|
|
| Author |
Thread Statistics | Show CCP posts - 1 post(s) |
Leither Yiltron
Molon Labe.
1
   |
Posted - 2015.06.30 05:53:00 -
[1] - Quote
deezy dabest wrote:Kain Spero wrote:Deezy, except for no one that knows what their talking about is worried about matches actually occurring. The arguments that I've heard about team deploy is that there is concern that there will be instances of 16 vs 16 randoms or a variation there off.
If you are going to try and argue a point at least try to use a valid argument that takes into account with a rolling player population in FW of greater than two teams your problem doesn't exist. Why is it that no one in your corp can understand simple mathematics?
I'm actually due to study mathematics as a graduate student at the Uni. of Oxford this upcoming year. As such, I feel reasonably confident when I say what you describe as "basic" mathematics in your posts is neither mathematics in the strictest sense of the rigor required for proper mathematical discourse, nor does it resemble mathematics that is "basic".
Mathematical modelling is an entire discipline that takes serious study to understand properly. In this case we'd need to discuss a sincere amount of probability theory in deriving an appropriate approximation of the main thing we'd want to know, which is average wait time on a by-group-size basis. Realistically there are variables which can only be guessed experimentally, such as the average size of a group in the Factional Warfare matching pool. We could wax eloquent all day about how we might look to model player queuing, and queue theory in and of itself is literally its own area of research (https://en.wikipedia.org/wiki/Queueing_theory). I think it's clear that none of us have the time to write a thesis on the subject.
You did have the seeds of a model that I think you were trying to use to make conclusions about potential waiting time situations. It was flawed in that you assumed that matches won't start without full 16 man teams on each side, which I believe is experimentally not the case.
Given the complexity of analyzing a queuing system like this from a mathematical standpoint, you'd probably be better served by discussing some other aspect of FW grouping concerns.
Have a pony
|
Leither Yiltron
Molon Labe.
1
|
Posted - 2015.06.30 06:13:00 -
[2] - Quote
On the one hand, a 16 man queue requirement is not what I've observed experimentally, though there's always room for an investigation of legitimate evidence. On the other even assuming that the threshold is 16, making intricate assumptions about arrival times of groups of particular sizes isn't something to do without justification.
There are plenty of interesting discussion points already on the table in regards to this thread's topic, but I virtually guarantee that mathematical modelling of wait times isn't one of them.
Have a pony
|
Leither Yiltron
Molon Labe.
1
|
Posted - 2015.06.30 06:30:00 -
[3] - Quote
deezy dabest wrote:Leither Yiltron wrote:On the one hand, a 16 man queue requirement is not what I've observed experimentally, though there's always room for an investigation of legitimate evidence. On the other even assuming that the threshold is 16, making intricate assumptions about arrival times of groups of particular sizes isn't something to do without justification. There are plenty of interesting discussion points already on the table in regards to this thread's topic, but I virtually guarantee that mathematical modelling of wait times isn't one of them. Proof is a helluva drug. It does not take a fully flushed out model to see the chances of it really messing up the system. I think that anyone can look at the basic example scenarios and see where flaws are in the system.
This really isn't a trivial mathematical claim. Your concern would be a component of modelling wait times for a scenario this complex, but the actual probabilities would require a full analysis. The "basic examples" you present are meaningless without probability distributions associated with their likelihood.
Have a pony
|
Leither Yiltron
Molon Labe.
1
|
Posted - 2015.06.30 06:53:00 -
[4] - Quote
deezy dabest wrote:Leither Yiltron wrote:deezy dabest wrote:Leither Yiltron wrote:On the one hand, a 16 man queue requirement is not what I've observed experimentally, though there's always room for an investigation of legitimate evidence. On the other even assuming that the threshold is 16, making intricate assumptions about arrival times of groups of particular sizes isn't something to do without justification. There are plenty of interesting discussion points already on the table in regards to this thread's topic, but I virtually guarantee that mathematical modelling of wait times isn't one of them. Proof is a helluva drug. It does not take a fully flushed out model to see the chances of it really messing up the system. I think that anyone can look at the basic example scenarios and see where flaws are in the system. This really isn't a trivial mathematical claim. Your concern would be a component of modelling wait times for a scenario this complex, but the actual probabilities would require a full analysis. The "basic examples" you present are meaningless without probability distributions associated with their likelihood. Okay so give me a basic example of how queue times get screwed over in 8 v 8 and lets compare. Oh wait with no number greater than 50% of the required total there is not a single such example of more than enough people searching for battle without a battle taking place. That is all that needs to be said.
Unfortunately I'm going to have to stop responding for awhile with the note that it's easier to make incorrect statements like yours than it is to debunk them. If there's a 0% probability of something happening, the fact that it exists as a configuration of the given system is irrelevant in a reasonable model. Additionally, purporting that queue times get "screwed up" in particular configurations without anything approaching rigorous reasoning is incorrect and irresponsible.
