Read: Survarium Update 0.53 Preview

Discussion of the latest news and events related to Survarium.

14.07.2018, 20:55

chicken_feet wrote:How many months until it will be ready?


The current goal is for implementation with 0.54. However, plans can change. It all depends how things go.
User avatar
joewillburn
Vostok Games
 
Posts: 6578
Joined: 03.04.2013, 11:25

14.07.2018, 20:57

I wonder if we're getting rid of the jumping in air mechanics. And maybe even vaulting animations in 0.56 or something? :P
Last edited by Chandrian on 14.07.2018, 21:02, edited 1 time in total.
User avatar
Chandrian
Supermoderator
Survivor
 
Posts: 3213
Joined: 03.04.2013, 18:33
Location: Spain

14.07.2018, 22:28

joewillburn:

The current goal is for implementation with 0.54. However, plans can change. It all depends how things go.

So probably October-December of this year, given previous time between version patches?

Chandrian:I wonder if we're getting rid of the jumping in air mechanics. And maybe even vaulting animations in 0.56 or something? :P

I think climbing ladders would be a good addition ;)
User avatar
chicken_feet
 
Posts: 5754
Joined: 22.02.2014, 18:39
Location: skeetskeetskeet

14.07.2018, 22:41

chicken_feet:I think climbing ladders would be a good addition ;)

What ladders xD
User avatar
Chandrian
Supermoderator
Survivor
 
Posts: 3213
Joined: 03.04.2013, 18:33
Location: Spain

14.07.2018, 23:18

joewillburn:

As I said, the ELO isn't working as well as it could be at this time. The fact there are multiple tiers, with multiple levels of XP and multiple ELO ratings all sitting in the queue at once it is not possible at this time for the ELO to come into proper effect. ...


Sorry, i must be a noob (not sarcasm, i can admit that i'm not good enough in something), but i don't understand that multiple tiers thing... That's okay there are 5 tiers now, but how is it affecting the matchmaking system? :O Sorry, but all the players in a match are searching games for the same tier... So, as i brought some exact examples, in tier 5... Who cares about players searching in another tiers (MM shouldn't!)? :(
I brought an existing match (match ID '11521514' if you want to dig yourself into it deeper), where there were enough player in the searching queue (MM could start a 8v8 game), but the MM ultimately FAILED to balance the game, because the only data was the ELO what MM used.
I made some exact calculations (some comments above) that shows You that the game's balance gave ALL the favour to the "team B" side (but MM thought that "team A" is stronger, because the higher ELO value).

And You said it yourself above: "with multiple levels of XP"!
XP is one of the most important thing in this (and any other) game! But the MM system avoids that important value!

A new player can reach high ELO very fast in the lower tiers (among low level, unexperienced players)! I had 3150 ELO on that match with my level 67 and with less(!) than 5 days ingame experience. лека had 2999 ELO on that match with level 100 and with 81(!) days ingame time.
So, if you are correct, i am a much more better player than him. This is the truth. The truth, based on ELO.
But sorry to say that, this truth must be a joke...

So, you asked (not from me) for a "better solution" to forward that to the Developers.

Here are mines:
Matchmaking system MUST calculate WITH the followings at the same time:
- ELO (elo is just the basic elo)
- LEVELS OF XP (as You mentioned absolute correctly!) (overall XP/10,000,000+1)

and if You want it more accurate, add the followings into the mixture:
- squads (if you are in 2-member squad, your ELO should get a multiplier of an exact 1.2 for example, if you are in a 3-member squad, you should get a 1.3 multiplier)
- average score / game (score/1000+1)

I give you an exact example of my opinion (all the datas are existing, exact values, You can find it at: https://survarium.pro/players/_ChE4ty_Br3Eki_)
My ELO is 3315 now. My main account is at level 94 (with 4,421,901 experience points). For the example, i'm in a squad with my 2 friends. My average score / game is 454.
The MM says now that i am a 3315 ELO player.

