I’m not sure how much point there is studying the end of the world given that, should the end of the world come about, all bets will be off. But such thoughts of futility do not deter academics, and their latest wheeze is to have a look at what happens in an MMO game that is going to shut off: A virtual end of the world affair.

The game in question is the Korean MMO ArcheAge and players were studied during a beta test that was scheduled to last 11 weeks. Thus the study had a decent time range and could look at a whole host of (anonymised) player data over this period. Players knew that the beta would end at the point it did, and so the academics reckoned that, with a whole range of caveats, it might give us some idea of what will happen when the asteroid is about to hit, or Musk accidentally turns on Skynet.

The paper (spotted by IFLScience) has the not-very-heartening title “I Would Not Plant Apple Trees If the World Will Be Wiped: Analyzing Hundreds of Millions of Behavioral Records of Players During an MMORPG Beta Test”. An important note is that it has been published but not peer-reviewed, so think twice before planning your doomsday strategy around the findings.

The abstract reads, in part:

In this work, we use player behavior during the closed beta test of the MMORPG ArcheAge as a proxy for an extreme situation […] We analyzed 270 million records of player behavior [and] Our findings show that there are no apparent pandemic behavior changes, but some outliers were more likely to exhibit anti-social behavior (e.g., player killing).

So the top-level finding is: Most of us are fine, but a handful are going to go full berserker mode. The paper’s title also plays on the adage “Even if I knew the world would go to pieces tomorrow, I would still plant my apple tree,” because it turns out most MMO players don’t, abandoning character progression, showing that “players abandoned character progression, showing a drastic decrease in quest completion, leveling, and ability changes”.

Well that’s all cheery enough. “But it’s just a game,” jeers the crowd, and of course it is, but studies like this rely on the mapping principle, which boils down to the behaviour of players in online games not being all-that-far from the behaviour humans exhibit in real life. So no academic would ever claim this kind of thing is a 1:1, only that it is possible to glean some behavioural insight from it.

We’ll get to the red meat of player killing in a moment, but there were positives about the end times in ArcheAge. Chat content (measured by a valance score, essentially analysing for ‘positive’ words and sentiments) hit a slightly positive trend as the world’s demise grew closer, and players increase their social interactions: Sending more in-game messages, and creating more parties for group activities. 

One significant difference was between players that voluntarily leave the game (who the academics refer to as “churners”) and those who stay until the end of the beta. It’s the churners that are way more likely to show anti-social behaviour nearer the end (including, yes, player killing), while in contrast those who stick around “continue to behave within accepted social norms”.

So we come to the juicy murderising. There’s an interesting arc here, where the researchers found that murders were more common at the beginning of the beta, before decreasing until roughly the last third of the timeline, when they suddenly spike again. The paper theorises that the earlier peak is due both to players trying out PvP, and the age-old habit of experienced MMO players griefing newbies.

“The increasing trend at the end of the timeline is an indication that players might be reverting to more ‘savage’ tendencies as well,” reads the paper. “As we expected, players are more likely to perform anti-social behavior when no penalty will be imposed”.

The paper notes there were 334 main murderers at the end of days, “relatively few”, and then examines how these outliers behaved throughout the timeline. Were they always bad, in other words, or was it the impending shutdown that made them go psycho killer?

The researchers use an algorithm to cluster players based on four aspects of their behaviour over time, resulting in Fig 4: “a histogram of the murders performed per cluster divided into three intervals (early-, mid- and end-times)”.

“Interestingly, we see some differences in when the different clusters performed their murders,” note our scientific heroes. One cluster, naughty old cluster 4, was killing people basically all the time. Cluster 2, on the other hand, exhibited a dramatic increase in murderous behaviour as the end came closer. I can’t really beat the two-part academic conclusion here so present it in all its glory:

• 1) not all murderers are alike, but there do seem to be some archetypes they can be clustered under, and
• 2) clearly there are some players who, although they did not quite go from pacifists to serial killers, did in fact show an increase in murderous tendencies as the end of the world drew near.

So pour yourself a stiff one because, as you hug your babies close and the sun flares for one final time, that nice old lady next door may well walk up and shank you for the hell of it.

Perhaps the more interesting element to tease out, however, and obviously this is a study of an MMO, is the difference between players who showed some attachment to the ‘world’ and those churners who could take it or leave it. The difference in behaviour is quite profound, and suggests those who feel some sort of connection or loyalty are overall more peaceful and better behaved than those who lack a similar attachment: The ones that go on the killing sprees.

It’s little surprise that, when the ‘world’ is ending, players don’t want to spend time improving their long-term stats or completing quests. Will you really be rushing to the gym as the Four Horsemen sally forth?