Statistic tracking... For good?

"O ho," I hear you chortle, "Stat tracking? Why, this is not only a duplicate thread, but the idea has been bitchslapped down from on high! The thread will be locked within minutes!"

Well, I'm not talking about normal stat tracking. I'm not even talking about using it for what it is traditionally used for. Let me walk you through the convoluted mental processes that spawned the concept before anything else.

Over the past few weeks, simply attempting to read the threads in the Suggestion Forum has become an exercise in futility. Every topic devolves into baseless assertions and endless back and forth, and no conclusion is ever reached. One person will claim that the grenadier class is underused, another will claim it is overused; one person will cite the weapon config files and claim one rifle is more powerful than the other because its damage is higher, and another person will cite the same file and claim that the different accuracies make them balanced. The fact of the matter is that no one person has any hard numers with which to bolster arguments, and such arguments are sorely needed. That's where the old idea of stat tracking comes in.


Here's how it would work: a simple script would run in the background on the official Empires server (since it gets the most traffic and displays the most regular gameplay). The script would sit back, watch what was going on in the server, read the logs, munch on a bag of Doritos, pretty much do the usual statistics tracking thing. However, it would not record names, Steam IDs, IPs, nuthin' that could be used to identify players. What it would record is the behavior of players over the course of the game, paying attention to such things as:

• Classes - which ones people are picking over others on different maps.
• Weapon use - what weapons are being picked over others, how many kills are attributed to specific weapons over an amount of time, the actual accuracy of weapons, etcetera, etcetera.
• Vehicle use - how many vehicles of each type are being purchased over the course of a match, how early or late into the match certain vehicles appear, the weapon loadouts chosen, the armor types used, and the kills accumulated by the vehicle.
• Research - exactly what is being researched and what isn't, as well as when things are being researched.
• Match timeline - the time of the first major engagement and subsequent battles, as well as the amount of people and vehicles on the field over the course of the game.
• Team victories and losses - a simple record of which team wins most of the time on certain maps.

After the map ends, the script would write it to a simple text file in an archive, timestamp it, and move on to tracking the next round. No uploading the stats to a webserver in real-time, no artsy little siggies for people on forums, just simply a file of what actually went on in the game. At this point, another little program would swoop in. It would search the fields in the generated logs for the statistics and crunch the numbers. It would, first and foremost, calculate the average statistics for that week or month, and maybe even pop out a scatter graph or something for more interesting statistics. It would write the results to the same archive, and there it would sit, the final product; hard numbers detailing the course of the average game.


Now here's where it gets interesting. Every week or month or so, the end file is uploaded to the Empires site for everyone to see, as well as the archive from which it was derived. The people on the forums are then free to pore over it and use it to bolster their suggestions and arguments on the forums. This would hopefully lead to an actual resolution for some of the suggestion threads, as it is hard to argue that a certain weapon is not overpowered when it is claiming half the kills in an average game, and only one out of eight players actually use it.

It would not just serve the community, obviously; the developer team could find the information very useful for tweaks they've applied. If a change to a map suddenly makes the win ratio between the teams level out, then obviously they are free to turn their attention elsewhere. If a weapon balance tweak triggers a drastic spike in the ratios of a certain firearm, then they need to get that fixed immediately.

Furthermore, giving players information like this simply for the benefit of their knowledge would help contribute to an air of openness. Letting your base know what's going on never hurts, and builds a healthy community that is comfortable with communicating with both the developers and each other.

I hope that you all give this proposal serious consideration, and that if you see a problem with this idea (besides making poor Krenzo do all this work), you not only voice it, but propose a solution. Thank you for taking the time to read through this in its entirety; I await your productive (please, god, let them be productive) comments below.

 

*After reading the first line of the post*

Aww, the thread's not locked yet. Damn. Guess I'll have to read the whole thing.

*After reading the full post* Gotta give you credit, I like the idea. My problem is minor. Comparing the On|E server (Clanners versus public players) versus the Empires server, I've noticed just by playing on both that the equiptment the players use is drastically different. We're all riding in Light Tanks with absorbant armor and advanced coolant (we being On|E). While the Pubs(on our server usually, and on the Empires) are all riding around in mediums with gas turbine and composite.

