Matchmaking and Threat

[MattScott 15242]

Hi everyone,

 

I'd like to start by thanking all the players who threw out suggestions about topics they would like to discuss.

 

The point of this discussion is to peel the curtain back and attempt a two-way conversation with you guys. I'm going to make the disclaimer that while I strive to be as transparent as I can, there are some things I can't share for any number of reasons. If I point an area out that is off limits, please do not assume that I'm trying to hide some evil intention. My hope is that we can use this discussion to help inform my team to some of your concerns, needs, or wishes, and then we can try to take those into account as we move forward.

 

I saw a lot of comments on threat / matchmaking, so we'll try to tackle that one first.

 

For most of these, I'm going to start by giving my opinion/context on the subject. Again, I'm going to make the disclaimer that I am not an expert on APB. I have employees who are that design these systems. Instead, I am the final vote. So it's important that I be as current as possible with the game and the players.

 

The purpose of me exposing my knowledge (or lack there of) is so we can have a discussion on common ground.

If there is a concept that I'm missing, then I trust that you, the players, will point it out and fill me in.

 

So.. IMO:

 

1) Why is matchmaking so important in APB versus other games? 

APB is somewhat unique in requiring PvP conflict to progress through the game. Any time someone wins, someone also loses. In my opinion, there is nothing wrong with losing matches. That's how we learn. But in an ideal world, those matches would be close, so that both sides get an opportunity to learn and get better.

 

2) What are my requirements for good matchmaking?

- Matching based on skill that goes up or down each match based on an accurate assessment of how you performed as an individual and how you performed in your team

- Matching based on latency so that all players have roughly the same response time from the server
- Matching across the widest pool of players to narrow the difference in skill as much as possible

 

3) What are the problems I see in APB's current matchmaking?

- Due to Unreal 3's notoriously slow loading times, and the fact that "phasing" or switching to another instance of the same district, isn't supported in Unreal 3 - APB was designed to only match one player against another player inside the same district on the same server.

- Working off that limitation, they needed some way to segregate districts based on "threat" or some other skill rating. In theory, with balanced districts, the players are already close in skill to any other player, so that matching can work again.

- Exposing the threat levels and giving players a choice on which districts they enter only breeds a way for players to undermine the necessary balance in districts.

 

Is Threat broken? Yes. Is it frustrating? Yes.

 

Here are my current thoughts on how to fix it.

 

"Cross district matchmaking"

The concept is to allow players to join whatever action district they want. No spamming servers to get into the one full district. They simply join, drive around and get to interact as part of the district while waiting for a group. Players are matched with other players across all instances of the same action district, so we can ideally narrow the range of skill as much as possible. Once they are matched, the server moves both teams to a different server where the mission plays out.

In my opinion, this is the holy grail. We eliminate segregated districts, because it doesn't matter which one you join. We eliminate visible threat levels, because it's not information you can act on. We increase the pool of players you match against for even matches.

 

What are the problems with Cross-district matchmaking?

First, and most importantly, it requires us to finish upgrading to Unreal 3.5 which supports Phasing. We need the ability to move players from one instance of a district to another instance of the same district without unloading all the art. Second, we need to add latency to our matchmaking algorithm so that you are all grouped on an instance of the district that is closest to most of the players.

 

Lastly, in an ideal scenario, we would collapse down to a single World with multiple servers in North America, Europe and Asia. Then we can distribute Regional districts in as many locations as possible for proper latency / skill / matchmaking.

 

Thoughts?

 

Thanks,

Matt

[Hexerin 1140]

Even if you hide threat level and remove district instance choice, it's still possible (and easy) to dethreat because it's an automated system.

[CookiePuss 5379]

As a Jericho player, I wonder how much phasing will even help considering we can barely fill a single silver  district anyways.

Edited October 19, 2018 by CookiePuss

[vsb 6171]

2 minutes ago, Hexerin said:

Even if you hide threat level and remove district instance choice, it's still possible (and easy) to dethreat because it's an automated system.

i think mattscott has said in the past he’d like threat change to be more gradual, which would make dethreating more difficult 

 

i honestly don’t think there’s any way to make dethreating impossible without it also affecting regular players 

[ky4 27]

1 minute ago, CookiePuss said:

As a Jericho player, I wonder how much phasing will even help considering we can barely fill a single silver  district anyways.

If we ever get to a point when the engine upgrade is complete and these changes are possible then I'm sure there will already be way more people back in the game. 