After my amendments "i become" a player like this:

basic ELO = 3315.000
XP multiplier: 1.442
squad multiplier = 1.300
avg. score multiplier: 1.454
in total: 3315*1.442*1.300*1.454 = 9,035.591 (without decimals: 9,036)

And with a non-existing fictional example:
Somebody with 2695 ELO, level 100 (6,810,030 xp), in a 2-member squad and with 357 avg. score / game.
in total he will get: 2695*1.681*1.200*1.357 = 7,377.
He should be a more experienced player than me, but with slower reflexes for example, and he has only 1 squad member (i had 2 in my example)...

And an existing example (from the real match mentioned above):
ELO 800, level 53 (232,717 xp), wasn't in a squad (1.000 "multiplier"), 91 average score...

800*1.023*1.000*1.091 = 893.

And yes, sorry to say that, without any offense, i am a 10 times better player than that guy... So, my "new" 9,036 ELO is very accurate compared to his "new" 893 ELO!

OR, leave squad and score multipliers behind... just calculate with the ELO and the XP points.
example 1 (myself): 3,315*1.442 = 4,780
example 2 (fictional): 2,695*1.681 = 4.530
example 2 (real guy): 800*1.023 = 818

Both of my calculations result a more accurate upshot!

Both of the calculation shows more REALISTIC difference between an experienced (high level) player and a rookie (low level) player, and between a good shooter and a bad shooter!

But now, the MM says that a level 50 (around 180k xp needed) player with 2500 ELO is AS GOOD AS a level 100 (exactly 6,810k xp needed) player with 2500 ELO...
And, sorry again, they are REALLY NOT as good as each other...

As "my system" will say:
2,500*1.018= 2,545 ELO
compared to
2,500*1,681= 4,203 ELO

And yes! It shows the difference between their ingame experience more accurate! Both of them are able to play in a tier 5 match, but surely they are not equal!

EDIT: And thank Dima, Ivan and You for the answers, and be sure guys i want this game to be a great game, that's why i spent more than 3x the Far Cry 5 Gold Edition's price on Survarium!
Last edited by _ChE4ty_Br3Eki_ on 14.07.2018, 23:40, edited 6 times in total.
User avatar
_ChE4ty_Br3Eki_
 
Posts: 64
Joined: 18.05.2014, 12:45
Location: [DREAM] [MEK] [Bier]

14.07.2018, 23:44

*grabs popcorn*
User avatar
Niruxu
 
Posts: 138
Joined: 03.04.2013, 18:43
Location: Germany, Dortmund, North Rhine-Westphalia

14.07.2018, 23:57

_ChE4ty_Br3Eki_ wrote: That's okay there are 5 tiers now, but how is it affecting the matchmaking system?

Currently match maker must generate matches across all 5 tiers. And if low-skilled player wants to play on tier 5 matchmaker will have to put him in some match. As result we end up having matches like the one you've described - a match with both low-skilled players (elo below 1000) as well as high skilled players (elo over 3000).

One tier would allow us to create "virtual" tiers instead - limiting max ELO difference in match will lead to matches where all players have roughly the same skill.
- LEVELS OF XP (as You mentioned absolute correctly!) (overall XP/10,000,000+1)

Using XP as basis for matchmaking is not a good idea. Experienced player who created a new character (or simply experienced FPS player who came to Survarium) will have almost 100% winrate until he finally reaches high level. And in the same time player who simply played huge amount of time without having good skills will stuck at lose streak simply because he reached high level and now game places him against high-elo players.

So while I can't say that current situation is perfect introducing another variable in form of level or k/d won't help much.

ivan_vg
Vostok Games
 
Posts: 329
Joined: 24.10.2016, 10:24

15.07.2018, 01:14

ivan_vg, so what happens in situations where a player from NA wants to play a match, and only on NA server, in this new one-tier system?

I specify NA only because it is historically the least played server. Perhaps the SEA server is less. Pertinent is that we are talking about a server that has very finite amounts of active players searching for a match at a given time.

If there is not enough players to balance this "virtual" tiers for ELO, we still suffer the same problem: low ELO and high ELO players will intermix in matches.

