Some data on Malton population has been collected on a more-or-less weekly basis (on Monday) since the inception of the statistics page, but daily numbers have been saved only since the 2nd of January. The numbers are checked in the morning, PST.
Zombie numbers peaked in the week following the sack of Giddings Mall on 1 Nov and then began a long downturn during the First Siege of Caiger Mall, finally bottoming out just before the XP/AP change to Headshot (and the subsequent player strike) while humans peaked at the same time.
Then, Feeding Groan was implemented on 22 Dec and the trend almost immediately reversed. The subsequent large building code change a week later does not appear to have been as significant, however, as the rate of survivor loss was already slowing. Despite the massacres being perpetrated by the Mall Tour horde, numbers were stabilizing in mid-January. While "zombie" numbers were way up by this time, much of that was simply dead survivors standing around designated revive points for days at a time, begging for revivification.
Just as the population curves had flattened and stabilized, NecroNet Access was introduced on 19 Jan, and with it, manufacturing of revival syringes. A large number of players who previously eschewed searching for random syringe drops began actively making needles and putting them to use at the queued-up revive points, and the population began to shift heavily back in the direction of the survivors (many of whom were active zombies revived involuntarily).
With the introduction of increased AP cost to use those syringes, however, the trend once again fishtailed. Survivor numbers were down to an all-time low and zombie numbers were up (though not to their all-time high). Since the recent change in find rates within powered buildings, however, the population has shifted back in the direction of survivors once again. We're not just seeing zombies being revived, however; the rate of zombie loss is greater than the rate of human recovery.
A very simple formula produces a fairly clear curve for charting the general "health" of the game, in terms of balance of power. (A calculated trendline has been added as well, more as a clarifier than as a predictor.) Additional analysis of day-to-day change provides additional insight.
The total number of survivors plus revived corpses, less the number of active zombies and killed corpses, clearly shows the slow of zombie victory in January; then, the introduction of syringe manufacture on 19 Jan seals the deal. Interestingly, extrapolating old data from the current data using the new way corpses are reported, it appears that in January, those who were surviving were in fact outnumbered by those who were dead by as many as 4000. By the time we reached March, the living and revived were outnumbering the dead and killed by almost 12,000; in the space of a single month, the game experienced something like 16,000 net revivifications.
The chart on the right shows daily change in human and zombie population, in absolute numbers-- again, with trendlines to somewhat clarify overall tendencies. When the trendlines are widely divergent, one side or the other is probably experiencing too much success and the population balance is moving rapidly one way or the other.
The "live gap" was already decreasing at a slow but steady rate well before the revive nerf. It is not yet clear what the causes were, except in the overall sense that it was clear that zombie numbers were rising, more than survivor numbers were falling. The turnaround began even before the RRF (a very large zombie horde) began its recent mobile suburb destruction event. At any rate, the RRF excursion has been very localized at any given time and wasn't affecting revival patterns on its own, so it cannot have been the sole factor in the trend. It has been suggested that we were seeing large numbers of Caiger Mall survivor "zergs" idling out now that the Second Caiger Siege is over, but that would result in a significant fall in survivor numbers (and possibly zombie numbers as well), not in a slight decrease accompanied by a zombie growth. (I do not see any evidence in the population trends to support an assertion that there were very large numbers of zergs in use at Caiger, on either side.)
Regardless of the cause of the original decrease in the gap, the syringe nerf certainly accelerated the process for a while. Then it slowed to a point of balance in late April. Then it began a slide back in the direction of survivors right as powered building searches were introduced. It remains to be seen if we will reach a new stability point soon.
The activity dip in late December stems from players going on holiday. The activity spike around 9 Feb coincides exactly with the announcement of the Mall Tour finishing their rampage in Darvall Heights and the beginning of the Second Siege of Caiger. The subsequent drop-off in activity is very reminiscent of the same decrease in activity that marked the First Siege of Caiger. It would appear that the Caiger wars drive some players away. Once Caiger II broke up, there was some recovery. However, over the last few weeks since the revive nerf, there has been another extended decrease in activity, which appears to be continuing. We are now at an all-time low level of activity and continuing to lose players.
