Steam Gauge: Addressing your questions and concerns

Plus, we offer an expanded list of the Top 100 most played Steam games.

Read the whole story

 

[hairyfeet]

There is an easy way to compare, ask AMD Raptr users to compare their data as well as asking Raptr for their game usage data and comparing it to Steam. I have noticed since using Raptr that my Steam time (compared to Raptr's reporting of my Steam time) can be off from anywhere from a few minutes to over a half an hour.

Now why that is? Heck if I know. I have noticed that those times when Steam connection is sketchy that I lose more time, perhaps Steam tosses that data when the connection is off? I have also noticed those times I have had to reinstall a game if it doesn't have Steam cloud saves likewise it'll reset the hours played, so I'd take Steam time with a serious grain of salt.

 

[ChrisSD]

pqr[/url]"frkt4y1]Kyle, thanks for log-lin plot.

Thanks from me too. It's rare to see such a welcome sight in a more journalistic setting. It's a shame about the pie chart though. They make me cry.

 

[lowlymarine]

It's interesting to see SMNC on this list, since that game is dead as a doornail these days. I tried a couple times in the last couple of weeks to get a game going, but gave up each time after ~20 minutes in the queue. It's a pity; that game was my favorite "DOTA-like" but Uber just left it to rot.

It makes me wonder how many of the other games still in the top 100 don't have anyone still actually playing them.

 

[CatG]

Kyle Orland[/url]":1zq7aygt]

Zandros[/url]":1zq7aygt]It is somewhat surprising that KSP doesn't show up in any of the charts. I've no idea how much it's sold, but it seems to me like it could've been a contender in the median number of hours played.

KSP is 168th in number of players, with about 645K by our estimates. Not too bad.

It's important to note too that KSP is also sold outside of steam directly from Squad. Really, the community of people that play it seems to be small, but fierce and devoted. It has by far the most hours played in my Steam library.

As an aside to Peter Bright - how long until we see the ArsTechnica Dota2 team trying to break into The International? And have you thought of casually suggesting to Conde Nast that esports might be a place for sponsorship and exposure to a young audience?

 

[Turin Turambar]

Thanks for the extra data. I still would like to know the exact number of games with +1M copies, as the graphic can be judged in several ways.

 

[AstroGrover]

What's with the ones where 100% of all owners have played it? You can't tell me there's not a single person that hasn't played their copy of TF2.

 

[FoolishOwl]

Huh. So my taste in games is pretty mainstream, as it turns out. I've got more hours logged in Skyrim than in anything else on my Steam account, with Civilization V in second place -- which surprises me, by the way, as I don't remember playing Civ V more than a couple of times.

I've had Team Fortress 2 almost since it was released, but I've never been able to get into it somehow. The pace of gameplay seems off, somehow.

I'm actually quite surprised about DOTA 2. I'd seen references to it here and there -- like Jinx, the online tee-shirt vendor, selling shirts with DOTA 2 logos. I got around to looking it, and found out it was free, only about two weeks ago, and got through some of the tutorials. It seemed like a decent game, though my kids assure me that it's much too complicated and that they don't like it as much as Smite.

 

[vartec]

AstroGrover[/url]":2if4cw9c]What's with the ones where 100% of all owners have played it? You can't tell me there's not a single person that hasn't played their copy of TF2.

The ones that have 100% are F2P, so I guess Steam only counts people who actually play them.

 

[Pantagruel]

I personally think the "games bought but never played" data from the other article is one of the most important take outs of all this. It seems Steam Sales create an opportunity for titles to generate revenue from audiences they don't even actually appeal to. I really can't decide if this is good or bad.

 

[vartec]

Pantagruel[/url]":1qlwang6]I personally think the "games bought but never played" data from the other article is one of the most important take outs of all this. It seems Steam Sales create an opportunity for titles to generate revenue from audiences they don't even actually appeal to. I really can't decide if this is good or bad.

Well, there are games that I bought but haven't time to play, but that's not very typical. Actually I think most of owned-not-played come from bundles. I've bought quite a few bundles of 6-8 games just to play 2-3 of them.

 

[Shusien]

Also note games with launchers do not always accurately track time spent playing. Warframe for instance (in the 64bit build) originally did not track time played (outside of the launcher) and even did not have the Steam Overlay. I wracked up probably 60 hours before it ever started working. Another game comes to mind was Ragnarok 2 (only for a little bit). The launcher would be tracked as the game for about 2 minutes of updating then stop once the actual game was launched.

