Friday Facts #292 - Inching closer to stable

[BlueTemplar]

At Paradox they get some fairly detailed information about player behavior in game, Hearts of Iron 4 specifically comes to mind. They mention it occasionally in their dev diaries (here and here for example). I'm sure if you ask they'd give you some idea on where and how they get it, they have their own store/launcher thing so I suspect it isn't driven via steam, although I might be wrong.

Also that ray casting: just wow.

Lots of games have user statistics like that, however I've always been one against such things. Use my computer to render the game rather then send statistics. It would be nice if you could fully opt out from non-essential network activity and vice versa.

On the flipside of that I love local user data, would love to see a more robust stat screen for current game, and all time. E.g. total hand crafts, total x made, total trees caught on fire, etc. But this would most likely have it's own performance impact as well. Especially with such a dynamic game such as factorio.

Not to verge further off topic but I love their forum layouts, often wish to "upvote" post here rather then commenting.

Well, with GDPR, if I'm not mistaken, this kind of data gathering has to be opt-in. Certainly if a personal identifier is involved... (that would allow to count the number of unique players per version of the game - and even then, for accuracy Wube might actually need player-hours !)

Note that the data that Paradox shows is in the form of percentages : they probably don't know the exact number of players either, and like Wube, rely on opt-in information.

[Maddhawk]

Most people play through Steam, and so far we have found no way of determining how many people are opted-in to the 0.17 experimental through Steam. It could be that the game is more stable, or it could be that less people are playing.

How can you not know this ? is there any way we can help with this on our end ?

Allow me to help. I play Factorio through Steam and I am opted in to the experimental releases. To date, I have had 1 and only 1 crash and it was related to the mod Merging Chests. It crashed when I tried to make a 78x78 merged chest array for storing my excess uranium refining production.

Edit: Otherwise it has been rock solid stable for me every single iteration of 0.17.x so far.

Mods I use:
Concretexture 1.1.1
Flare Stack 2.2.3
Infinizoom 0.1.3
MegaBotStart 0.17.0
My_Tree_Mod 1.1.13
Power Armor MK3 0.2.6
Turret-Shields 0.17.35
WideChests 2.2.0 (Merged Chests mod)
base 0.17.33

I let Steam keep it auto-updated to the latest experimental version.

That said, there are some thing I haven't explored yet in any game to date. Rail and logistics networks. I have yet to build a train or a single logistics robot. I have played 3 different maps in 0.17 and launched a rocket in all of them. Currently expanding my base in my 3rd map and at 30+ launches so far. I have also completed the new campaign in its initial configuration after 0.17 released.

Other mods I have used, but currently not using include:
Creative Mode
BigBrother

Had no issues with them as well, other than that disabling BigBrother removed ALL of my radars and I had to run around and manually replace them. >>

[exi2163]

Klonan

This last weekend, we had over 500 total crashes reported, which is a slight improvement over the prior weekend's ~650. One thing that makes our progress hard to evaluate is that we don't know how many people are actually playing experimental. Most people play through Steam, and so far we have found no way of determining how many people are opted-in to the 0.17 experimental through Steam. It could be that the game is more stable, or it could be that less people are playing.

How about including the game version number in the user auth request if the submit-crash-report option is active?
This would yield an exact ratio for your statistics and include the possibility to opt-out if someone does not want that information to be sent.
Just make sure to communicate this change in big letters to avoid GDPR backlash.

[orzelek]

Klonan

This last weekend, we had over 500 total crashes reported, which is a slight improvement over the prior weekend's ~650. One thing that makes our progress hard to evaluate is that we don't know how many people are actually playing experimental. Most people play through Steam, and so far we have found no way of determining how many people are opted-in to the 0.17 experimental through Steam. It could be that the game is more stable, or it could be that less people are playing.

How about including the game version number in the user auth request if the submit-crash-report option is active?
This would yield an exact ratio for your statistics and include the possibility to opt-out if someone does not want that information to be sent.
Just make sure to communicate this change in big letters to avoid GDPR backlash.

I'm not sure if game version would be a GDPR thing. Version of game you are using doesn't identify you personally in any way.. but lawyer would be needed to confirm that.

[BlueTemplar]

If there's no identifying information, how would they be able to count ?

[fendy3002]

You're all lying. That amount of bug fix and patch, under illness wave and easter vacation? Please tell me that you're lying, please...

[bobingabout]

One thing about the multiplayer stress tests.

It's a good thing and all that, but you need to also do that sort of thing with modpacks, otherwise there's a whole heap of desyncs that could happen with mods that you're not going to find this way.

I think that is up to the mod developers to fix. I think a lot of those desyncs occur because of small errors in the mod.

in some cases yes, in other cases it's because there are features in the game engine, that aren't in the game itself. think of things like Loaders, Void or Fluid energy source, etc. Even scripting. There's a lot of modder requested scripting interfaces that aren't used by the base game.
Of course, if they're not used by the base game, the incentive for the developers to fix it is lower, because it affects fewer players, but either way, you're not going to find ALL bugs if your tests don't include all features.

So doing a test run like this that includes mods would be a good thing.

[Riktol]

One thing about the multiplayer stress tests.

It's a good thing and all that, but you need to also do that sort of thing with modpacks, otherwise there's a whole heap of desyncs that could happen with mods that you're not going to find this way.