Looking at the daily growth rates, we can see the quick rise to peak popularity last fall, then a drop off after the removal of the ability to choose your starting neighborhood. The big spike that suddenly appears at the start of November was caused by someone unleashing a character creation bot for two extended periods in a 48-hour window; to what end is unclear. Finally, in late February there was a big drop-off in creations and we've been on a slow deflation ever since.
I had, until quite recently, thought the decline in player activity was mostly the result of this lower character creation rate, but recognized that there might be an elevated rate of player departure contributing as well. sunnyside, on the desensitised forums, pointed out that some math could give us some sense of whether there really is a player exodus underway. The result surprised me: for the past six weeks, the rate of daily idle-out has not only shifted to a lower level, but it also become a lot more reliably bound. At the start of the year, we were seeing wide swings from 1500 to 2500 per day, but since about 2/26 it is varying between only 1000 and 1500 per day, and in the last few weeks since the syringe nerf there has been no noticeable difference!
Some more detailed views into the same daily data. The first graph shows relative change in size of the three kinds of survivors, expressed as a percentage of their own total type. The second shows the same data for humans, but also adds zombies, shown in stacks. The third shows the change in all humans (lumped together as a percentage of total humans), all zombies, and (additionally) all corpses (of both types).
The anomalously-high spike on 27 Feb appears to have been the result of a one-day rush of character creation, nearly 600 worth. I suspect that UD got mentioned in the press somewhere very high-profile and we saw a bunch of daytrippers trying out the game for a day before dropping it.
These are some graphs that I put together looking for patterns involving number of corpses. It is not clear that any of them are particularly instructive, but they each have their interesting traits. The first two start at 2/9 because reporting of corpse numbers and hunters changed dramatically at that point, and it's very hard to reliably correlate numbers from before with numbers after.
The first graph shows what percentage of all standing survivors are hunters, and what percentage of all corpses laying in the street are revived instead of killed. Both were about a third at the start of February and, until quite recently, both were climbing. Since then, they have diverged: Revives were already falling off and now the recent syringe nerf has really decreased them, while Hunters were rapidly growing as a percentage of the total survivor population. In the last week, however, the hunter percentage has leveled off and revives have made a slight comeback.
The second shows something similar: the total absolute number of active hunters, and the total number of bodies (both revived and killed) on the ground. They were practically the same value (about 9200 of each) at that point, after which they widely diverged. Absolute numbers of hunters has come way down from its all-time high, however, and the two stats are much closer once again.
The third tracks the percentage change in number of active characters from day to day parallel to the percentage change in number of corpses of both types. Again, I don't think there's an actual correlation to necessarily be drawn here, but I find it interesting that the changes match very closely in a few places.
The following three graphs show the relative representation of each of the three survivor types in the total survivor population. Note that the vertical scale in each case is exaggerated; you're only looking at changes of a couple percent at most.
Military characters had their strongest peak between the intro of Feeding Groan and NecroNet, with a noticeable resurgence right at the time when the Mall Tour hit Caiger in early February.
Scientists were at their all-time largest representation back when stats were first implemented, with a long but steady drop-off over the next two months. Then, they immediately resurge again when NecroNet is introduced, and start falling off thereafter.
Civilians (which, it must be remembered, also include original zombies who are revived) slowly rose in representation during the First Siege of Caiger, and fell off after that. They began rising again with the introduction of NecroNet and syringe manufacture, as more and more zombies were being combat-revived back into human status. With increased syringe use costs, random drive-by revives became rarer and there was a drop off.
Should require little explanation. I have only been keeping phone coverage stats since early March.
Note that there is a great deal of variability from day to day, but there have definitely been two noticeable overalls drop in the past six weeks, one right after the syringe nerf, the other about ten days later.
The active population of Malton as distributed across levels. Note the spikes that represent maxed-out humans with no zombie skills and fully maxed characters, plus the somewhat softer "lump" of more-or-less maxed-out zombies with some human skills (such as Bodybuilding). The graph on the left is current (as of the listed date); the graph on the right is the original distribution at the inception of statistics on 27 Oct 2005, for comparison. If you open both images and toggle between them, you can see a definitely "sliding" effect as the body of low-level characters become higher-level ones. Note the loss of sharpness around the highest-level characters in the past month.