 

[Astramancer]

Kyle Orland[/url]":1uzgiygy]

Archangel Mychael[/url]":1uzgiygy]I'd like to see some Linux data to see how well it's doing.

I'd love that too, but SteamCommunity doesn't break out which platform version of a game a person is or isn't playing...

Are there any linux-only games? It wouldn't be a totally accurate picture of who's playing on linux, but if they have linux-only games, you at least have a minimum number. Maybe if someone has a high percentage of crossplatform games and a linux-only game?

 

[Pyros]

I can confirm that most of my steam hours played numbers are accurate, but a couple titles like TF2 are off a bit. I suppose that may be the 2009 thing though.

 

[rimbaud]

Would you please indicate which titles are available on STEAMOS? Would be very interesting to see how much work valve has to do on the top 100...

 

[skicow]

Wow, Half-Life 2 was released almost 10 years ago....

 

[Kyle Orland]

Hydrargyrum[/url]":12jbsvwo]Something's gone wrong with the Publisher and Developer columns towards the end of the list - I know that Trine 2 is a Frozenbyte game and Dishonored is Arkane Studios, but this chart has the credits offset by one place, so Dishonored is credited to Frozenbyte and Amnesia: The Dark Descent is credited to Arkane Studios. I'm not sure exactly where the rot sets in but I suspect "War Inc: Battlezone" is not a Valve game.

Yes, there was an error in my copy/pasting that caused a misalignment initially. It was fixed within a couple hours of the original post. Thanks for the note!

 

[eXceLon]

Steam doesn't measure my gameplay hours correctly, even since 2009. Is this common?

I'll throw my anecdotal example in here.

Steam Community: eXceLon

My gameplay time generally seems accurate, with the exception of the time logged for Battlefield: Bad Company 2. As much as I enjoyed that game back in 2011, I know I didn't put nearly 300 hours into that game. That's nearly three times as many hours as I've put into Skyrim.

Even counting that time I spent two hours on an empty server and used C4 to destroy every single destroyable building in town...

 

[danstl]

All I can say is TF2 has got some legs - that game is fun, and the enhancements made over the years have been great. I am always shocked how many people (including myself) still play it!

 

[Orestus]

This is a fascinating article/series and a great piece of reporting and investigation. Bravo!

Is there any explanation as to why DOTA 2 and TF2's owner numbers and player numbers are identical. It would seem to me to be statistically impossible for the 25,926,966 estimated people owning DOTA 2 that every single one of them has in fact played the game. Same w/ TF2 (I have TF2 as an "owned game" in steam and have definitely never played it for instance) I understand the final numbers are estimates based on your polling, but is it really the case that every single account you've polled, which still seems to be a huge volume of accounts, that owned these games has also played them?

I suspect Steam is going to quickly take action to stop this type of reporting considering how tightly they guard their sales numbers, but run w/ this while you can.

 

[robotic_tourist]

From the original article there were a lot of people commenting to ask why the Steam API wasn't used. Any more information on this?

 

[Kyle Orland]

Orestus[/url]":2h4t8zmw]This is a fascinating article/series and a great piece of reporting and investigation. Bravo!

Is there any explanation as to why DOTA 2 and TF2's owner numbers and player numbers are identical. It would seem to me to be statistically impossible for the 25,926,966 estimated people owning DOTA 2 that every single one of them has in fact played the game. Same w/ TF2 (I have TF2 as an "owned game" in steam and have definitely never played it for instance) I understand the final numbers are estimates based on your polling, but is it really the case that every single account you've polled, which still seems to be a huge volume of accounts, that owned these games has also played them?

I suspect Steam is going to quickly take action to stop this type of reporting considering how tightly they guard their sales numbers, but run w/ this while you can.

I've added an answer regarding why free-to-play games show the same number of players/owners in the piece above just now.

 

[Kyle Orland]

robotic_tourist[/url]":47tna2mt]From the original article there were a lot of people commenting to ask why the Steam API wasn't used. Any more information on this?

I'm a bit of a hack programmer?

Seriously, when I started on this, I wasn't even aware that this info was available via API, in addition to the web site. By the time I found out about it it seemed relatively pointless to start over and change everything up... the data is the same either way, right?

 

[TylerH]

You got the release date for TES V: Skyrim incorrect; it's 11/11/11, not 11/10/11.