I think that is up to the mod developers to fix. I think a lot of those desyncs occur because of small errors in the mod.

in some cases yes, in other cases it's because there are features in the game engine, that aren't in the game itself. think of things like Loaders, Void or Fluid energy source, etc. Even scripting. There's a lot of modder requested scripting interfaces that aren't used by the base game.
Of course, if they're not used by the base game, the incentive for the developers to fix it is lower, because it affects fewer players, but either way, you're not going to find ALL bugs if your tests don't include all features.

So doing a test run like this that includes mods would be a good thing.

The event is not run by the devs, they're just taking an interest because they can get useful information from it. So if you want to host a modded large scale multiplayer event and invite the devs then I'm sure they will give you a plug in the next FFF.

[exi2163]

If there's no identifying information, how would they be able to count ?

If a multiplayer server is configured to verify the user identity then the game client is connecting to the factorio auth server to get an authentication token. This token is presented to the server, its signature checked and the client allowed if it is ok. It is the only anti-piracy feature that is enabled in factorio (afaik).
To get the auth token the client supplies the auth server with your factorio username and password (So it does not have to be sent to the game server for obvious reasons). As almost every server uses this verification process this would cover all online multiplayer games.
The auth server can then put an entry into a database for the combination of date/ver/user to track that this user was active and later count all active users per version/date and correlate this to the automated debug logs database to crosscheck for crashs per version per active player.

If a debug report is coming in from a non-tracked user he either has only played single player or has no valid factorio account. As a debug report indicates the player has played and the version used is reported you stil can add this datapoint to the database.
If automatic debug uploading is disabled we don't care about the user because he is either developing mods and crashing constantly or not interested in helping the devs making a better game.

[BlueTemplar]

If a debug report is coming in from a non-tracked user he either has only played single player or has no valid factorio account. As a debug report indicates the player has played and the version used is reported you stil can add this datapoint to the database.

Yeah, this is what I mean - you have no way to know if each of these datapoints comes from separate users...

I guess that this issue is moot since most users are going to be registered (via Steam) ?
But then I don't understand why Wube can't get the information they need from their update server ?
(Which a registered user is generally going to ping on startup.)
Because then they still have no idea for how long that Factorio instance is going to stay open ?
(Well, until they receive a second ping...)

[<NO_NAME>]

I really think player AI in this game should be nerfed. The game is just broken right now, balance-wise. I mean, how this move with building 3D engine is even legal? How am I supposed to compete with that?

[Zavian]

I guess that this issue is moot since most users are going to be registered (via Steam) ?
But then I don't understand why Wube can't get the information they need from their update server ?
(Which a registered user is generally going to ping on startup.)

For the Steam version updates are handled via Steam, so there is no ingame updater, and hence no need to ping the update server.

[BlueTemplar]

Oh, right ! :facepalm:

[gaelyte]

Speaking of mods, it could be good if we were able to enable, disable, download and update some mods without having to load all the mods for nothing, especially with the 0.17 where the mods loading take one or two minutes and where the game freeze for an additional minute when you reload the mods

[Unknow0059]

This one was really uneventful. More so than usual when it comes to recent bug report posts.

What a shame. I really look forward to these. It's like my morning newspaper (except it's got good news) every Saturday (except i missed it this saturday).

Anyways, i can't complain. There's not much to say about bug fixing anyway. I guess you guys could tells us some of the more interesting bugs, but even then...

[Dominik]

One thing about the multiplayer stress tests.

It's a good thing and all that, but you need to also do that sort of thing with modpacks, otherwise there's a whole heap of desyncs that could happen with mods that you're not going to find this way.

While testing with mods have its purpose too, the aim of these tests was elsewhere. It was about handling the large numbers if input actions and data transfers - finding bottlenecks etc. Feature richness does not really affect that, so more simple and stable game was preferable to test it.

[bobingabout]

One thing about the multiplayer stress tests.

It's a good thing and all that, but you need to also do that sort of thing with modpacks, otherwise there's a whole heap of desyncs that could happen with mods that you're not going to find this way.

While testing with mods have its purpose too, the aim of these tests was elsewhere. It was about handling the large numbers if input actions and data transfers - finding bottlenecks etc. Feature richness does not really affect that, so more simple and stable game was preferable to test it.

And I do understand that. The base game being stable is a very important first step. Do that first, then look at the other features used by mods.

[Ormy]

Speaking of mods, it could be good if we were able to enable, disable, download and update some mods without having to load all the mods for nothing, especially with the 0.17 where the mods loading take one or two minutes and where the game freeze for an additional minute when you reload the mods

I can second this, great idea. Just being able to enable/disable already-installed mods without loading the whole game and all my mods (I use a lot of mods) would save lots of time.

If I sit down to play factorio, knowing last time I played I had my mods enabled but now I want to play a vanilla game, I still have to load all the mods (which can take 2 or 3 minutes) just to then disable them and load the game all over again. Being to disable/enable mods from a separate GUI outside the game itself would be amazing. Still having to load the full game to install and update mods would be a lesser inconvenience.

[boran_blok]

Just FYI, you can specify the used mod folder as a commandline argument to the game, for this I have multiple mod folders setup for different games I play. (One for Seablock, one for MultiPlayer, one for Creative, etc) each modfolder is completely separate from the others so you dont have to enable/disable mods all the time.

[NelsonSKA]

The Ray-casting game inside Factorio is very impressive, I saw the video in youtube and it's very interesant. I want to see how the project will look.