Request DDOAudit math help

[DDO Gaming]

I logged in literally every character I transferred on all 6 accounts last Saturday to unpack them. They were logged in, unpacked, and then their bank was checked to make sure their stuff was there. Behavior like this totally messes with your data. They were on for less than a minute each (assuming the lag permitted that quick an unpacking and bank check).

so we'll return to the debate next week. I would be surprised if the graphs change much unless the lags are fixed inwhich that 2000 average will probably rise to 3000+

 

[SquireZed]

I only considered 64-bit data (you can choose to include only 64bit...besides 32 only adds upto 100 extra per day)

Ok so let's consider three seperate scenarios:

1. a player logs into 1 character and plays quests over a 3 hour period

2. a player logs into 5 characters and plays quests over a 3 hour period

3. a player logs in then AFKs for 3 hours (I find this difficult to believe because...I get booted off the system if I AFK more than 10 minutes which is reasonable since during the AFK period the server is still updating my system)

4. a player multiboxes so they can use dungeonhelper to play 4 characters simultaenously

Scenario 3 is dismissed because its only relevant for calculating lag factors (and wasting precious servertime which inturn contributes to lag. SSG are you reading this?)

From the perpective of "how many characters online within a 1 hour-block) scenarious 1 and 2 are the same. However counting unique characters is suddenly where these two scenarous become important

Scenario 4 contributes to both "number of characters online within a 1 hour-block" and "how many unique characters online" however I haven't seen many examples of multiboxing (4 on Cormyr...in 8 months) so I propose this covers a small percentage of the gamerbase

Scenario 2 is actually interesting because it explains why there are so many unique characters and I admit I didn't consider this possibility

So there are a bunch of problems here.
1. You can't choose to only consider 64-bit data when you are considering a data point that doesn't discriminate. You can look at player counts for the 64-bit server but the "In the last quarter, we've cataloged 688,409 unique characters and 28,289 unique guilds" bit you reference does not. That's like saying "Oh I only looked at the oranges I have in this basket of fruit" when the number you're comparing it to is the total number of apples and oranges in the basket.
2. You are making huge assumptions about AFK because you just want to. I have AFK'ed for more than an hour on accident because I forgot I left the game running, never got booted. It is also unlikely that AFK players (in public instances) actually contribute to lag significantly. A lot of effects pause in public areas and the issues that seem to be related to the performance issues based on recent posts is largely due to memory garbage collection, where an AFK player likely does not contribute, so pointing it out to SSG is moot. If AFK players were the problem they can just auto-disconnect AFK players more aggressively. While instances have a 30 minute AFK timeout (60 for VIP) in no case is it 10 minutes and I believe that only applies when you're in an instance and not a public area (or I've grossly underestimated my own AFK time, which I find unlikely). You've also chosen to arbitrarily exclude that number for... reasons? But if you log in and AFK, you're still going to count in the number of online players and your login is going to count for unique characters for each character you AFK on.
3. You are still just applying random numbers to other random numbers in ways that don't critically think about how those numbers impact each other. Again, someone might log into 5 unique characters in a 3 hour period or 1 unique character in a 1 hour period, but *NEITHER* of these figures really matters that much about the total unique character count because I can play on two unique characters for 400 hours in three months and they still count as two, or I can play on 400 unique characters in two hours by just quickly swapping between them (well, ignoring that it's probably improbable to manage that amount of character switching with the servers as they are) and that would be hugely impactful. Online players and unique characters ***ARE NOT RELATED*** the way your interpretation of the data implies. There are obviously some baseline overlap in so far as someone who doesn't play the game in the quarter does not have any data represented and someone who plays some during the measured period will be represented at least once, but using number of online players to explain the total unique characters is a fundamentally flawed assumption.
Also you're still asking me to consider scenarios that are not a comparison between the data you're analyzing in any kind of responsive way. All four of these scenarios only look at online character count, and then you try to count unique characters, which is not something that data supports, even if you try to make it work that way. It's how you wind up with crazy data validation errors where you're an order of magnitude high on your estimate, because you are not interpreting the data right, and no amount of trying to massage online player counts to get unique characters will work when the realistic answer is that unique character counts are not directly correlated to active player counts because of a different thing going on (character transfers that require a character to be logged into twice in a way that probably counts twice) and trying to extrapolate that data from a different point is an exercise in fundamental denial of basic data analysis.
To massage the data you want to get the answer you want you would need to make all the incorrect assumptions you've been making so far, and then claim absurdly that the average player plays eight hour sessions and plays a different character each time. That just doesn't happen and is observably not accurate.