 

[Kyle Orland]

Astramancer[/url]":300kfpab]

Kyle Orland[/url]":300kfpab]

Archangel Mychael[/url]":300kfpab]I'd like to see some Linux data to see how well it's doing.

I'd love that too, but SteamCommunity doesn't break out which platform version of a game a person is or isn't playing...

Are there any linux-only games? It wouldn't be a totally accurate picture of who's playing on linux, but if they have linux-only games, you at least have a minimum number. Maybe if someone has a high percentage of crossplatform games and a linux-only game?

There are almost no Linux-only games on Steam... they are by and large ports of existing PC/Mac games.

 

[Kyle Orland]

Sufinsil[/url]":1sml2fli]Any thoughts on separating F2P games from each analysis?


Did not see it addressed in this follow up: How did you treat games with multiple App IDs? Examples: Civ IV, Betas for multiple games.

Civ IV
http://steamdb.info/app/8800/
http://steamdb.info/app/34460/

Betas:
http://steamdb.info/app/223530/
http://steamdb.info/app/219540/

Not sure how I got opted into the Betas to show up, but they certainly would count as a registered unplayed game by default.

If the beta and the real game have separate app ids, they are tracked separately in our stats. That said, it seems a lot of these betas are removed from the Steam Community pages once the actual game comes out -- my sample didn't register a single person owning a game with the two beta app IDs you listed, for instance. So Steam seems to self-correct for this kind of thing, to some extent.

Edit: Actually, I take that back -- these betas do show up in our scraping, but the system treats those old betas as "inactive" games because they don't have an active Steam page, meaning they were temporary or have since been pulled from the store. Those inactive games are simply stripped out of the data. In the future I may go back through these titles and cross-reference with SteamDB to see what games they were, but they are not a significant part of the aggregate in any case.

 

[Sufinsil]

You must not have hit my profile.

Civ 4 + expansions is the only game I have that has 2 app ids for the same exact name. I think I got a 2nd appid from the Humble Bundle? I forget when it happened.

Company of Heroes has the standard and Steam matchmaking version listed. But all Steam owners got the new version.


Ah, that could be the case for the beta. I have not played recently any TF2, L4D2, and the few other games I have a beta version for, which show up on my Community page. Rare enough to not really effect the data really.

 

[Eisenhower]

Bleeprz"I'm on IE11 on WP8.1.

I'm so sorry.....

 

[kwazycoolkid88]

Schwanz[/url]":3dmub0t9]The DOTA2 being super popular thing baffles me. I ran it up a couple of times, and I can't even figure out WTF is going on in that game. If I could delete it from my steam account, I would.

I was considering installing it, but figured I'd be in the same boat as you.

 

[Makuta]

I also think the "Genre" make up statistics would be interesting!

 

[Machtyn]

My $0.02

0.01 - That's interesting, the top 200 games have had 1,000,000 or more sales. It's still difficult to see, but I'm supposing the top 50 or top 20 games have had over 9,000,000 sales. However, if you're in the top 80% of sales, you are almost guaranteed 10,000 sales. And if you're an indie company of 1 or 2 people, that makes for a great payday (depending on how much time you put into it.) Of course, these simple numbers don't equate to the realities of game dev and what it takes to create even a top 80% game. (See that bottom 10%?)

0.02 - I'm disappointed to see Portal 2 did not do as well as Portal. Granted, Portal was an add-on with the Orange Box bundle when first released. I purchased the Orange Box specifically for TF2 and was VERY pleasantly surprised with Portal. Portal 2, while not as innovative a single player game as Portal, did add some nice innovation to the multiplayer. Also, Portal 2's story is much better and more in depth than the first. (Both situations, Portal being more innovative and Portal 2 having better story, are natural progressions for a game.)

0.03 ooh, you get an extra penny! Speaking of Valve bundles, I know I have at least 2 copies of the entire Valve library since the last Valve bundle released with Half-Life 2: Ep2. Yes, that was quite a bit of time ago. But that should also clue one in as to how often I purchase Valve Software packed bundles any more. I generally pick a few games or less a year and enjoy them for quite a long time. (Skyrim currently, I still need to finish Fallout: New Vegas, get my money's worth out of Civ5 which seems broken for me, and put more time into the Battlefield game (WW2 flight sim) and RIFT, a free-to-play MMO that is a really good alternative to WoW.)

