move, the Icelandic game company released a lot of information about the operation, including the effects on the Tranquility shard, home to all of EVE's players outside of the People's Republic of China.

From CCP's Unholy Rage Developer's Blog

The company immediately noticed a massive drop in server load:
"Now, that is a beautiful graph if I ever saw one.  Check out that purple line representing average CPU Per User. This clearly shows the very disproportionate load the RMT type accounts imposed on our system.  While the number of accounts banned in the opening phase of the operation constituted around 2% of the total active registered accounts, the CPU per user usage was cut by a good 30%. That is a whole lot of CPU for the rest of you to play with, people."
The developer's blog also pointed out the effects on those systems with the highest concentrations of bots:
"Load issues for certain solar systems that were hubs for RMT activity have also dramatically altered.  Anyone who has travelled to Ingunn in the past can testify that this particular solar system was well clogged up with haulers and shuttles running missions pretty much 23/7.  Here's a little graph showing the population in Ingunn before and after June 22nd.
From CCP's Unholy Rage Developer's Blog
"Clearly, the state of Ingunn has improved so dramatically that one can hardly recognize it as the same system.  The same is true for many other systems previously chock full of macro miners and mission farmers."
So real money trading can cause behavior, including hacking and botting, designed to maximize the amount of virtual currency gathered for sale.  The gold farming activity can affect the play of players though the monopolization of content, like resource nodes or even valuable NPCs that spawn in the open world that are required to progress through the game.  Also, poorly written bots, as the data from CCP demonstrates, can indirectly affect the game experience of players by using an inordinate amount of server resources.  For game companies, reducing this type of negative behavior is beneficial both for the smooth running of their virtual worlds as well as keeping their customers happy.


  1. Again, I must point out that RMT does not cause behavior in a vacuum. Rather, RMT in combination with the game design and implementation creates incentives to act in certain ways.

    Why does this matter? It matters because game developers (a) do control the game design and implementation, absolutely, whereas (b) they do not control the desire in human beings for real-world money, nor do they control the actions of their players in any substantial way.

    So there is a tradeoff. There is nothing about players or bots "harvesting resources" that is wrong per se. It's only a problem when the game design is such that this causes problems for some reason. I feel that fixing the game design is often overlooked as a solution, because game design and implementation are hard whereas banning people is easy. But this too has its costs.

    Just to be clear, there are many game designs that can work well with human players but not with bots. (I.e: as described above, a valuable resource that is always in the exact same place, ungated, and first player to click is winner and gets all.) These may be very hard to remove from the game. Then, the tradeoff is probably best solved by banning bots. But you never have to design a game using such techniques.

  2. Less rudimentary bots don't care about fixed locations, but will read the position of a resource node/mob/other directly out of memory.

    Only CAPTCHA gives the current generation of bots issues on an accuracy basis - because the "answer" doesn't exists the client side memory.
    But if there's money to be made, this too could quickly become untrue - my guess.
    One case for Streaming games is that you can't read vital information from memory when all you get is an image.
    You'd have to use image interpretation, which humans are still superior at.
    My point here being that bots are hard to beat with simple design changes.
    Borrowing a term from Ramin Shokrizade (sp): "Industrialized RMT farming" is the real scourge - hundreds, or thousands of bots utilizing state of the art technology to farm gold/etc.
    They're investing in it and your design won't keep up.

  3. You can make a case that game design helps lead to RMT, or at least makes the practice easier than it should be., I will disagree with you about the bots. Unless a game is specifically designed with bots in mind, bots shouldn't be allowed to operate in games.

  4. Von Keigei,

    Nicely nuanced review of your standard stump speech as it clearly clarifies your ‘problem definition’ approach to the subject. If I may elaborate a little . . .

    1) Nosy has been kind enough to signpost his approach to the ‘Rabbit Hole’ series. As he explains, he’ll be exploring “five reasons why real money trading (RMT) has negative effects on massively multi-player online games.” I know it’s tempting to simply waive away the whole series right from the get go as you believe you already have ‘The Best Answer’ in your pocket but for the rest of us who remain interested in Nosy’s exploration that offhand waive away gets a little old.

    2) If I may ask, if your ‘game design and implementation’ approach is both evident and implementable why hasn’t it already been done? While I’ll happily agree with ‘evident’, I, much like Thomas in this thread, have serious doubts about ‘implementability.’ Creating a successful MMO that both supports human avarice (no avarice, no desire, no desire, no economy) while simultaneously stunting the bad sides of that very same avarice (farming, botting, cheating . . .) is no easy matter. We have an entire real life human history of struggling with just these matters and as best I can tell no society has yet managed to find a perfect (or even terribly good) solution. Amazingly, MMOs struggle with surprisingly similar conundrums.

    3) I’m not saying you’re wrong Von Keigei nor am I saying your comments are unhelpful but I do find your approach a touch cavalier. ‘Better design and implementation’ is easy to toss out in the abstract. Luckily, we’re in the fortunate situation as Nosy plans an entire series on just these matters which gives us opportunity to further explore in much greater, wonderfully concrete, detail.

  5. Actually I agree with you about bots -- they should be banned in many or most games.

    I think where we may differ, and it's not a huge difference, is in dealing with them. My attitude is that just by saying "you should not do this", you don't solve a problem. It's like saying "don't take money from the till". Neither do you solve the problem by making it illegal, although this does at least deter. You solve it by making it impossible to take money from the till. Or, if that's not feasible, by making it very hard to take money from the till without being observed. So you should design the game to make undesired behavior be very hard to profit from.

    Also, I'd add that there is a second way to deal with bots, which is to build automation into your game. I.e., in EVE the autopilot.