Support NeoGAF

[dLMN8R]

This thread is partially a response to this thread, and was inspired by this article on PC Gamer's web site:

http://www.pcgamer.com/2010/09/20/around-266505-people-played-an-fps-on-a-pc-today/

It's also a thread where we analyze Steam statistics. Fun! :lol But seriously, I want others to criticize my logic and conclusions. I've thought about this for a while, but still can't find a hole in my logic as long as my suppositions are relatively accurate.



http://store.steampowered.com/stats/

Top games (shooters only)

Current | Peak | Game
31,368 | 67,306 | Counter-Strike
29,303 | 72,730 | Counter-Strike: Source
29,045 | 75,250 | Call of Duty: Modern Warfare 2 - Multiplayer
11,471 | 19,237 | Team Fortress 2
8,569 | 14,293 | Left 4 Dead 2
4,557 | 8,282 | Garry's Mod
4,494 | 6,415 | Battlefield: Bad Company 2
3,482 | 7,232 | Condition Zero
3,395 | 9,565 | Mafia II
2,654 | 7,457 | Call of Duty: Modern Warfare 2 - Single Player
2,464 | 5,325 | Day of Defeat: Source
2,440 | 3,845 | Left 4 Dead



How can we estimate unique players over 24 hours from concurrent player numbers?
1) X = average concurrent players over 24 hours
2) Y = for every person who played a given game, the average amount of time they played it for (in hours)
3) Total unique players for a day = X * (24/Y)

Does this make sense? Is there something I'm missing here in that equation?

Look at it this way - if a game is played on average 1.5 hours, and the average concurrent player number is 7,000, then you need 24/1.6=16 unique groups of people to fulfill that 7,000 average for the entire 24 hour day, right?

Still, I'm not confident in my math, please rip it apart :lol



Examples
Left 4 Dead 2: based on the above numbers, I think it's safe to say that 9,000 people were concurrently playing, on average, throughout the last 24-hour period.

How long does the average gamer play Left 4 Dead 2? 30 minutes? 1 hour? 2 hours? The higher the number, the lower the number of concurrent players. For the sake of fairness since there's no way to know this number, let's say 2 hours.

That would mean that 9,000 * 24/2 = 108,000 unique people played Left 4 Dead 2 over the last 24 hours.

And that's the low-end based on pure guesses...I highly doubt the average person who played Left 4 Dead 2 yesterday, actually played it for 2 full hours.


What about Modern Warfare 2? 40,000 people on average? 50,000 people on average? Using the same figures as above (~2 hours average for each person), that would represent 480,000 to 600,000 unique people who played Modern Warfare 2 on PC yesterday.



Conclusion
I'm fully aware that there's no way for me to know how long an average play session is. That's why I estimated (what seems to me) very high. I can't imagine that the average Left 4 Dead 2 player yesterday actually played for 2 hours. But still, my guesses are strongly inflated, which in the end would hurt my conclusion (millions of people playing an FPS on PC).

Yet even by weighing every single guess I have to make against my conclusion, the conclusion is still the same. Add together all of the numbers from Counter-Strike, Call of Duty, TF2, Left 4 Dead 1/2, Bad Company 2, and more - you easily have millions of people who played an FPS on PC yesterday. Legally, of course, since only legal purchases are tracked through Steam.


And yes, I also know that 360 numbers for Modern Warfare 2 and Halo 3/ODST/Reach are a lot higher. I'm not here to argue that obvious number.

But the question of whether anyone plays FPSs on PC anymore...well....I think the answer to that is quite clear


Final point:
Remember that, while Modern Warfare 2, TF2, L4D, and CS are all tied to Steam, Bad Company 2 has a shitload of people playing it who didn't buy it on Steam. They are therefore not tracked in Steam's stats.

http://bfbcs.com/

Numbers are surprisingly close among all 3 platforms.



</over-analyzing>

 

[lastplayed]

Cool idea.