 

[fxds]

star skies[/url]":20usr3ur]Wow, Deus Ex: Human Revolution isn't even on there? Not even a million players through Steam? I'm finding it hard to believe that FTL: Faster Than Light did so much better, as much as I love that space sim.

Apparently the Director's Cut version of DE:HR made some questionable aesthetic changes and added more serious bugs than it fixed. And apparently Eidos has no intention of putting any more work into fixes.

At least, that's what I took away from the Steam forum discussions about the game. I suspect I'm not the only one who chose not to purchase it because of that.

 

[pappypappy]

Machtyn[/url]":2ec6wrly]My $0.02

0.01 - That's interesting, the top 200 games have had 1,000,000 or more sales. It's still difficult to see, but I'm supposing the top 50 or top 20 games have had over 9,000,000 sales. However, if you're in the top 80% of sales, you are almost guaranteed 10,000 sales. And if you're an indie company of 1 or 2 people, that makes for a great payday (depending on how much time you put into it.) Of course, these simple numbers don't equate to the realities of game dev and what it takes to create even a top 80% game. (See that bottom 10%?)

0.02 - I'm disappointed to see Portal 2 did not do as well as Portal. Granted, Portal was an add-on with the Orange Box bundle when first released. I purchased the Orange Box specifically for TF2 and was VERY pleasantly surprised with Portal. Portal 2, while not as innovative a single player game as Portal, did add some nice innovation to the multiplayer. Also, Portal 2's story is much better and more in depth than the first. (Both situations, Portal being more innovative and Portal 2 having better story, are natural progressions for a game.)

0.03 ooh, you get an extra penny! Speaking of Valve bundles, I know I have at least 2 copies of the entire Valve library since the last Valve bundle released with Half-Life 2: Ep2. Yes, that was quite a bit of time ago. But that should also clue one in as to how often I purchase Valve Software packed bundles any more. I generally pick a few games or less a year and enjoy them for quite a long time. (Skyrim currently, I still need to finish Fallout: New Vegas, get my money's worth out of Civ5 which seems broken for me, and put more time into the Battlefield game (WW2 flight sim) and RIFT, a free-to-play MMO that is a really good alternative to WoW.)

It would be interesting to see a revenue chart (although probably not possible) rather than sales. If you are an indie with 2 staff then 10000 sales seems a lot until you realise that Steam is so very price sensitive that they might mostly be at $5 or $10 a pop, at which point a year's work isn't all that much of a payday at all, after Steam takes their cut...

 

[Flit]

Here's something I'd like to see with your programming kung-fu.

For every game on the profile, check to see if there are achievements for that game. If so, how many people have at least one achievement, and also, how many people have all achievements.

If they have no achievements, we might use that data to narrow the "never played" metric with another datapoint, since most games played (more than just launching the game for Cards or whatever) have an achievement fairly early in the game.

 

[Munin]

One thing that I would have been curious about is whether and/or how the addition of Steam cards impacted play time and the percentage of games that have been "played". I do know that they did encourage several people I know to idle in games so that they could get them.

 

[fxds]

vartec[/url]":2pcsnqpb]

Pantagruel[/url]":2pcsnqpb]I personally think the "games bought but never played" data from the other article is one of the most important take outs of all this. It seems Steam Sales create an opportunity for titles to generate revenue from audiences they don't even actually appeal to. I really can't decide if this is good or bad.

Well, there are games that I bought but haven't time to play, but that's not very typical. Actually I think most of owned-not-played come from bundles. I've bought quite a few bundles of 6-8 games just to play 2-3 of them.

Bundles surely explain some, but not all of it. And I definitely don't think the rest is explained by people just randomly deciding to act irrationally during sales and buy games that don't even appeal to them. For example, I just looked at my own Steam library. I own approximately 40 games, of which I have played only about 41%. So my unplayed percentage, at about 59%, is far higher than average. But only three of the games in my library came from a bundle, and I've certainly never decided to buy games that don't appeal to me just because they were on sale. What I have done is buy games that do appeal to me, that I'd like to play someday, even though I don't know when exactly that someday is going to come. For someone like myself, a working adult with more discretionary income than free time, that's a perfectly rational choice.

Obviously I don't have any demographic information for Steam's player base, but given all the stories on gaming and tech sites in recent years about studies documenting the spread of gaming into the mainstream adult market, I strongly suspect there are a lot of people in the same general situation as myself.

 

[Voo42]

Kyle Orland[/url]":4d0wealv]

robotic_tourist[/url]":4d0wealv]From the original article there were a lot of people commenting to ask why the Steam API wasn't used. Any more information on this?