So does a one tier solution not solve this problem? It does with enough players, but it seems like one of the big problems for Survarium currently is balancing mechanics around a player population that may be stable but is too low to support the game currently. Perhaps less players than we had in previous years, as it wasn't as apparent a problem in previous years, to me, especially after Survarium entered open beta testing in early 2015.

Is the one tier solution good only for servers and game modes that are actively supported by the playerbase? Will we see the same ELO intermix problem at times of day when player numbers are fewest?
User avatar
chicken_feet
 
Posts: 5754
Joined: 22.02.2014, 18:39
Location: skeetskeetskeet

15.07.2018, 19:20

joewillburn:I quote "Better Idea"

RiotOfDoom:As I tried to propose: It better should judge by k/d instead of by how often you won a match being first, second, third etc.

Read carefully. I know it sounds too simple and too obvious. It didn´t require me to be an expert in coding. It´s simple gamer experience.
Did Dima and Co. already try to introduce MM balance that way? I don´t remember that to be since 2013. Maybe it was tried in 2012? I can´t recall that.

Btw. you should use a steady email adress. I remember having had three email conversation with you and every time you had a different email adress.
Also the Vostok Games company page lacks full information on their postal coordinates and the list of their staff with their names and their purpose what doesn´t make it look legal.
Here´s an example of how to do it right: https://imgur.com/Mq5p30N
That´s what serious investors/ customers look for at first
Even the co-developer of FTW "Focus Home Interactive" has done it right: https://imgur.com/zkMb6YY
Last edited by RiotOfDoom on 15.07.2018, 19:38, edited 1 time in total.
User avatar
RiotOfDoom
Survivor
 
Posts: 577
Joined: 11.05.2013, 14:32
Location: Phobos

16.07.2018, 10:52

RiotOfDoom wrote:Even the co-developer


They're a publisher not developer.

RiotOfDoom wrote:lso the Vostok Games company page lacks full information


We're in the process of a redesign of the corporate website.
User avatar
joewillburn
Vostok Games
 
Posts: 6578
Joined: 03.04.2013, 11:25

16.07.2018, 16:14

ivan_vg:or k/d won't help much.

There´s always a new patch or PTS that can help test that. Why don´t you just try and hear out the feedback?
User avatar
RiotOfDoom
Survivor
 
Posts: 577
Joined: 11.05.2013, 14:32
Location: Phobos

16.07.2018, 18:56

Let me explain how current matchmaking is working:
1) players in squads are taken into account with a factor more than 1 to compensate for their possible better teamwork
2) matchmaker tries to balance teams' sum of square roots of their Elo ratings for a particular game mode

What issues we see with it:
1) the same Elo rating is used for every equipment level
2) points scored on bots are not differentiated from the points scored on actual players and therefore distort "real" Elo rating
3) the players' skill points are not taken into account
4) squad doesn't always mean a better cooperation

What we propose:
1) balance weapons to be able to provide matchmaking at a single level to eliminate the issue #1
2) treat points on bots and players differently to eliminate the issue #2
3) due to a single matchmaking level we expect to not have the issue #3: the players' skill points will be accounted in player Elo rating. I mean if a player has less shooting skills but more game skill points this may counterweight the other player more shooting skills with less game skill points and they may have similar Elo rating and have similar game performance and therefore can be gathered into the very same match
4) due to a single matchmaking level we may restrict matchmaking to not create matches where players have a too big difference in Elo rating
User avatar
dima
 
Posts: 2550
Joined: 03.04.2013, 09:52
Location: Kyiv

16.07.2018, 19:26

dima:L
3) due to a single matchmaking level we expect to not have the issue #3: the players' skill points will be accounted in player Elo rating. I mean if a player has less shooting skills but more game skill points this may counterweight the other player more shooting skills with less game skill points and they may have similar Elo rating and have similar game performance and therefore can be gathered into the very same match


Interesting... are you able to go into more depth? What does game skill points mean? The Physical skill tree and the anomaly skill tree? You are comparing two skill trees against one. For higher level players, that will almost always be more game skill if true. (Even if a player has all the slots in the shooting skill tree filled in, past 75 levels, they're going to have way more into the game skill trees.
User avatar
VioletRose
 