[MattScott 15242]

19 minutes ago, Hexerin said:

Even if you hide threat level and remove district instance choice, it's still possible (and easy) to dethreat because it's an automated system.

 

15 minutes ago, BXNNXD said:

i think mattscott has said in the past he’d like threat change to be more gradual, which would make dethreating more difficult 

 

i honestly don’t think there’s any way to make dethreating impossible without it also affecting regular players 

I think a lot of work needs to be done on how threat or skill is calculated. It feels like we should be measuring activity better, and change should be more gradual (like BXNNXD referred from an earlier post of mine).

[MattScott 15242]

22 minutes ago, CookiePuss said:

As a Jericho player, I wonder how much phasing will even help considering we can barely fill a single silver  district anyways.

My first reaction is: "It wont hurt"

I feel like filling a Silver district right now is a bad criteria because it is affected by other issues like dethreating and general server population issues.

 

The concept of "Silver" or "Gold" goes away.

Instead I think players would choose to enter the type of action district they want.. Financial or Waterfront.

Then we would load balance and put the player in whatever instance makes sense. That could be 3 instances of Financial with 40 players each. It could be 2 instance of Financial with 60 players each.

Once they want a mission, we can match their skill across all of the players in any instance of the same type.

[Zanting 11]

42 minutes ago, MattScott said:

- Matching based on latency so that all players have roughly the same response time from the server

This is an interesting idea, in the instance that latency is unstable, would the average or highest value be the determining factor? Might be worth considering that other highly variable things affect this aside from just latency, like which resolution a player uses, or if a player has better client performance and stability.

[Hexerin 1140]

54 minutes ago, CookiePuss said:

As a Jericho player, I wonder how much phasing will even help considering we can barely fill a single silver  district anyways.

As @ky4 stated, the upgrade to UE3.5 and matchmaking/threat changes will draw many players back. Especially if @MattScott pushes an advertisement campaign about it.

[Optimus_Crime 81]

The 'roughly-the-same latency' part with our already low population could introduce new problems like barely finding any populated matches or having the same squad and opponents which some people might not enjoy playing with.

 

By 'seperate server' you mean an instance with only say the 4 players on one side and 4 on the other in the entire district kind of thing? I mean this could possibly eliminate mission griefing, but wouldn't it feel a bit isolating and lonely? i'd miss the lollygagging in /d chat but then again what are action districts for if not fighting i suppose.

 

Will the threat level actually be removed or will it just be simply hidden from others? I feel like the former would be a better idea and just let randoms be matched with randoms, it's a 50/50 chance of having a really great match or just a terrible one in general like in old source engine valve games.

[Similarities 226]

Matching based on latency will make it very difficult for players from the SEA area of the world to find matches because they don't have their own servers and their pop is already REAL dead. That's the only worry I can really express there.

[ky4 27]

5 minutes ago, K3ith said:

By 'seperate server' you mean an instance with only say the 4 players on one side and 4 on the other in the entire district kind of thing? I mean this could possibly eliminate mission griefing, but wouldn't it feel a bit isolating and lonely? i'd miss the lollygagging in /d chat but then again what are action districts for if not fighting i suppose.

I was thinking about this as well. Phasing is a good idea but it also takes away some of the open world elements which (in my opinion) are core parts of the game. Perhaps it would work to include multiple matches in the same phase instead of having a phase for each individual match?

[Psygo 15]

Hopefully it doesn't phase people to a separate instance, I'll miss the randomness that is APB.

and what about when your entire clan of 2-3 groups joins the same instance, they'll be split up after the mission starts, and I'll miss running into them and sending them flying into space

[yood 345]

your scenario of the game looks so that bronze, solo players will resist to gold, clan groups .Yes, this is a reasonable and adequate solution to the problem .  

 

 Posted 23 February 2012 - 07:33 AM Do you think it is a good balance? 

 

https://forums-old.gamersfirst.com/topic/257903-do-you-think-it-is-a-good-balance/?p=2460078

 

I understand that this is a difficult and challenging task .but with the solution to this problem you will get an increase in the player base which is also a problem for the game at the moment .

Edited October 20, 2018 by Yood

[MattScott 15242]

44 minutes ago, Similarities said:

Matching based on latency will make it very difficult for players from the SEA area of the world to find matches because they don't have their own servers and their pop is already REAL dead. That's the only worry I can really express there.

