A20 Developer Diary Discussions

[MonkCrimson]

On 12/28/2020 at 10:38 PM, n2n1 said:

fatal, please pay attention to the random number generator algorithms.

Because of its "stuck", it often happens that events that depend on randomness are repeated.

(or a set of loot in the same types containers, if they are robbed in a row - is repeated).

 

 

 

 

 

 

Yep i agree , that happens to me . Sometimes i get the same loot at multiple containers and i'm not playing with the loot on 100%.

[Bertlor]

Could be a long shot but regarding the "randomization" seeming to be a pattern of sort in looting the houses / reward system in multiplayer... could it be as simple as the RNG wasn't initialized? I'm not too familiar with the Unity editor/programming, however, I have experience in the past with other languages that when calling for a random number it may appear random until you realize it's a preset pattern each time the application was loaded.

 

For example... launch the application and call the randomized function 5 times which may return; 7, 3, 18, 6, and 1. It appears to be random at first, however, you launch the application a 2nd time and call the same function 5 times again resulting in; 7, 3, 18, 6, and 1. I then found the randomize function needed declared/initialized beforehand to truly be random. Once this happened then calling the randomize function 5 times would return, 8, 3, 16, 7, and 2. The next time the application was launched the same process would then yield; 13, 6, 9, 20, and 3 (for example).

 

I bring this all up because I've kicked myself after realizing it was as simple as adding Randomize() before using Rnd() to get my expected results/behavior;

' Initialize RNG
Randomize()

' Generate random value between 1 and 6.
Dim value As Integer = CInt(Int((6 * Rnd()) + 1))

[Ricowan]

Fatal already replied to this with "Now the seed used for the random generator could cause issues if not varied enough or reused/repeated."  The "seed" is what is used to change where in the pre-generated list of random numbers the game starts, when picking new random numbers.  The seed probably just needs to be changed more often, and it sounds like the devs are looking into that.

[meganoth]

48 minutes ago, Ricowan said:

Fatal already replied to this with "Now the seed used for the random generator could cause issues if not varied enough or reused/repeated."  The "seed" is what is used to change where in the pre-generated list of random numbers the game starts, when picking new random numbers.  The seed probably just needs to be changed more often, and it sounds like the devs are looking into that.

 

The seed does not need to be changed at all, it just needs to be sufficiently random when you start the game. This is why date+time plus timings of user input and mouse movements are often combined to get a good seed.

 

To my knowledge Fataal did not inditacte anywhere that they are looking into that. Until now nobody could show them that a problem really exists.

 

Players often fall into the psychological trap of seeing patterns that they deem non-random, but the occurence of patterns IS truely a sign of randomness. If you throw 5 dice and get all 6's, then it doesn't tell you the dice are faulty. But if you threw those dice a million times and it would never show that pattern, THAT would indicate they are not random.

 

Look into the forum of every RPG or wargame and you will find people who complain about the RNG not being random because they saw a spectacular series of good or bad events. The problem is that when 1 million players play a game, there is a good chance that one of them will see a one-in-a-million event

Edited January 4, 2021 by meganoth (see edit history)

[Blake_]

2 hours ago, meganoth said:

 

The seed does not need to be changed at all, it just needs to be sufficiently random when you start the game. This is why date+time plus timings of user input and mouse movements are often combined to get a good seed.

 

Interesting explanation.

 

[meganoth]

5 minutes ago, Blake_ said:

Interesting explanation.

 

FAKE! A very deterministic robot. Show it a bone and it will immediately wag its tail.

 

 

[n2n1]

Just by the way:

I can give an example, when in a discussion about the distribution of biomes i was shown a map.

https://community.7daystodie.com/topic/22253-classic-style-hardcore-mod/?do=findComment&comment=414735

 

To check (without knowing seed) i generated any two maps.... and one of them - was exactly the same as i was shown....

https://community.7daystodie.com/topic/22253-classic-style-hardcore-mod/?do=findComment&comment=414901

 

A coincidence? I do not think so.

 

 

I just want to repeat once again - it doesn't matter how the algorithms work, whether they are a real random or a masked sequence of numbers.... good - this is when her work is not thrown in the eye. And that the player would not pay attention to it. And that the player would not pay attention to it.

Edited January 4, 2021 by n2n1 (see edit history)

[meganoth]

52 minutes ago, n2n1 said:

Just by the way:

I can give an example, when in a discussion about the distribution of biomes i was shown a map.

https://community.7daystodie.com/topic/22253-classic-style-hardcore-mod/?do=findComment&comment=414735

 

To check (without knowing seed) i generated any two maps.... and one of them - was exactly the same as i was shown....

https://community.7daystodie.com/topic/22253-classic-style-hardcore-mod/?do=findComment&comment=414901

 

A coincidence? I do not think so.