I love Steam's stats. There are some games in the list that appear in places you just wouldn't expect.

 

[Mr Sandman]

Remember that, while Modern Warfare 2, TF2, L4D, and CS are all tied to Steam, Bad Company 2 has a shitload of people playing it who didn't buy it on Steam. They are therefore not tracked in Steam's stats.

Makes sense. I did notice that this afternoon on the stats site it said close to 25,000 players were on BFBC2 PC. Steam is only saying 6,415 at peak. Hmm. My friend and I play retail btw.

 

[kamspy]

someone check this math, it's hurting my brain.

 

[dLMN8R]

someone check this math, it's hurting my brain.

I'll try to summarize...


1) 9,000 people playing at any given time on average
2) average person plays for 2 hours
3) Day is 24 hours long
4) To satisfy an average of 9,000 people playing, need 24/2 different "segments" to fill the day
5) Therefore, 9,000 * 12 = 108,000 total unique players

 

[Stabby McSter]

Looks like PC gaming is dead!

Spoiler

i played l4d2 just yesterday

 

[obonicus]

This is just math on made up numbers. I realize you put a caveat in your conclusion, but it's still made up numbers and thus meaningless.

 

[Ashes]

I'll try to summarize...


1) 9,000 people playing at any given time on average
2) average person plays for 2 hours
3) Day is 24 hours long
4) To satisfy an average of 9,000 people playing, need 24/2 different "segments" to fill the day
5) Therefore, 9,000 * 12 = 108,000 total unique players

Except steam isn't availible through out the known world to cover the 24 hour period; and niether Steam, or fps's are universally popular everywhere. Fps is not big in asia for example. Is steam even available in asia?
The extrapolation throughout a 24 hours period without dips and highs are woefully misleading.
If somebody just asked valve, I'm sure they could offer a better run down.

 

[kamspy]

Rationalize it like you would to someone who just got some really good indo after having some really bad indo for the whole weekend. The last thing I want to do right now is trying and deduce something, but I do want to hear good news about PC gaming.

So there's my quandary.

 

[dLMN8R]

This is just math on made up numbers. I realize you put a caveat in your conclusion, but it's still made up numbers and thus meaningless.

The only number I made up is the average playtime. Unless people play games for 8 hours a day on average, or the underlying math is entirely wrong, there's no way that my conclusion in the subject title is wrong.


Really, it's more of a discussion into what concurrent numbers mean for total daily uniques than anything. Frankly, I've thought for years that Steam's concurrent numbers do a disservice at representing how many people are actually playing the game.

Similarly, I think that common indicators in Call of Duty and Halo games on 360 have gotten gamers to over-estimate the number of people who actually play games online on average, for other multiplayer games that aren't Halo and Call of Duty.


It's just that Halo and Call of Duty are such tremendous outliers that it skews perception of multiplayer games on 360 as a whole...there's probably a reason why most games don't state # concurrent players, and that reason would probably be that not many people are playing them.

 

[SlipperySlope]

Can't extrapolate anything from these numbers other than relative popularity. I'm surprised CS and CS:S are so close.

 

[dLMN8R]

Except steam isn't availible through out the known world to cover the 24 hour period; nor is it, nor fps's universally popular everywhere. Fps is not big in asia for example. Is steam even available in asia?
The extrapolation throughout a 24 hours period without dips and highs are woefully misleading.
If somebody just asked valve, I'm sure they could offer a better run down.

Counter-Strike is massive across the world at lan cafes and such. Even Modern Warfare 2 is huge in the middle of the North American night.


Anyway, look at Steam concurrent numbers in the middle of the night and you'll still see them pretty large. Plenty large enough to still reach the same conclusion. The "average concurrent players" I chose is toward the low end based on what you see throughout the day.

 

[lastplayed]

I'll try to summarize...


