1

State of the Game #260: Looking at the Data

StateofTheGameHeader_Ice

Welcome back for this weeks State of the Game! While I have been very hard at work on the next update for MAV, I wanted to take a look back at the January update to see how it went.

So, part of why I like to track data and look at data is because I not only use it to help make decisions, but also validate the choices that I do make. Not every choice can be 100% grounded in data, so it’s always good to validate after the fact to ensure your ‘gut’ is still calibrated correctly. This was the case with leaving Steam as a platform [Which is working out fantastically I might add!] and for sure a big change like the January update, this is a good use case.

So to recap, we have 3 major additions in the update, the inclusion of the No Man’s Pass level, the Skynet AI update, and the large balance update. I want to first look at these in total, then breakdown each addition itself.

Overall Update

Overall the release of the update saw a 300% rise in daily active players, generating a new all time high of active players in MAV. So far January is set to increase the number of active players by 30% over the December numbers. This is good news as it means I am doing something right and getting more and more people not only joining MAV but also getting into the game and playing. As we all know, MAV fun increases the more people you have to play with! In addition to this, players are tending to play MAV about 10% longer in January then they ever have before. This is helpful as it helps increase the number of ‘collisions’ between players online, increasing the chances they will play together online.

No Man’s Pass

This level was a long time in the making and it was always a disappointment to me that it was not finished for the Beta launch. Looking at the numbers, it is currently the 4th most popular level. Part of this is just the ‘new’ factor of the level and the cult like following it had. I will keep monitoring this metric to see how well it performs overall. It was primarily designed as a siege style map and as that game type becomes more popular my hope is that the popularity of No Mans Pass will also increase.

 

Skynet AI

The skynet system seems to be a very big success so far. The Skynet system has examined ~15,000 builds and narrowed this down to a selection of ~1000 ‘winning’ build designs. From this it has created 475 viable builds that it cycles into play. There is currently no metric to see how effective a Skynet build is as it’s cycled in other than it getting repooled back into the ‘winning’ build pool. I am not sure if I will add this fitness function to the core skynet logic or just leave the ‘cycle’ affect in play as the mechanism to generate build churn. I think I will wait until I have a full month of data before approaching the problem.

 

The Balance Update

This one is a bit trickier to digest into just pure numbers. The first metric I have been monitoring is the Time to Death and Time to Kill metric. These seem very similar, but do measure slightly different items. The time to death starts the timer as soon as a player is spawned. It continues to count up time until that player is destroyed. The time to kill metric only starts the timer when the first damage has been received. By using these metrics together we can determine the time till engagement as well! The good news is, the lowest time to kill reported is 25 seconds, meaning there has not been a SINGLE 1 hit death since the new balance changes. This is a significant achievement of the balance changes and something I was specifically trying to address. One item that is NOT captured by data and I will need to continue to monitor is the feeling of ‘power’ behind weapons. Blowing stuff up in a single hit tends to make people feel more powerful, so I want to ensure I am still indexing how the player experience of feeling like you are in a big powerful mech. Part of this is the environment destruction and allowing those 1 hit kills to happen outside of core gameplay.

Another key indicator for the time to kill is the player level. Currently, the player level seems to give zero indications into what the time to kill would be. What would be bad is if there was a clear line of kills getting significantly easier as you leveled up [this would indicate you got better parts] or if time to kill was still very low at low levels [this would indicate splash damage was still a problem at low levels]. So far, it seems like the balance pass has done a significant step forward to achieving the goals I had. One side effect of the balance pass is the time players spend in the garage is up 60% week over week! Everyone is figuring out new ways to use the parts!

Overall, that is a good overview of the metrics I have been gathering and monitoring to validate the balance choices and production choices for January. While I know the January update has received some good praise from the playerbase directly, I do always want to have some automated metrics to act as guide lines so I can ensure I am meeting the expectations of the less vocal audience as well. Happy MAVing! See you next week!

Comments 1

  1. Fenix

    Yeah, overall I’ve been loving this update.
    While I don’t entirely understand why the aim stability was drastically reduced for the Arrow/Dart cockpits and why it gets worse with the heavier armoured variants (seeing as these cockpits are generally worse than an Ivan in durability and some Hawks are lighter), I like almost every other change that was made.

    The amount of parts that make me think “why would you ever use this?” has really gone down, many chassis variants are very comparable to one another with an almost hand-crafted level of uniqueness in their advantages and disadvantages; take the HEX-Stomper and Boosted-HEX for example, they have near identical load caps but the former has more armour, better durability and a very high recoil stability, while the latter is much much faster. It’s kind of like you partially implemented my suggestion for Hexes and went a little further (even if the names didn’t make it *shrug*).

    On another note, I’m loving the Skynet AI, they put up a very nice fight and I’ve seen a re-emergence of many old favourite builds from the alpha among their ranks that were utterly extinct not long ago such as; Serge’s shotty kart (and a pile variant), Delvin’s Corvette, my old Desolator, even three different Spartan variants (one of which looked exactly like the Spartan V3 but with X-Cons instead of HDs).
    Not to mention many new builds including these shotgun-cannon bipeds that were clearly made by Delvin, and oddly I came across what looked like the Riot Sniper V3 (which was odd because I never used it and the build thread is 8 pages buried while the V4 can be found much more easily).

    It’s all pretty inspiring to jump into a singleplayer game and come out of that match headed straight to the garage with new ideas.

Leave a Reply