Basically, what I think I'm trying to get at is, stretch it to all the servers aside from just the official server. Because I'm sure we're not the only clan that swears against the norm, and we should be represented even if we're an outlying variable.

 

The problem is that clan play is drastically different than normal play. The official server is where most of the casual players go, and it probably represents a good segment of the Empires population. I would certainly see nothing wrong with you being able to run the same program on your server, but I would think that including clan servers in the official tracking could significantly skew the statistics.

Clans also have a way of adapting to any sort of changes thrown their way. It's getting it balanced for all the incoming players that I worry about, honestly. You doods will persevere.

On a final note, if the base grows significantly within the next few releases, I could certainly see the program being run on multiple servers. If a clan server happened to be where a good segment of the normal population was playing, I don't think that it would skew it too much.

 

I think this is a great idea, how hard would it be to make?

 

o think it shoud be for all servers

i really like it

not even steam has such an advanced way of "product evaluation"

it woud really help krenzo massively, not even in balancing, but in actually seeing how players play the game itself

i say

10/10

 

I doubt you could find all those statistics reliable

 

I don't think it would be hard so much as time consuming. Churning out the code for the gathering program would involve hours upon hours of hammering away on the keyboard, specifying exactly which fields to search for what numbers and what to do to them to calculate the actual output. The number-cruncher program would probably take less time, but it would still be a significant amount of work. Getting it to grab all the right fields from the archived files and making it run though the math would be pretty easy, but it would certainly take quite a few hours to get all the code laid out and logic-error free.

That's why you would have the second proggy average them all out, see? The ratio of deviant games to normal games would be pretty low, and averaging it would smooth out the wrinkles. Plus, if there's one match that's totally throwing the curve, you could just exclude that file and try again (although you wouldn't want to delete the file; it should be included in the archive of files provided to the community no matter what, just marked so that people know it wasn't crunched with the rest).

 

i disagree, basically the server console keeps track of all the wanted data
we need somthing to write it down thoug

 

It's definately possible, Dystopia did it a long time ago, and I think it would be a good addition. What I also think would be good to steal from Dystopia is their suggestion forum layout (http://dystopia-game.com/forum/viewforum.php?f=46). This way we can actually get some decent feedback on the suggestions, and know what to stay clear of.

 

so you noticed huh ?

but yeah its a great idea would love to see it implemented and would be a really useful feature for a game.

productive section commences here:

we need an average playing time as well, i know you said it would record the amount of people but i think it should take an average of how long each person plays for as well. (we end up with a figure like 75.4 minutes not 30 hundred different figures)

not sure if you meant for that figure to be included anyway was just making sure.

 

I love that idea... Would certanly help everyone, as well as a plan to devise new strategies...

Oh and I play on the On|E server almost exclusivly and to be honest, clanners don't really play that often on it, most of the time there are alot of pubbies on it...

It sounds like a fairly simple script to implement as well...

Im hands down all for this...

 

Must remember to do significance testing...nothing would bother me more than deriving conclusions from a bunch of means alone. So long as you get a statistician (like me ) to go over the results, I am all for it. Would be very useful for balancing

 

I never read anything in this thread but krenzo did seem interested in creating a stat tracking system.

 

Hate you. Hate you so very much.

Taco, the idea is indeed to involve the community in interpreting the statistics. You would be beyond welcome to go through the archived statistics, as would the rest of the community. That's the whole point of making them publicly available!

 

i wish i had something to say, but everything i had to say has been said. so ill just say good job, i like it.

 

its quite a strange feeling isint it?

im having it too...

better post this idea on the idea collecting tread

 

Eesh, is this what happens to good ideas? Everybody agrees, so it just gets ignored? Bah! Bah, I say!

I would kind of like to hear Krenzo's take on this, since he would be the one cranking this out. Assuming he doesn't smack it down, it would be nice to know that it's actually on some list, somewhere.

 

I'm going to have to agree to the previous posts that say that there's nothing really left to say here, except that I agree.

 

I am indeed planning on implementing a stat tracking system and charging money bags for it to support me building a top of the line server to hold the stats as well as migrating the official server to it.

 

...and gold-plated diapers. I'll pay!