The average level of all active members of each character class.
Note that after an extended period of being the least-advancing group, zombie levels climbed rapidly after Headshot was changed and Feeding Groan was introduced. This was, in some large part, the result of many high-level humans being killed, not just zombies quickly gaining experience. With the introduction of NecroNet syringe manufacture and the clearing out of the revive queues, many high level "zombies" went back to being survivors and there was a brief dip in average zombie level again before returning to everyone advancing as before.
This is clearer in the second graph, which shows a rough approximation of how much change there was in the average character level each day, contrasting zombies specifically with all characters put together (i.e. zombies and humans both). When the zombie line is higher, zombies are leveling faster than survivors, and vice versa.
The following graphs do not separate anything out by character type; that is, they do not currently distinguish zombies from survivors, or (say) military from civilian. This is mostly because it would be four times as much work to graph all the character types and look for the ones that show the most interesting results (or each graph would be four times as cluttered, which would be virtually unreadable). They are not intended so much as evidence that something should or should not be done about the population balance; they are more interesting as indicators of how characters are behaving overall, and what they tend to do.
The first trend I looked at was the change in the number of lowest-level characters. It is clear from the population across all levels graph that there are many fewer bottom-level characters than there used to be. In graphing the actual week-to-week changes, however, some interesting things emerge.
In the graph on the left, we see the absolute number of characters at each of the lowest six levels, since the inception of stats. The main thing to note here is that levels 1 and 2 fluctuate much more than levels 3 through 6. This isn't particularly surprising; we know that the character database grows by an average of 1500 new toons every single day, and that every night at rollover it idles out a similar number. It is hard to estimate how much of this comes from first-time players who drop the game after just a little bit of use, and how much of it is from existing players making new alts. Regardless, it appears that only about half of all 1st level characters make it to 2nd, and half of all 2nds make it to 3rd. Once a character gets to 3rd, however, their numbers are much less volatile.
How much less volatile? Well, in the first chart, it looks very close to flat, but if we exaggerate vertical scale, we can see some points of difference. My original interest in drawing this data out was to explore the question of whether the NecroNet skill had created the rise of so-called "syringe alts", that is, new NT characters who existed solely to get the three NT skills and then only manufacture and use syringes thereafter. These would be 3rd-level characters. So, in the graph above right, we see a graph of 3rd-level activity. And indeed, starting at 1/19 and the introduction of NecroNet, we do see a noticeable peak in 3rd level characters. However, note two other things:
So let's look at the question of whether those theoretical syringe alts might have continued to level up.
Here we see levels 3 through 6 with exaggerated vertical scale, to make them more distinct. The bump in level-3 characters after 1/19 is very apparent, but as it goes down, there isn't a corresponding rise of the same size in levels 4 and 5. This suggests a die-off of those 3rd-level characters. Looking back at the 3rd-level graph again, I think we can see exactly when it happened: there was a rise of almost exactly 300 from 1/19 to 2/2, then a slight decrease, then a few steady days, and then a rapid fall-off of almost exactly 300 characters after 2/16. Compared to the timing of the end of the Second Siege of Caiger, perhaps a number of them really were being run out of the NTs around the mall. We'll never know for sure; it may also have been a very distributed phenomenon, and they began idling out as the syringe glut made dedicated NT work entirely redundant.
Having expanded out these less-populated levels, we can also see that they are not flat lines, but are in fact all trending downwards. Does this mean characters are leaving? Well, it does for some-- there are only half as many level 6 characters as there are level 3-- but mostly, characters who reach 3rd level continue to level up. The second graph shows the continuation of this progress from levels 7 through 13, all of them having slight ups and downs, but trending down in the big picture. Some quit, but where are the rest going?
Maxing Yourself Out
Here's where things get interesting to look at again. There are three big collection points, where some large number of characters stop advancing. The first two are the highest possible zombie level, and the highest possible survivor level. Each one halts a few thousand characters, who are content with being the best of their type, but uninterested in picking up much (if anything) from the other side.