I think we can balance it so that latency is only part of the algorithm. At first SEA players might find themselves matched on servers that are farther, giving them higher latency. But I believe over time, since those players are still able to engage and play, we can market and attract new players in smaller regions to help fill in more local opponents.

 

This is basically a chicken or egg problem. Cross district/region matching allows us to start the cycle that leads to higher populations / better matches.

 

44 minutes ago, ky4 said:

I was thinking about this as well. Phasing is a good idea but it also takes away some of the open world elements which (in my opinion) are core parts of the game. Perhaps it would work to include multiple matches in the same phase instead of having a phase for each individual match?

We have to include multiple matches in the same instance, otherwise the server cost would be excessive. I also agree that we need to preserve the open world elements of having players in other matches running around as nice ambience to your own match.

[Salvick 248]

2 hours ago, MattScott said:

Lastly, in an ideal scenario, we would collapse down to a single World with multiple servers in North America, Europe and Asia. Then we can distribute Regional districts in as many locations as possible for proper latency / skill / matchmaking.

 

I'd like to bring a question, I always wonder why South America is barely considered at all by any gaming company. Is not like there are so few players from this region as you could easily find yourself if you could gather these stats.

 

I do understand instead that it is related with the costs to keep a server up it implies, while I'd let in your hands the idea to evaluate the cost/benefits rate of such investment since a bunch of us have been loyal to this game for years spending investing above the average in our purchases, and a lot of South American players left the game or ended up playing less often when the DDoS mitigation and server relocation started in the G1 era.

 

And for those who always think of "Brazil" when someone says something about South America well, just take a look at the map to figure out we are way more than just Brazilians. Regards everyone.

[yood 345]

in the dynamics of this issue, I see only one solution - I will reconsider the tactics of the game - I'll team up with strong players , and will DESTROY a New POPULATION of players . as this make the rest .  we'll tick the box, and we'll discuss it next time . Matt. I wish you good luck .

Edited October 20, 2018 by Yood

[MattScott 15242]

10 minutes ago, Salvick said:

 

I'd like to bring a question, I always wonder why South America is barely considered at all by any gaming company. Is not like there are so few players from this region as you could easily find yourself if you could gather these stats.

 

I do understand instead that it is related with the costs to keep a server up it implies, while I'd let in your hands the idea to evaluate the cost/benefits rate of such investment since a bunch of us have been loyal to this game for years spending investing above the average in our purchases, and a lot of South American players left the game or ended up playing less often when the DDoS mitigation and server relocation started in the G1 era.

 

And for those who always think of "Brazil" when someone says something about South America well, just take a look at the map to figure out we are way more than just Brazilians. Regards everyone.

We investigated hosting servers in Brazil almost immediately, and the cost was very high. However I think with a smaller footprint of 6-8 servers to support player districts it becomes more feasible.

[CaptainSloth2Guns 15]

2 hours ago, MattScott said:

Hi everyone,

 

I'd like to start by thanking all the players who threw out suggestions about topics they would like to discuss.

 

The point of this discussion is to peel the curtain back and attempt a two-way conversation with you guys. I'm going to make the disclaimer that while I strive to be as transparent as I can, there are some things I can't share for any number of reasons. If I point an area out that is off limits, please do not assume that I'm trying to hide some evil intention. My hope is that we can use this discussion to help inform my team to some of your concerns, needs, or wishes, and then we can try to take those into account as we move forward.

 

I saw a lot of comments on threat / matchmaking, so we'll try to tackle that one first.

 

For most of these, I'm going to start by giving my opinion/context on the subject. Again, I'm going to make the disclaimer that I am not an expert on APB. I have employees who are that design these systems. Instead, I am the final vote. So it's important that I be as current as possible with the game and the players.

 

The purpose of me exposing my knowledge (or lack there of) is so we can have a discussion on common ground.

If there is a concept that I'm missing, then I trust that you, the players, will point it out and fill me in.

 

So.. IMO:

 

1) Why is matchmaking so important in APB versus other games? 

APB is somewhat unique in requiring PvP conflict to progress through the game. Any time someone wins, someone also loses. In my opinion, there is nothing wrong with losing matches. That's how we learn. But in an ideal world, those matches would be close, so that both sides get an opportunity to learn and get better.

 

2) What are my requirements for good matchmaking?

- Matching based on skill that goes up or down each match based on an accurate assessment of how you performed as an individual and how you performed in your team

- Matching based on latency so that all players have roughly the same response time from the server
- Matching across the widest pool of players to narrow the difference in skill as much as possible

 

3) What are the problems I see in APB's current matchmaking?