 

 

I just want to repeat once again - it doesn't matter how the algorithms work, whether they are a real random or a masked sequence of numbers.... good - this is when her work is not thrown in the eye. And that the player would not pay attention to it. And that the player would not pay attention to it.

 

This is not a good example. RWG world generation is done very different from loot randomization.

 

For starters the biome map is not generated, it is AFAIK one of a few preset maps. I don't know how many, at one time there was even only one. So it is very very probable that you will see the same biome map when generating multiple worlds.

 

Generally, when generating worlds, YOU set the seed for world generation. The name you give the world is the seed. Another seed you set is used for grass, plants and stone distribution. These seeds are definitely not random numbers and deliberately so.

 

Please correct me if I somehow misunderstood you.

 

 

 

 

Edited January 4, 2021 by meganoth (see edit history)

[n2n1]

21 minutes ago, meganoth said:

 

THIS is not a good example. RWG world generation is done very different from loot randomization.

 

For starters the biome map is not generated, it is AFAIK one of a few preset maps. I don't know how many, at one time there was even only one. So it is very very probable that you will see the same biome map when generating multiple worlds.

 

Generally, when generating worlds, YOU set the seed for world generation. The name you give the world is the seed. Another seed you set is used for grass, plants and stone distribution. These seeds are definitely not random numbers and deliberately so.

 

Please correct me if I somehow misunderstood you.

 

 

 

 

No - it's not. I can change the ratio, size, and shape of biomes for generation - so this is definitely not a presets. What You're talking about - was earlier.

But i agree that it doesn't have much to do with loot. I just wanted to show that now there is nothing random, and very frequent recurrence in everything.

(of course, there is a mod here, but we are talking about algorithms that lie deeper)

Edited January 5, 2021 by n2n1 (see edit history)

[Jost Amman]

1 hour ago, art1336 said:

Console did sell great...but then it got dropped and console players felt extremely shafted for paying money for a product that was never delivered.

I get what you're saying, and it's true that it "did" sell great before the Great Shafting, but it might seem callous to some to point that out,

Only if it was TFP's fault... 

[n2n1]

1 hour ago, meganoth said:

 

This is not a good example. RWG world generation is done very different from loot randomization.

 

For starters the biome map is not generated, it is AFAIK one of a few preset maps. I don't know how many, at one time there was even only one. So it is very very probable that you will see the same biome map when generating multiple worlds.

 

Generally, when generating worlds, YOU set the seed for world generation. The name you give the world is the seed. Another seed you set is used for grass, plants and stone distribution. These seeds are definitely not random numbers and deliberately so.

 

Please correct me if I somehow misunderstood you.

 

 

 

 

.... there is no point in having a conversation here - messages are transferred to another branch - the meaning of what was said is lost...

 

it turns out some nonsense...

 

If you(moderators) do it this way, then disallow any messages that are not addressed to developers so that conversations  between users do not start at all.

Edited January 5, 2021 by n2n1 (see edit history)

[khzmusik]

4 hours ago, n2n1 said:

Just by the way:

I can give an example, when in a discussion about the distribution of biomes i was shown a map.

https://community.7daystodie.com/topic/22253-classic-style-hardcore-mod/?do=findComment&comment=414735

 

To check (without knowing seed) i generated any two maps.... and one of them - was exactly the same as i was shown....

https://community.7daystodie.com/topic/22253-classic-style-hardcore-mod/?do=findComment&comment=414901

 

A coincidence? I do not think so.

 

 

I just want to repeat once again - it doesn't matter how the algorithms work, whether they are a real random or a masked sequence of numbers.... good - this is when her work is not thrown in the eye. And that the player would not pay attention to it. And that the player would not pay attention to it.

 

The biome placement example might not be a good one. I believe that biome placements in RWG aren't actually random. At least an alpha or two ago, they were the same biome pattern all the time, except perhaps rotated. I don't know if that's still true, but it might be, or they might have just added a few more pre-determined biome maps to choose from.

 

I personally agree that the issue is not randomness but perceived randomness. I think of it as the "playlist shuffle" problem. Many music apps shuffle a playlist by just choosing a random song in the playlist. But that means it's a near-certainty that many songs will be played multiple times, before all the songs in the playlist are played at least once.

 

What most people expect is "shuffling" like a deck of cards - they don't want the songs to be randomized, they want the song order to be randomized. If they hear the same song twice, they think the shuffle "isn't working."

