> 2533274801176260;8424:
> On the other hand, how would it be possible for the peak population to decrease without the overall number of players to do as well? If one gamer plays for X hours, his playtime will overlap with a certain amount of other people playing. At some point during 24 hours, this overlap is maximal, and this is where that value is recorded. If the number of players stayed the same, the amount of time X that they play would need to decrease in order to account for a lower max population. This time decrease cannot sink arbitrarily low, with the low-end minimum being the time it takes to complete one match, which is ~6 minutes (not counting booting the game and search time). Players that change their playing habit from weekdays to weekends will skew the curve shape towards a rise on weekends, which actually can be obeserved in the final weeks of the graphs. It is the weekend max UUs that rise above 20k, so we already accounted for that by taking the upper estimate instead of the lower one.
Consider this simplified scenario: we have N players each of which logs in for some fraction T of the length of a day at an arbitrary time. What is the average number of overlapping players at any given time? The probability of finding any given player online at some moment in time is T/(1 + T) (to explain the 1 + T, consider a player signing in at 23:59). The probability of finding k players online is given by the binomial distribution with p = T/(1 + T). And as we know, the expected value of k is Np, which is a strictly increasing function of T. So, we can decrease the expected value of players found at any given time by decreasing T, without touching the total number of players.
Now, of course there are many things here which don’t hold up in the real world. But I hope this shows that it’s indeed possible for players playing fewer hours per day to decrease the expected population at any given time (and therefore also at the peak time) even when the total amount of players per day remains constant.
> 2533274801176260;8424:
> Now I fully admit that in the heat of the moment, I misworded myself by inferring that 95% of the population left. But, while it is not a direct proportionality between the population and the max UUs (and not even to the 24 hours cumulative players), there is still a strong correlation that can be used to at least indicate a trendline, which in the end was all that I was going for, as I merely wanted to show that Halo 4 also suffered a strong decline without the issues of lacking content and bad story (which doesn’t effect multiplayer that much anyways). However, now that you linked your data, I added a bookmark to it on my browser (I did read it previously, but just couldn’t find it later on) and will try to use that as a basis for future arguments from here on out.
There’s definitely a correlation, but that wasn’t really my main concern. My main concern are actually people who take what you said literally without any sanity checks, and then go around declaring that Halo 4 only had 20,000 players in total after one year. I’ve seen way too many of those people, and I at least try to prevent sensible people from spreading that any further.
But thanks for considering using my collection of data in the future. When it comes to the facts, I think this community needs neutral sources more than anything.