- Due to Unreal 3's notoriously slow loading times, and the fact that "phasing" or switching to another instance of the same district, isn't supported in Unreal 3 - APB was designed to only match one player against another player inside the same district on the same server.

- Working off that limitation, they needed some way to segregate districts based on "threat" or some other skill rating. In theory, with balanced districts, the players are already close in skill to any other player, so that matching can work again.

- Exposing the threat levels and giving players a choice on which districts they enter only breeds a way for players to undermine the necessary balance in districts.

 

Is Threat broken? Yes. Is it frustrating? Yes.

 

Here are my current thoughts on how to fix it.

 

"Cross district matchmaking"

The concept is to allow players to join whatever action district they want. No spamming servers to get into the one full district. They simply join, drive around and get to interact as part of the district while waiting for a group. Players are matched with other players across all instances of the same action district, so we can ideally narrow the range of skill as much as possible. Once they are matched, the server moves both teams to a different server where the mission plays out.

In my opinion, this is the holy grail. We eliminate segregated districts, because it doesn't matter which one you join. We eliminate visible threat levels, because it's not information you can act on. We increase the pool of players you match against for even matches.

 

What are the problems with Cross-district matchmaking?

First, and most importantly, it requires us to finish upgrading to Unreal 3.5 which supports Phasing. We need the ability to move players from one instance of a district to another instance of the same district without unloading all the art. Second, we need to add latency to our matchmaking algorithm so that you are all grouped on an instance of the district that is closest to most of the players.

 

Lastly, in an ideal scenario, we would collapse down to a single World with multiple servers in North America, Europe and Asia. Then we can distribute Regional districts in as many locations as possible for proper latency / skill / matchmaking.

 

Thoughts?

 

Thanks,

Matt

7

Matchmaking is the key to everything successful with APB and the most important aspect. 

 

Why might you ask?

 

GOLDS V GOLDS

CHEATERS V CHEATERS

BRONZE V BRONZE

PREMADES V PREMADES2

 

The worst thing to come from this is peoples egos having to deal with no gold, silver or bronze.

The games would become so more enjoyable and also newbie friendly. If it was 80% effective that would be amazing.

 

 You could even put in a progression system that has no impact on matches but helps peoples ego's for the ones that need it.

 

 

 

 

 

Edited October 20, 2018 by CaptainSloth2Guns

[CookiePuss 5379]

Sounds like a plan.

I wish you all luck!

[haunta 10]

Not a fan of phasing really, would this include us having to connect and wait whilst we connect to the instance?

[IamGalaxy 2]

1 minute ago, haunta said:

Not a fan of phasing really, would this include us having to connect and wait whilst we connect to the instance?

I dont think so, correct me if im wrong please, but i think phasing is instant. if its anything like WoW, i understand that Blizzard is a much more big company then this but it still would be the same right?

[haunta 10]

Just now, Kusarixx said:

I dont think so, correct me if im wrong please, but i think phasing is instant. if its anything like WoW, i understand that Blizzard is a much more big company then this but it still would be the same right?

 

This is what I was hoping for but I still feel like I am losing that populated district feel. When we have finished the mission do we stay in that instance?

 

Also people are still going to be able to get into the lower instances.

[vsb 6171]

1 hour ago, Salvick said:

investing

lol no

[Salvick 248]

1 hour ago, MattScott said:

We investigated hosting servers in Brazil almost immediately, and the cost was very high. However I think with a smaller footprint of 6-8 servers to support player districts it becomes more feasible.

Sounds just amazing to know that at least you looked into it. And why not in other countries? Such as Argentina, Chile, Uruguay. Specially with the depreciation of the currency against the dollar here in Argentine it could be a good chance to get better opportunities, although Amazon and Google data centers are located mostly in Chile and Brazil and we don't have any of these here afaik. Also, not sure but I think there's this company called "Level3" that is the main wholesale provider for Telecom and Telefonica here in Argentina, and they offer premium data center services, but I'm not sure if that has anything to do with server hosting, I think is just a distributor or something like that.

 

 

11 minutes ago, BXNNXD said:

lol no

 

Yeah, fun is an investment, never an expense, that's what I mean. As I see it, I don't make "purchases" of fun, but I invest on it instead. I wasn't meaning it like if we were part of the company.

Nice forum posts count btw.