 

That's certainly true of quest rewards. You can specify in XML that certain kinds of rewards are "fixed" (always appear), and other kinds have a limited supply to draw from. It's not particularly surprising that the same rewards would be offered to more than one player.

 

But IMHO "fixing" this is more trouble than it's worth. Like the "playlist shuffle" issue, 7D2D would somehow have to keep track of which items had already been given to someone, and make sure the next "random" item wouldn't match any of those. It would take time to program, would be easy to game, and probably wouldn't add much.

 

On the other hand, if it's as simple as re-seeding the RNG before each use - and/or not getting a shared seed from the server - then that would probably be worth it, since it's trivial. But I think it's probably not that simple.

[Roland]

2 hours ago, n2n1 said:

.... there is no point in having a conversation here - messages are transferred to another branch - the meaning of what was said is lost...

 

it turns out some nonsense...

 

If you(moderators) do it this way, then disallow any messages that are not addressed to developers so that conversations  between users do not start at all.

 

Sorry, I meant to move the entire conversation but missed some which made it confusing. It is all here now and you are welcome to carry it on. I am not going to dissallow messages that are not addressed to the developers but messages that are not addressed to the developers will be moved out.

[Zipcore]

There are some things missing for me which would make this game a lot better for me (at least) and propbaly others.

 

1. Rotating prefabs in world editor

- Replacing, deleting etc works but could be improved, but I have to rotate them manually each time using xml files and need to reload the game to take effect.

 

2. Ability to remove items using the API.cs

- Would be the seed for TONS of cool features me & the community could develop.

 

3. Extended drop on death setting: keep equipment only

 

I hope I'm not asking for 2 much but those additions would make me really happy.

In case those features somehow already exist I would like to have some info how to make them work.

Edited January 5, 2021 by Zipcore (see edit history)

[n2n1]

51 minutes ago, khzmusik said:

/////

I personally agree that the issue is not randomness but perceived randomness. I think of it as the "playlist shuffle" problem. Many music apps shuffle a playlist by just choosing a random song in the playlist. But that means it's a near-certainty that many songs will be played multiple times, before all the songs in the playlist are played at least once.

 

What most people expect is "shuffling" like a deck of cards - they don't want the songs to be randomized, they want the song order to be randomized. If they hear the same song twice, they think the shuffle "isn't working."

//////

 

Right.... Only in our case it turned out that the algorithm designed to solve "imaginary repetitions "generates "real repetitions".

I would like the developers to pay attention to this.

And what algorithm they will eventually choose-we should not care.  And what algorithm they will eventually choose-we should not care.

Edited January 5, 2021 by n2n1 (see edit history)

[meganoth]

5 hours ago, n2n1 said:

No - it's not. I can change the ratio, size, and shape of biomes for generation - so this is definitely not a presets. 

Where can you do that ?

 

5 hours ago, n2n1 said:

What You're talking about - was earlier.

But i agree that it doesn't have much to do with loot. I just wanted to show that now there is nothing random, and very frequent recurrence in everything.

(of course, there is a mod here, but we are talking about algorithms that lie deeper)

Mod? Are you talking about Nitrogen?

[Outlaw_187]

6 hours ago, art1336 said:

Console did sell great...but then it got dropped and console players felt extremely shafted for paying money for a product that was never delivered.

I get what you're saying, and it's true that it "did" sell great before the Great Shafting, but it might seem callous to some to point that out,

Careful, you're gonna get these PC boys fired up bringing that console sold so well. 

The PC version is by far superior however I started on ps4 & loved the game. I'm grateful it was released on console otherwise I may not have experienced this game as I'm not a pc gamer. 

 

On a side note, hopefully this post doesn't get moved, but do any of the devs know if there will be any additional experimental past 19.3? I actually haven't restarted since 19.1 and was considering restarting but I'll wait if there's gonna be a 19.4. 

Thanks Fun Pimps, love the game & look forward to Alpha 20!!

[n2n1]

18 minutes ago, meganoth said:

Where can you do that ?

 

Mod? Are you talking about Nitrogen?

In rwgmixer.xml. I use the very old code (similar the one that was in A16) - it still works.

And what you described - was in A17 (static map of biomes).

Of course mod, he did it based on the capabilities of the game itself. No third-party utilities and integrations are used.

Just take a look at the screenshots i made there.

Edited January 5, 2021 by n2n1 (see edit history)

[Zark]

Unfortunately, A20 might be the update where I decide to quit because the climate survival I once enjoyed and loved has been stripped away.