Posts: 469
Joined: 22.04.2014, 15:44
Location: Yannov Station

16.07.2018, 20:05

dima:the same Elo rating is used for every equipment level

That, again means re-balancing of every item, what costs valuable time of development. Some time ago I proposed an introduction of a field which shows how much players play on which tier and which server.
This may have made it easier for players to find a match. They only would´ve had to switch to the required equipment level and click find match

dima:skill points

Is that a physical number resulting from overall gathered points like XP, k/d, headshots, wins/losses, or is it something else?

dima:due to a single matchmaking level we may restrict matchmaking to not create matches where players have a too big difference in Elo rating

That, again, depends on how much players with similar elo´re searching for a match atm ofc.

dima:I mean if a player has less shooting skills but more game skill points this may counterweight the other player more shooting skills with less game skill points and they may have similar Elo rating and have similar game performance and therefore can be gathered into the very same match

Please don´t forget that players that´re skilled shooters´re also skilled in completing game tasks. The shooting in Survarium´s different to other games and it takes players some time to adapt. But: Players with good shooting skill´re experienced players at the same time and they automatically have the knowledge about how to manage priorities to win a game.

Therefore they always will have a potential advantage over the player that´ll try to complete the game-tasks to win the game.
So if you have one player on your team that knows what to do but lacks shooting skill, while on the other team you have two skilled shooters, the game result´s easily predictable. That will not work, if the opponent´s playing purely for kills ofc.

dima:due to a single matchmaking level we may restrict matchmaking to not create matches where players have a too big difference in Elo rating

Yes. That´ll slightly decrease the chance for such a possibility, but again, that all depends on the number of players that have similar settings for a matchmaking like you atm of search. I didn´t encounter more than 200 players per server yet. If you´d have about 1000 players per server or even more, then this system´d definitely work as you predicted
User avatar
RiotOfDoom
Survivor
 
Posts: 577
Joined: 11.05.2013, 14:32
Location: Phobos

17.07.2018, 00:42

dima:What issues we see with it:
2) points scored on bots are not differentiated from the points scored on actual players and therefore distort "real" Elo rating

What we propose:
2) treat points on bots and players differently to eliminate the issue #2

I tend to come back to Alliance of Valiant Arms for my examples but it's just that game had so many things done correctly. It had amazing russian dub, good lobby browser to find a match quickly even with a low population, good weapon feel, nice maps and lots of modes and also only 1 tier and good matchmaking.

In the Team Deathmatch mode (called Annihilation, with [AI] if the map has +bots variants), which I think was the only mode that had bots in that game, you wouldn't get kills for killing bots and your deaths wouldn't count for dying against them... but their TDM used a score system instead of tickets that run out with each death. So, for each killed bot or enemy you would get +1 point and when a team reached 300 or so (the score to reach was dynamic and depended on the number of players in the match), the match resulted in a win. The bots weren't even on the scoreboard, they just were there to fill the place and get some action going if the match didn't have many players, so even a 2v2 + bots felt alright.
They also had a dog-tags sytem in place, each dead soldier (enemy or teammate) would drop them and if you picked up 3 without dying that would be +1 point towards the goal.

Consider this when looking for solutions for this "bots affecting ELO" issue.
Last edited by Chandrian on 17.07.2018, 00:53, edited 5 times in total.
User avatar
Chandrian
Supermoderator
Survivor
 
Posts: 3213
Joined: 03.04.2013, 18:33
Location: Spain

17.07.2018, 12:01

RiotOfDoom wrote:again means re-balancing of every item


Work on that has been and is being done.
User avatar
joewillburn
Vostok Games
 
Posts: 6578
Joined: 03.04.2013, 11:25

17.07.2018, 19:58

ivan_vg:

- LEVELS OF XP (as You mentioned absolute correctly!) (overall XP/10,000,000+1)
Using XP as basis for matchmaking is not a good idea. Experienced player who created a new character (or simply experienced FPS player who came to Survarium) will have almost 100% winrate until he finally reaches high level. And in the same time player who simply played huge amount of time without having good skills will stuck at lose streak simply because he reached high level and now game places him against high-elo players.


