Post-AvA Rogue Phase Analysis

A week has passed since we threw a little caution to the wind and embarked on a Rogue journey together. We can’t say thank you enough to those who took a chance and joined in, and we also appreciate the quality and quantity of feedback we received. Whether positive or negative, we learned a lot with and from you in the Rogue phase. This post attempts to highlight some of what we learned along the way.


Levels of Participation and Action

We were pleased to see that a total of 705 Alliances took part in the experimental Rogue Phase, approximately 80-90% of the typical participation rate.  

Rogue Partipation for post-ava blog.png

Within Rogue, we were not surprised to see that the unique changes we made to actions for the phase resulted in their levels of use looking different from other AvAs.  

When these actions are consolidated and graphed by hour, the strong initial engagement only dropped off as night fell in North America, coming back the following morning and steadily increasing until the end of the phase.

And, when the consolidated actions of each of the regional Alliance winners is similarly graphed, you really get a sense of the different Rogue experiences that each Alliance must have faced.

Just as unique were the amounts of VP earned in Rogue from Alliance to Alliance. Unlike regular AvAs, the Rogue regional leaderboards were far tighter. Check out this interesting graphic that was created by one of our data scientists. No ‘megas’ running away with the phase this time! But, as we discuss further down this blog, the Rogue leaderboards had their own problems, not visible in a static graphic.


Passive to Active VP Imbalance

It wasn’t long into Rogue, when it became clear that an imbalance of Passive to Active VP would prove an issue for many Alliances. Passive VP quickly dwarfed Active VP, leading to all sorts of unforeseen and negative consequences, such as ‘active’ Alliances scoring below their more ‘passive’ counterparts on the leaderboard (a result of becoming targets and getting attacked hard, as is described below).

The cause for this imbalance was a result of the difficulty to predict how these values would play out in the actual phase and our inability to do any testing ahead of the real-world testing that was Rogue. So, in our efforts to not set passive VP too low, we set it too high.

The lesson for us here is that were we to run a phase like this again, we would work to improve the VP balance. We would lower Passive VP, and look for a way to mitigate against any unwanted impact on Threshold Rewards by an increase in Active VP. We would also work to create better modeling tools to test the balance theoretically before the phase began.


Leaderboard Fluctuation

Another common challenge for many players, hinted above, were the fluctuations of the Leaderboard results. Alliances would battle hard to make it to the top of the leaderboard only to see their standing drop precipitously on the next update of the passive VP timer. Top Alliances would drop dramatically, and lower alliances would rise. Worse still, once an Alliance had dropped down the leaderboard, because of the constraints on healing, it was difficult to rise again.

Let’s unpack this a bit.  First off, like in regular AvAs, Alliances at the top of the leaderboard were disproportionately attacked. No surprise there, though we believe many players might have erroneously believed that they would earn more VP the higher up the leaderboard they attacked. For the record, this wasn’t the case for Rogue (yes, perhaps we should have emphasized that more clearly ahead of the phase).

But why would top-ranked Alliances – even if attacked disproportionately – then drop so dramatically in their rank? The answer primarily lies the passive VP timer, the timer that dictates when the Leaderboard is updated. This timer is currently designed to update only once per hour. As a result, those at the top of the Leaderboard would incur damage throughout that entire hour, since they would appear to remain a top-ranked target throughout.  So by the time the VP Timer updated, the top alliances would have received an hour’s worth of heavy damage and would then drop far down the board. Moreover, once fallen, these ‘wounded’ Alliances found it difficult to heal themselves, since silver contributions were not allowed.

If the Passive VP Timer had cycled more frequently per hour, top Alliances would still have dropped, but not as far, since the amount of damage they would have taken since the last timer update would have been less. That is, a more frequent VP timer would have allowed the leaderboard to more accurately reflect the actual standing of each Alliance and not held them artificially at the top of the board to be pummeled.

We are currently looking at the technical feasibility of more frequent Passive VP timer updates and are even discussing its potential value in regular AvAs.


Regional Diversity

One aspect of Rogue that we thought was refreshing was the diversity of the regional leaderboards. No Alliance won more than a single region, and no AvA ever had so many first-time winners. It was the implementation of flat GP, of course, that lead to this outcome (and lowered overall server demand, allowing for one of the smoothest AvAs in terms of performance).  Nevertheless, we will not be implementing flat GP in regular AvAs, as we are exploring other, less drastic ways to achieve these positive outcomes.


Looking Forward

The Rogue phase was intended to be a unique AvA experience from which we might play with the mechanics and take away lessons. On that score, it delivered as promised and while a number of players might have found it wanting, many others found the challenge of Rogue fun and refreshing.  Either way, we were thrilled by the intensity, intelligence and constructiveness of so much of your feedback and continue to digest it as we look for ways to make AvA more fun and fair for all.