 

I loved having the amount of clothing slots because I like tweaking the settings to also make it a winter survival game. Having 4 slots will ruin this mechanic and the game for me. Ever since A16 or A17, the "weather snow 0.4" command that added a layer of snow depth no longer worked. That disappointed me because I enjoyed playing with the weather settings to add a unique level of challenge for my friends where there would be cold fronts and snow in the desert or pine.

 

Edited January 5, 2021 by Zark (see edit history)

[MechanicalLens]

2 minutes ago, Zark said:

Unfortunately, A20 might be the update where I decide to quit because the climate survival I once enjoyed and loved has been stripped away.

 

I loved having the amount of clothing slots because I like tweaking the settings to also make it a winter survival game. Having 4 slots will ruin this mechanic and the game for me. Ever since A16 or A17, the "weather snow 0.4" command that added a layer of snow depth no longer worked. That disappointed me because I enjoyed playing with the weather settings to add a unique level of challenge for my friends where there would be cold fronts and snow in the desert or pine.

 

 

No offense, but really, this is what would make you quit? To each their own, but it is obvious that your loyalty is lacking.

[Diragor]

42 minutes ago, MechanicalLens said:

 

No offense, but really, this is what would make you quit? To each their own, but it is obvious that your loyalty is lacking.

Well I must say he's not wrong about it. I loved that snowing too. It added variety to the world. Now you always see the same world with the same textures. I wish snowy days would come back. But I wouldn't quit cause of the lack of it^^ I want to see where the game goes

[Xtrakicking]

44 minutes ago, MechanicalLens said:

 

No offense, but really, this is what would make you quit? To each their own, but it is obvious that your loyalty is lacking.

Loyalty? Starting to sound like a cult to me. Lol.

 

I mean, I think it's perfectly reasonable that someone would quit if the devs take away what made the game fun for them, don't you think?

[MechanicalLens]

1 minute ago, Xtrakicking said:

Loyalty? Starting to sound like a cult to me. Lol.

 

I mean, I think it's perfectly reasonable that someone would quit if the devs take away what made the game fun for them, don't you think?

 

Nevertheless, quitting a game due to one simple change almost seems a bit petty to me. Perhaps there were other changes that they cannot currently recall that they were also not a fan of- a snowball effect if you will.

 

If you get married to any system in an alpha game, then you are just setting yourself up for disappointment.

[MechanicalLens]

MUAHAHAHAHAHAHA!

 

[meganoth]

7 hours ago, n2n1 said:

In rwgmixer.xml. I use the very old code (similar the one that was in A16) - it still works.

And what you described - was in A17 (static map of biomes).

Of course mod, he did it based on the capabilities of the game itself. No third-party utilities and integrations are used.

Just take a look at the screenshots i made there.

 

Ok. I have never played around with rwgmixer. I can't assess how much randomness would be in the old biome generation and how probable it was that you saw the same biome map.

 

1) Who knows, the generator might just select between 10 or 20 fixed versions of a biome map internally. Very unlikely since someone should have noticed this in A16 already.

2) Since all the rest of RWG is procedurally generated I would assume biome generation also uses the (user-provided) seed that is used for the rest of world generation. Are you sure that the other poster used a different name for WorldGenSeed AND GameName than you? Are you sure your biome does not depend on the WorldGenSeed or GameName?

3) A possibility is that your mod itself shrinked the possible biomes that can be generated by the biome generator.

4) Another possibility is that the old unused code that generates the biomes in your mod is buggy and removes all the randomness it would get from the random number generator.

 

I'm not saying you are wrong. This could be the basis of a proof, but at the moment this is just another weak indication there could be something wrong, with too much uncertainty what it really means. I'm sure you know more about this than I do and might be able to dismiss one or more of the possibilities I listed. But to dismiss all of them you would need access to the internal biome generator code and have analyzed it.

 

In summary, to prove it you have to give a lot more details and in this case only TFP internally could find out if this is a problem of its RNG in general or just some bug in old code that isn't used anymore. And how likely is it that they will try to follow any argument that leads to analyzing unused code that is probably on the way out of the code base? If you are really sure then you could make a bug report, but I suspect it likely will be ignored because of using old code.

 

If this were a murder investigation we would be at the point where we have a missing person that doesn't answer any calls. And we have a few suspects, one of them so old he will be dead before you can convict him. But neither do we have a dead body nor any proof of foul play at all. We need the dead body, not more people who tell us they didn't see him since yesterday. 😉

 

Edited January 5, 2021 by meganoth (see edit history)