1) 9,000 people playing at any given time on average
2) average person plays for 2 hours
3) Day is 24 hours long
4) To satisfy an average of 9,000 people playing, need 24/2 different "segments" to fill the day
5) Therefore, 9,000 * 12 = 108,000 total unique players

Does this take into account people playing multiple sessions per day? All my friends, myself included, will boot up a game multiple times even in the same night.

 

[tokkun]

Made up playtimes aside, the problem with your math is that you add the numbers at the end. That suggests that there is no one who plays more than one game.

 

[kamspy]

alright so maybe he's higher than me.

makes me feel a little better.

 

[Ashes]

Counter-Strike is massive across the world at lan cafes and such. Even Modern Warfare 2 is huge in the middle of the North American night.


Anyway, look at Steam concurrent numbers in the middle of the night and you'll still see them pretty large. Plenty large enough to still reach the same conclusion. The "average concurrent players" I chose is toward the low end based on what you see throughout the day.

24 period covers all timezones. Whilst I agree with the premise that millions of people are playing on pc, I don't think you have adequately proven the case using your method.

 

[dLMN8R]

Does this take into account people playing multiple sessions per day? All my friends, myself included, will boot up a game multiple times even in the same night.

Sure, Steam concurrent numbers won't count you twice. It's all about the total amount of time that all those multiple sessions add up to.

I'm still guessing that, on average, most people who played the game will have played it for less than 2 hours in that day.

 

[kaizoku]

while numbers might be healthy the proportion between big brands and lesser games throughout the year will be drastically different. and the number of big hitters isn't great.

So you have less players in less territories spending less money on less games = games industry by and large doesn't give a shit about PC FPS anymore.

How's that for maths?

Not that there isn't opportunity in future. Internet cafes in asia are like the main way many people consume the net and are ripe environments for FPS gaming to take off. Problem when I was there was besides CS they played some weird unknown one made in korea or something. Left 4 Dead 2 was there too, but they didn't even know what COD was!!

But again in terms of revenue I'm not sure how much is made by devs from a market like this.

 

[SapientWolf]

This is just math on made up numbers. I realize you put a caveat in your conclusion, but it's still made up numbers and thus meaningless.

Welcome to the wonderful world of statistics.

 

[Fredescu]

24 period covers all timezones.

No idea what you're saying, but 24 hour period is easily covered by one unemployed CS player. You don't need Steam to be active in every single timezone for it to be active 24 hours a day.

 

[dwebo]

It's too bad Steam doesn't just publish daily unique players per game. If you can find another online FPS that displays concurrent player data And a 24H total, you should be able to get a better estimate for these Steam games.

 

[Ashes]

No idea what you're saying, but 24 hour period is easily covered by one unemployed CS player. You don't need Steam to be active in every single timezone for it to be active 24 hours a day.

If anything that proves the exception to the rule rather then the average. Although I cannot suggest anything about the average CS player myself.

 

[dLMN8R]

Made up playtimes aside, the problem with your math is that you add the numbers at the end. That suggests that there is no one who plays more than one game.

If they're not only playing a single game for 2 hours in a day, but multiple games for 2 hours each every single day, I think that person is an even bigger outlier.

My math still holds up as long as the average person is playing on average 2 hours total each day...which still sounds high.

 

[Sloegr]

while numbers might be healthy the proportion between big brands and lesser games throughout the year will be drastically different. and the number of big hitters isn't great.

So you have less players in less territories spending less money on less games = games industry by and large doesn't give a shit about PC FPS anymore.

How's that for maths?

Not that there isn't opportunity in future. Internet cafes in asia are like the main way many people consume the net and are ripe environments for FPS gaming to take off. Problem when I was there was besides CS they played some weird unknown one made in korea or something. Left 4 Dead 2 was there too, but they didn't even know what COD was!!

But again in terms of revenue I'm not sure how much is made by devs from a market like this.

Heaven on Earth!

 

[obonicus]

Welcome to the wonderful world of statistics.