Thanks for the extensive answer! You may be correct, but with the ELO, this is the same... Second accounts and/or FPS experienced players will get low WLO in the first, with 100% winrate. Like mine secondery (Steam version) account. I had 100% winrate for weeks! So, the ELO is not answering this "issue".

Now, i played a game again, where Team A got 50-80 lvl players and Team B got level 100 players again. Okay, The ELO was (almost) equal, but lol, the game was already decided at the beginning. The levels of players in team A were 27-31-65-68-78-79-88-91, and in team B were 17-24-31-55-80-88-100-100.
And i became the 1st in team A with my lvl 68 account... But we had no chance, and i know, the playerbase is separated into tiers, game modes, etc. It seems to me, ELO is not everything, and never will.
As i mentioned, second accounts will have low elo too, like low level or "level of xp". And a level 50 player with 2500 ELO is surely not as good as a level 100 player with 2500 ELO.

But, i hope, I AM NOT RIGHT, and you are! That will be fantastic, because now matches are really unbalanced. Winning or losing 100%-0% in research, no great 76%-71% "close" matches. If i win, i win flawlessly, if i loose, i will loose as a dog against an armoured division...
But, i hope!
Last edited by _ChE4ty_Br3Eki_ on 17.07.2018, 20:08, edited 1 time in total.
User avatar
_ChE4ty_Br3Eki_
 
Posts: 64
Joined: 18.05.2014, 12:45
Location: [DREAM] [MEK] [Bier]

18.07.2018, 11:12

_ChE4ty_Br3Eki_: And a level 50 player with 2500 ELO is surely not as good as a level 100 player with 2500 ELO.

It should be, but now we can pump ELO on bots...

_ChE4ty_Br3Eki_: The levels of players in team A were 27-31-65-68-78-79-88-91, and in team B were 17-24-31-55-80-88-100-100.
And i became the 1st in team A with my lvl 68 account...

So you (68 lvl) was better than someone 91 lvl - that shows how inaccurate is lvl based MM.
Elo adapt after few games, lvl don't adapt at all.
User avatar
Chrno
 
Posts: 551
Joined: 03.04.2013, 19:02

18.07.2018, 12:11

As I mentioned before, ELO is adapting too slow to current player's skill. I saw it many times in recent league, where guys with 1200-1300 ELO had the best score and easily destroying enemy team.
Last edited by Tanatloc on 18.07.2018, 12:12, edited 1 time in total.
User avatar
Tanatloc
 
Posts: 896
Joined: 08.03.2016, 17:01

18.07.2018, 15:15

Chrno:
It should be, but now we can pump ELO on bots...

Yes, thats why ELO is not accurate. Low levels jumping up their ELO with the speed of light in tier 1-2-3, and surprisingly they are become tier 5, and playing t5 with 2800 ELO and level 48... Against players with 2800 ELO and level 100. Level 100 will be the winner in that comparison.

Chrno:So you (68 lvl) was better than someone 91 lvl - that shows how inaccurate is lvl based MM.
Elo adapt after few games, lvl don't adapt at all.

Yes, i was, because that is my secondary (Steam version) account. My main account is lvl 94 (with very similar stats to my lvl 68 acc., for example 500 avg. score/game.).
And yes, you are correct, MM based on only level (level of experience) will be not accurate!
That"s why i told ELO+XP are must be better... Maybe ELO+XP+winrate, for example. But, as i told in my last comment, i HOPE that Dima, Ivan, and the others are right! And i'm not!
Last edited by _ChE4ty_Br3Eki_ on 18.07.2018, 15:30, edited 3 times in total.
User avatar
_ChE4ty_Br3Eki_
 
Posts: 64
Joined: 18.05.2014, 12:45
Location: [DREAM] [MEK] [Bier]

In order to reply you must be signed in.

Return to News