The truth about AC multiplayer

Latest Numbers:
https://www.assettocorsa.net/forum/i...about-ac-multiplayer.30472/page-6#post-743841

____________________________________________________________________

Hey guys,

end of last year I started my work on a still secret project. Unfortunately because of real life and some missing incentive to continue the work I doubt it will ever reach a state which correlates with the idea behind. First when I talked about it to @Minolin and asked for a good name his answer was Edwin. I still don't understand why, but whatever the basic logic is called Edwin since then ^^

So Edwin is a server plugin independent system to gather data and because of its nature to only be able to collect data from open accessible interfaces, the data isn't spectacular but complete. Edwin is alive since 2nd december 2015 and actual it's only used to identify some wreckers, when own server logs or speedwise or RSR or minorating fail to know a specific player, because Edwin knows them all. And today I decided to open my data in order to stop all these rumors or guessed numbers around when it comes to multiplayer usage. It also helps a little bit against seeing the work done as senseless Last week there was a little problem in the Edwin logic so the data isn't as good as it should be, but I think it still gives a basic idea about what is going on.

1. Player statistics:



After two months Edwin knows nearly 36k players who were playing AC multiplayer and actual the daily playerbase is around 4k players. When I compare this with some numbers from Steam Spy sites, my conclusion is that the multiplayer usage of the AC playerbase isn't 10% or whatever number was discussed here in the forum. It's more something between 30% to 50%.

2. Players online:



Please ignore the huge dropdown at the end. The blue area is the average concurrent player number of each day, the grey area is the maximum concurrent player number per day. The average seems to be very stable, but the release of DLC3 (15th december) had no effect to the multiplayer usage.

3. Servers online:



And again please ignore the drop down ^^ Blue is again average, grey is the peak value per day. While the player numbers are staying at the same level, the server number is growing slowly. But we've got more servers then concurrent players online, so there isn't at least 1 player for each server. The times when Edwin saw a higher player number than server number can be counted with both hands.

4. Most used tracks:



That's where I stopped the work, but maybe also a little bit interesting to see the top 15 used tracks. So around 10% of the servers are using Spa. And around 30% of all servers running of a pool of 5 tracks.

 

Players online last 24h, i couldnt believe it was such a high number. :O Nice! And great job!

Players online last month is also much higher than i could ever guessed. Because of sales/christmas maybe?

 

I don't think so... I don't save these statistics because the idea behind the project is a completely different one, so all can display is the live data (point 1) and some calculations of the tracking results (point 2 & 3). But maybe I should think about saving these statistics.

I also checked some old discussions with Minolin and the players online number after the first 22 hours of Edwin being alive: 3,584 players
So all numbers seem very stable and without any big impact because of DLC3 or christmas or whatever.

 

Is it possible to track how long each player spends on online session per day or week?

 

Great work, thanks.
This is an awesome piece of advice for Kunos, in order to eventually understand the actual weight of multiplayer platform of AC.

 

Possible is a lot of stuff, but like I said all this data is more a side effect of the things I need for the idea behind my project and it's not about tracking. For every tracking plugin which is running server side (e.g. stracker or speedwise) it's much easier to get what they want.

And there are some limiting factors like server ressources (you would need a much higher logic cycle for detailed tracking and also a more frequent and faster cycling to reach a useful data quality for your approach, which will end in much higher monthly costs) and on the other side also a much higher stress factor for the servers. My aim was to be able to gather the needed data as cheap as possible and without any noticeable impact for server owners. So Edwin is more like an old man (running on an old slow server) who is walking slowly around and have a look what is going on ^^ What he is seeing take its time to be calculated, but the frequency is high enough to get valid data.

 

Im sure Kunos have their own ways of seeing how the activity on mp vs sp is
What Jaye have done is interesting for us players though. Its much more activity online than i thought.

 

In reality you should thank @Michael Hornbuckle ^^ His mantra about less than 10% of AC players play online was a big impulse to gather more data I needed

 

I'm not so sure of this.
Michael said that thing about 10℅ of mp users because IIRC Stefano stated first it during a Q/A of his code-streamings.

 

You cant seriously believe that game developers doesnt have ways of seeing how active their game is in both mp and sp? They can probably see what kind of tyres that are used most if they want lol.

 

Pretty sure the devs know how many people have bought AC and never installed it. Are you sure that the stats gathering program isn't counting people that are pre-booked in booking servers because that would certainly skew the results.