What statistics are you referring to, where you guess the numbers and then try to draw a conclusion? It's not just the length of the play session we're unsure about, the average number of players during the day is also eyeballed. This is actually something that could be calculated, but it doesn't seem like the OP did that.

 

[Fredescu]

Edit: Actually, forget it. I'm not going to defend using made up bullshit stats to counter other made up bullshit.

 

[dLMN8R]

Let me put it in plainer terms. There is never a point during the day when current players reaches zero. There are 18k CS players right now, down from a daily peak of 72k. In fact, right now the numbers are a lot lower than what's in the OP. I wonder if the maths still apply if used against the new lower numbers:


18,337 72,730 Counter-Strike: Source
18,259 67,306 Counter-Strike
17,078 75,250 Call of Duty: Modern Warfare 2 - Multiplayer
12,313 19,237 Team Fortress 2
7,140 14,293 Left 4 Dead 2
5,678 21,986 Football Manager 2010
5,022 8,282 Garry's Mod
3,973 6,415 Battlefield: Bad Company 2
3,900 8,797 Empire: Total War
2,101 3,845 Left 4 Dead

I grabbed numbers a few hours ago when I started writing this post, then I had to...well...work :lol

It would be interesting to see numbers gathered every hour and then plotted out to find the average. Maybe that average is far lower than what I guessed at then.

Still ends up reaching millions throughout the day especially if you reconsider how long each of those people played for.

 

[lastplayed]

Empire Total War is always so high, I don't think it've ever seen it drop down.

 

[RedStep]

My math still holds up as long as the average person is playing on average 2 hours total each day...which still sounds high.

The other problem is that your assumed quantity of hourly players are guesses, which invalidates everything else. Taking a guess at the number of players and a guess at how long each played just ends up with a fictional number. You need more stats to anchor/extrapolate from.

As mentioned above, those numbers are much lower now (6PM PST).

 

[SapientWolf]

What statistics are you referring to, where you guess the numbers and then try to draw a conclusion? It's not just the length of the play session we're unsure about, the average number of players during the day is also eyeballed. This is actually something that could be calculated, but it doesn't seem like the OP did that.

You can get a more precise guess if you do more homework but the numbers will never be exact until Valve decides to release the data on the number of unique players on a given title in a given time period.

The field of statistical inference is basically just trying to make good guesses based on a set of information.

 

[Ashes]

Let me put it in plainer terms. There is never a point during the day when current players reaches zero. There are 18k CS players right now, down from a daily peak of 72k. In fact, right now the numbers are a lot lower than what's in the OP. I wonder if the maths still apply if used against the new lower numbers:


18,337 72,730 Counter-Strike: Source
18,259 67,306 Counter-Strike
17,078 75,250 Call of Duty: Modern Warfare 2 - Multiplayer
12,313 19,237 Team Fortress 2
7,140 14,293 Left 4 Dead 2
5,678 21,986 Football Manager 2010
5,022 8,282 Garry's Mod
3,973 6,415 Battlefield: Bad Company 2
3,900 8,797 Empire: Total War
2,101 3,845 Left 4 Dead

edit: never mind...

 

[Fredescu]

How do you propose to work out how many unique players there have been?

I edited too late. There is no way of knowing so we shouldn't bother.

 

[lastplayed]

It'd be interesting to know when the peak time for each game usually hits (if it's the same each day). I'm guessing it'd be around the afternoon/evening period between NA and Europe, which is only a few hours difference.

 

[Ashes]

It's on the link:
http://store.steampowered.com/stats/
The peak is 2.7 million users. It seems to peak everyday at around half eleven.
I'm at work but I'm sure you can click on to CS and by the side, there will be stats for that game.

 

[lastplayed]

It's on the link:
http://store.steampowered.com/stats/

Of course, forgot about that. :lol

 

[dwebo]

I'm surprised Monday's concurrency numbers aren't much lower than Sunday's. Then again, I wonder how many users just leave the sign-in on Windows startup option checked and are always online.