TL: DR
Until you give up on your fundamental assumptions and look at what the data actually says, you're just conjecturing and there's no information to take away from here, hypothetical scenarios and conjecture do not prove anything and when you use the wrong data to prove a result it obviously will not validate and intentionally misinterpreting that data to try to fix the data validation error will just result in a different wrong hypothesis when realistically the problem is that the two data sets are not correlated and a *MAJOR OBSERVABLE PHENOMENON* explains the disparity without requiring mystical math to try to solve the question.

 

[Lofen]

so we'll return to the debate next week. I would be surprised if the graphs change much unless the lags are fixed inwhich that 2000 average will probably rise to 3000+

You're ignoring the message that every poster here is giving you, the problem isn't that the time frame you have picked is wrong and messes with your math (although that for sure isn't helping), the problem is your assumptions. You assume too many things with nothing to base them on aside from the fact that they correlate with how YOU play the game.

In order to demonstrate how inaccurate this reasoning is people have given you examples left and right about different types of players and behaviors, showing that there is no "one" proper player, and that even if there was such an average player, that average might be far from indicative of the majority.

Returning to this experiment in a different time frame with the same assumptions (this is how I play therefore let's assume all do) will result again in a poor analysis.

Please try and articulate what you would like to know, what are you trying to find with all your assumptions and math, so we can see if it's attainable and hopefully help get it.

 

[DDO Gaming]

You're ignoring the message that every poster here is giving you, the problem isn't that the time frame you have picked is wrong and messes with your math (although that for sure isn't helping), the problem is your assumptions. You assume too many things with nothing to base them on aside from the fact that they correlate with how YOU play the game.

Every poster on this thread is claiming that every gamer's interaction with the DDO system is so unique that ultimately all the DDO Audit graphs data is worthless.
Its an interesting argument however:
1. if one creates dots representing every 5 minutes
2. then connects those dots together thereby creating one contiguous line

I'm pretty sure that contiguous line will flow rather than be jagged. That suggests gamer activity across the community is pretty common. I know that is a slightly complex argument BUT you can mimic it by throwing a D50 at random intervals then graphing it. Connecting those lines together will create the jadded effect you are claiming is the norm amongst the DDO community

 

[SquireZed]

Every poster on this thread is claiming that every gamer's interaction with the DDO system is so unique that ultimately all the DDO Audit graphs data is worthless.
Its an interesting argument however:
1. if one creates dots representing every 5 minutes
2. then connects those dots together thereby creating one contiguous line

I'm pretty sure that contiguous line will flow rather than be jagged. That suggests gamer activity across the community is pretty common. I know that is a slightly complex argument BUT you can mimic it by throwing a D50 at random intervals then graphing it. Connecting those lines together will create the jadded effect you are claiming is the norm amongst the DDO community

That's not what anyone is saying.

People are saying that different data points prove different things, and two different data points do not point to the same thing. You are intentionally silo'ing off observable phenomenon (transfers leading to dramatically increased unique character counts in a limited window) to wildly extrapolate from data that doesn't indicate anything that has any way to reflect unique character counts based on a sample size of one. That's just bad data science, sorry.

I can draw a contiguous line between days I eat lunch and then try to say it defines unique character counts in DDO because I eat lunch every day and over every day for the quarter I've had lunch roughly 90 times and based on that when I eat lunch roughly 7,650 unique characters log into the servers but that is no more accurate an analysis than trying to compare active player counts to unique characters because those two data points (while somewhat more correlated because of the nature of both being related to DDO's player base) are not directly related.

 

[DDO Gaming]

Week and month range data have remained consistent so we can dismiss 32bit server controibutions. Furthermore about TWENTY THOUSAND additional unique characters have logged in since the transfers were ended. One can claim SOME of them were unblocked character transfers from DDO hades however I think its perfectly acceptable to suggestion at least 50% (upto 10,000 characters) are newly created