Please explain how you have come up with the 30%-50% stat?

 

Yeah but Michael says a lot of things and one nice thing about numbers is that they can proof every point you want when you use them in another context...

Yes I'm sure, this was one of the basic features I needed. Without that being possible I would never have started to work on it. Sure there is some fuzziness, because each gather cycle needs a lot of time and between each cycle there is also some time. But it only counts real online people and don't count duplicates. So in bad cases I maybe don't get a few guys because of switchting servers in the gathering timeframe, but there is no case where I count too much players.

Steam spy says 300k owners of AC, Edwin knows 36k after tow months => at least 12,5% multiplayer users
Steam spy says 50k player last 2 weeks, Edwin says 17k => at least 35% multiplayer users

So nobody knows how accurate steam spy is or how bad the impact of 4 days of less tracking last week was, but the actual percentage of players also playing online is for sure higher than 30%. And in my opinion comparing overall owners with online users makes no sense when less than 20% of the owners played the game in the last 2 weeks...

Or take the live data:
Steam Spy says 1,607 AC players right now, Edwins last cycle found 461 players => 29%

I'm pretty sure it's something around 40% over a monthly timeframe, but we also can wait 2 weeks so I have complete tracking cycles again.

 

I just manually counted all the people in locked and unlocked non-booking servers and the total was 293; not sure where your program is seeing 461.

But even if for the sake of argument we agree that 40% of the people that have started AC in the last few days have joined an MP server, that is still a minority. So what exactly is your point?

 

I believe the statistic was that only 10% of people that own AC have used MP....

"Steam spy says 300k owners of AC, Edwin knows 36k after tow months => at least 12,5% multiplayer users" that would therefore line up....

 

And a bit of backstory that I think is relevant:
This whole discussion started because people were upset that Kunos was 'wasting' time working on AI and other SP features instead of spending all their time improving MP. The original 10% stat was given to me by Kunos to show that even though MP players are far more vocal about things they 'need', at the end of the day they are the minority and Kunos would actually be doing the majority of their customers wrong if they spent all their time on MP.

No matter how you twist the numbers the fact remains that the number of people that have bought AC and then clicked on the online tab, even just once, are a minority. All this effort spent coding Edwin has only reinforced that fact.

 

When you only count non-booking servers you nearly miss 10% of running servers. I checked my data, but unfortunately the column where I wanted to save the online clients shows still zero, so it was on my list but I didn't finished this feature. Maybe also the content filter active? And don't trust the lobby too much, it's not updated as fast as needed to get the right data. Because of that I (and the pre-booked problem) I need much more logic and process server for server in detail. This is also a weak spot of Edwin because I skip servers where the lobby says zero clients, but on the next gathering cycle I will get my data for these players.

Some technical numbers:

• Total: 15,661 cycles
• Cycle duration average: 22.81 seconds
• Cycle duration minimum: 6.22 seconds
• Cycle duration maximum: 242.56 seconds

So you see the system was hard-working the last 2 months and I think it is really working very accurate. By the way it says you was last time online 5th january, you should play more multiplayer

My point is just that I wanted to use my work for something useful instead of delete it some day because it will maybe never be finished. And that we have got valid numbers for further discussions instead of some voodoo numbers from over a year ago.

Yes, but does it make sense to compare it with the overall number instead of active players? The 200k players who didn't play AC in the last 2 months are surely not the target group for upcoming DLCs or whatever, because they won't buy it ^^

And the people who have bought the game and still playing it are also a minority (17% regarding to steam spy), so just stop AC? My effort to code it was because of a completely different idea and after seeing the first numbers coming in, there was some kind of disillusion to push it forward because only the basic work was finished and a lot of stuff todo for maybe in the end a couple of guys using it.

This 10% number is very old, was measured in a time where the multiplayer was really at a bad level compared to today. And today we have got an active playerbase of around 50k people with around 20k of these also playing online, that's the whole message behind the thread. Nothing about what Kunos should do or shouldn't do, no elitism, no discussion about if offline or online is more important. Just a snapshot of the actual situation (and some proof of my opinion that with the time every game will split offline and online usage by equal parts ^^).

 

But for prioritys I would say important are the actual active players and how many of these use MP. Then we get a 30% as it seems. So 30% is not a minority from my point of view. So maybe MP is more important than expected.

 

If its not a majority its a minority.

 

aah, thanks for the enlightenment.. maybe i should have said "its not a major minority"