 

[Draft]

This shit is looney tunes.

 

[speculawyer]

Counter-Strike is a like a cult. Wow. Of course, I used to be a member, but that was like 7 years ago. And people still play it like crazy today.

More people should play TF2. So awesome. Same for L4D2.

 

[tokkun]

If they're not only playing a single game for 2 hours in a day, but multiple games for 2 hours each every single day, I think that person is an even bigger outlier.

These numbers have nothing to do with what someone does every single day.

For example a person might only play games one day per week but play 6 hours on the day they do play.

When you are using daily statistics, you have no way to extrapolate the average amount a person plays per day. All you can do is look at "if someone played today, how long did they play for".

 

[Grimm Fandango]

Um... lol.

 

[I <3 Memes]

Counter-Strike is a like a cult. Wow. Of course, I used to be a member, but that was like 7 years ago. And people still play it like crazy today.

It really is an amaing game. :lol It's the king of kings and the game of games. All these other games come and go but CS just keeps rolling on.

I wouldnt be surprised if 10 years from now somebody starts a topic here about a 14 year old cousin who says "What's a Halo?" and we all rage over that. But CS will still be rolling on.

 

[Ashes]

The only thing we know to be sure is that there are at least 72,000 cs source players, because that was the peak for that game for that day. That was when you definitively had 72,000 unique players.
If you then find that there are 67,000 CS players, and this peak is at 67,000 during the exact same time as the above, then and only then, between those two sets of players you have at least 130,000 unique fps players giving time to that game that day at the same time. Therefore we would prove there to be at least 130,000 unique fps players playing on steam.
It's misleading to extrapolate that over a time period, say over a week.

 

[John]

Dodgy math analysis thread*

 

[Wolf Akela]

I do think those numbers are for unique players.

People who regularly play TF2 should know this. Remember the idling debacle? The idler program connected to Drunken Fool's idle server, and it actually reported roughly 15,000 players are usually connected at any given time. It ties with the average number reported on Steam stats.

And yes idling was really big back then, since there's practically no reason not to open the idler program if you don't wanna play, and that it doesn't eat up resources.

 

[Fredescu]

I do think those numbers are for unique players.

I don't think anyone doubts that. What's in question is whether you can take 70k now, and 50k in two hours time, and call than 120k unique players. Clearly you can't.

 

[dLMN8R]

I don't think anyone doubts that. What's in question is whether you can take 70k now, and 50k in two hours time, and call than 120k unique players. Clearly you can't.

I never suggested that it's this clear-cut. But I certainly can't imagine a situation where 20k people now represents anything less than 100,000 unique people throughout the entire day.


Look at how Modern Warfare 2's numbers fluctuate throughout the day. 15,973 right now, 75,250 at peak. Even if you just look at 15,973 right now, easily close to the minimum throughout the entire day. Even that much smaller number still represents roughly 100,000 unique people if every single one of those people played for 4 hours.


Anyway, like I said, I'm well aware that there are a lot of guesses behind these numbers. But I think GAF vastly over-estimates the average gamer if you really think 2 hours/day average is the norm.

 

[SuperSonic1305]

I thought the thread title said Doggy Style at first.

 

[dLMN8R]

I thought the thread title said Doggy Style at first.

Then Steam would have everything!

 

[Sent]

It really is an amaing game. :lol

And it was a mod at first which is nutty to think about now. Then again so was Killing Floor.

 

[Ashes]

Let's try another way. The peak is when most people play, how do you then have nearly 400,000 to nearly 600,000 unique people playing it every singly day?
You think pc steam sales for MW 2 were so high, that you have at least 400,000 people playing it on steam every single day?
The absolute peak is only about 2 million or so for every person *logged in* concurrently, on steam, then consider that CS is double MW's figures. The peak being when *most* people play at once.