Six months ago, a maxed out zombie was level 14, 15 with Bodybuilding (or 13 without Brain Rot). However, as each new zombie skill is introduced, there is a sudden decrease in the previous max-zombie level, and a corresponding increase in the new one. Currently, a maxed-out pure zombie is level 16. With Bodybuilding, that's level 17. Note also a rise in level 18, which tracks level 17 very closely. I believe those are level 17 zombies who are currently infected. At any rate, we can see that there is a clustering-point around which some number of zombies accumulate, and it moves up over time as the number of zombie skills increases. With each shift, the new group is slightly smaller than the last.
On the right, we see the same thing for maxed-out humans, and here it couldn't be clearer: In the last six months, maximum pure-survivor level has gone from 19 to 21, and each time there has been an enormous and immediate shift in the survivor population. Note that the vast majority of maxed-out survivors are clearly not also in possession of Lurching Gait and Ankle Grab (currently represented by levels 22 and 23), but it is a steadily growing group.
The Upper Class
Once you get past the maxed level of your character type of choice, it's a consistent rise to the top.
On the left, we see the ever-growing number of characters making their run for the highest levels. Note that the difference between the largest and smallest group of these characters are separated by only a couple hundred. (I compressed the color distinction between each level because, for the purposes of this graph, they are all pretty much the same.)
On the right, we see their destination. The highest possible character level currently is 37, but level 36 is the far larger group: everything except Brain Rot. Unlike the zombie cluster, however, this is a group that only gets bigger each time it shifts. This is the end of the road: our growing upper class. There are now more characters with almost all the skills, than there are maxed-out pure humans or maxed-out pure zombies, and this will only continue to be more true. Pure survivor players who, in the past, looked down on and PK'd survivors with zombie skills are now outnumbered.
To get a sense of how much this top-most bracket has grown over time, let's stack all the highest levels on top of each other, to get a cumulative count, instead of having them overlap:
From 250 to 4750, a growth factor of almost 20x, in five months. (Plus, the growing number of top characters with the revived and/or infect flags also set, which is shown in the second graph.)
Changes from Week to Week
We can also look at this data from the point of view of how much change there is from week to week (starting from mid December, which is when I began reliably getting data every week). There's a few things we can learn that aren't necessarily apparent looking at the total numbers. First, let's look at those low level characters again, tracking how much percentage change they've experienced each week:
The drop-off of 1st level characters after the December holiday is sharply apparent. The graph on the left is stacked bars, the one on the right is overlapping. Almost 20% of the 1st level character set left for Christmas; some of them came back and leveled up, as we see the positive peak of level 2, but many did not. We can also see a 7% increase in 3rd level characters in the week following the Necronet syringe skill. We can also see the effect of NecroNet on highest-level pure humans:
This graph is in absolute numbers of characters: almost 800 people left level 20 and joined level 21 at the same time. We can see a slightly less distinct example of the same thing for zombies:
When Groan is introduced in late December, level 14 has a negative spike, and levels 15 through 17 all get positive spikes. Then Tangling Grasp does the same thing, one level higher. However, because zombies are somewhat more distributed around their "maximum" point (because some also have Bodybuilding, and some have chosen not to take Brain Rot), the spikes are much less distinct. Zombies seem to show more variety in their make-up than survivors.
Sometimes, looking at relative graphs (i.e. by percentage) sends one message, while absolute numbers send another. It's usually a good idea to look at the data both ways.
Here we see the week-to-week change in the top-most character levels. Viewed as a percentage change, the largest spikes are at level 36 in late December (Feeding Groan) and lvl 38 in mid February (Tangling Grasp). However, these were not the actual largest changes in the population; if we look at absolute numbers, we see that level 35 had the biggest growth, after 1/19. These are, of course, top survivors without Brain Rot who picked up NecroNet. The place where level 36 has its biggest actual growth is in February, not in December, as a bunch of characters-- again, having everything except Brain Rot-- picked up Tangling Grasp. But this is distributed over a few weeks, not the sharp delineation of 1/19, suggesting that most top-level characters who prefer to play as survivors didn't rush to get killed so they could take the new skill.