I'm a bit of a hack programmer?

Seriously, when I started on this, I wasn't even aware that this info was available via API, in addition to the web site. By the time I found out about it it seemed relatively pointless to start over and change everything up... the data is the same either way, right?

Would've made it easier for you actually.

If anyone's interested, there's a very trivial minimal example in Python2 to actually get data from a steamuser - just wanted to experiment a bit with it, some of the other API calls also give interesting information.

Obviously for serious purposes using twisted or some other async IO solution would be vastly superior, but hey.

Spoiler

from __future__ import division
from __future__ import absolute_import
from __future__ import print_function
from __future__ import unicode_literals

from collections import namedtuple
import urllib2
import urllib
from contextlib import closing
import json

UserData = namedtuple('UserData', 'steamid games')
GameData = namedtuple('GameData', 'appid name playtime')

# get from http://steamcommunity.com/dev/apikey
key = get_key_from http://steamcommunity.com/dev/apikey


def get_data(interface, method, version, **additional_params):
    url = 'http://api.steampowered.com/{}/{}/v{:04d}/?'.format(interface, method, version)
    params = {
        'format': 'json',
        'key': key
    }
    params.update(additional_params)
    url = url + urllib.urlencode(params)
    try:
        with closing(urllib2.urlopen(url)) as stream:
            return stream.read()
    except urllib2.HTTPError:
        raise Exception("Invalid steamid.")

def get_owned_games(steamid):
    data = get_data('IPlayerService', 'GetOwnedGames', 1, steamid=steamid, include_appinfo='1', include_played_free_games='1')
    parsed = json.loads(data)
    games = [GameData(game['appid'], game['name'], game['playtime_forever']) for game in parsed['response']['games']]
    return UserData(steamid, games)

# random person for demonstration purposes
print(get_owned_games('76561197960435530'))

 

[AngelusInsomnus]

Sufinsil[/url]":34vvxv9p]You must not have hit my profile.

Civ 4 + expansions is the only game I have that has 2 app ids for the same exact name. I think I got a 2nd appid from the Humble Bundle? I forget when it happened.

Company of Heroes has the standard and Steam matchmaking version listed. But all Steam owners got the new version.


Ah, that could be the case for the beta. I have not played recently any TF2, L4D2, and the few other games I have a beta version for, which show up on my Community page. Rare enough to not really effect the data really.

For Civilization IV, one entry is for the Windows version, and the other is for the OS X version.

 

[Sufinsil]

That makes sense! Never noticed the Install button not showing up.

 

[artifex]

Machtyn[/url]":1y75qqtd]
0.02 - I'm disappointed to see Portal 2 did not do as well as Portal. Granted, Portal was an add-on with the Orange Box bundle when first released. I purchased the Orange Box specifically for TF2 and was VERY pleasantly surprised with Portal. Portal 2, while not as innovative a single player game as Portal, did add some nice innovation to the multiplayer. Also, Portal 2's story is much better and more in depth than the first. (Both situations, Portal being more innovative and Portal 2 having better story, are natural progressions for a game.)

When the Steam client for OS X first came out, they gave Portal away as an incentive to OS X users. (That's what first got me to install Steam, in fact, so the ploy worked.) I'm sure plenty of Windows-only users managed to borrow the use of a Mac long enough to qualify, too.

 

[Vala]

pqr[/url]":1e604kvf]By the way, small request. Could you post most hours played for those top 100 games? Would tell us how devoted/addicted/loving we are compared to most devoted/addicted/loving person for each game.

This kind of sampling and extrapolating doesn't lend itself well to finding outliers in the full data set, so even if they did post the "most" hours played, there'd be no way to know if it was even close to accurate. Basically, without checking everyone's play time for a game, how will you know one of the ones you haven't checked yet doesn't beat your current highest value by orders of magnitude?

vartec[/url]":1e604kvf]

AstroGrover[/url]":1e604kvf]What's with the ones where 100% of all owners have played it? You can't tell me there's not a single person that hasn't played their copy of TF2.

The ones that have 100% are F2P, so I guess Steam only counts people who actually play them.

TF2 wasn't free to play though - I bought it on release way back when. And it was sold as part of the orange box (among others?) which would make me expect there to be a fair few people that owned it before it was F2P that never played it. That might have been before 2009 and not part of the stats though, I'm not sure.