Have a pony
|
Leither Yiltron
Molon Labe.
1
|
Posted - 2015.06.30 20:00:00 -
[5] - Quote
Piercing Serenity wrote:This argument is not a discussion of two sides of the same coin. Some people like Kain and Bright are arguing: "DUST deserves a game mode that supports TD". Other people like me are arguing: "The current implementation of FW cannot support TD". Although these argument are related, they are not in opposition. I don't think Deezy, Mina or I are against team deploy. We are just opposed to putting TD into a FW system where there is a limit on players.
At the end of the day, I agree with Kevall. If we implement team deploy in FW's current state , we're going to see "CCP fix FW now!", "[Suggestion] Improving FW incentives" and "[Faction 1] / [Faction 2] kickstart event" threads on the forums.
That's a bit of a null implication, since those threads should be present with or without any changes to deploying mechanics. FW isn't in a fantastic state, and needs to be in contention for the next "big thing" to be looked at after PC. It probably should have come before PC.
If you have a good logical reason why people will vacate FW given a change to 16 players instead of 8, by all means. But I don't think that FW's reward structure being broken is a product of any grouping concerns.
Have a pony
|
Leither Yiltron
Molon Labe.
1
|
Posted - 2015.06.30 21:17:00 -
[6] - Quote
Piercing Serenity wrote:Leither Yiltron wrote:Piercing Serenity wrote:This argument is not a discussion of two sides of the same coin. Some people like Kain and Bright are arguing: "DUST deserves a game mode that supports TD". Other people like me are arguing: "The current implementation of FW cannot support TD". Although these argument are related, they are not in opposition. I don't think Deezy, Mina or I are against team deploy. We are just opposed to putting TD into a FW system where there is a limit on players.
At the end of the day, I agree with Kevall. If we implement team deploy in FW's current state , we're going to see "CCP fix FW now!", "[Suggestion] Improving FW incentives" and "[Faction 1] / [Faction 2] kickstart event" threads on the forums. That's a bit of a null implication, since those threads should be present with or without any changes to deploying mechanics. FW isn't in a fantastic state, and needs to be in contention for the next "big thing" to be looked at after PC. It probably should have come before PC. If you have a good logical reason why people will vacate FW given a change to 16 players instead of 8, by all means. But I don't think that FW's reward structure being broken is a product of any grouping concerns. I think that people will leave more often with 16 man squads for some of the same reasons that newer corps don't go into PC: there isn't a reasonable expectation of an even fight. I understand that FW is not designed to be even all of the time. However, think one of the conditions for queuing for FW is: "I believe that, more often than not, me and my team mates will get good fights instead of bad ones", where "Good fights" could be defined as "Fight that we win", "Fights that give me lots of LP", "Fights that are close", etc. I don't think that the aforementioned condition will be satisfied for a group of 4-8 people who want to play FW. They will be going up against a much more organized team, and (I contend) will lose more often than not. Instead of queuing for a game mode where the reasonable expectation is: "I will, more often than not, get into bad fights instead of good ones", people will queue for game modes where that condition is satisfied. And that is going to be Pubs.
The same's true of, for instance, FW in Eve. Invariably the smaller organizations who don't want to disband end up coordinating with others who can bolster their ranks to take on the larger corps.
Without a much larger player base than it's reasonable to hope coalesces it's unreasonable to expect that segmenting up the number of squads people have to put together for a sync will discernibly impact the number of "high grade" syncs like NF's or Lucent Echelon's that will be entering FW.
I said it in 2012, and it has been true for the 2.5 years since: If you leave communities without an easy-to-organize way to get into matches with a full team on short notice, those communities will atrophy and die. Looking at our corporate landscape, that's exactly what has happened.
Have a pony
|
Leither Yiltron
Molon Labe.
1
|
Posted - 2015.07.05 21:13:00 -
[7] - Quote
Adipem Nothi wrote:Kain Spero wrote:At it's core though the indication is that issues with the team builder breaking down are not based in any kind of real math. This is a textbook optimization problem, and optimization is a very real kind of math, Kain Spero. Restricting input sizes to 4, 8 or 16 would absolutely result in less process time and - more importantly - substantially lesser odds of oddly shaped inputs getting hung up in queue while awaiting complementary pairs. If you think I'm trying to fool you, ask Demens Grimwulff. Or get Leither in here to defend his theory. Leither's math is fun and appears to be accurate, but for these purposes, it is incomplete and assumptive. Edit: You've got a billion Isk, right? Care to make a wager?
You're completely right that the problem is irrelevant, but you kinda say: "The problem is irrelevant" and then "The problem is actually relevant" in two subsequent sentences.
Your statement "substantially lesser odds of oddly shaped inputs getting hung up in queue while awaiting complementary pairs" is unclear for two competing reasons:
- It is precisely a statement which is equivalent to the counting problem I attempted to detail in my discussion. It's clear that the problem is difficult, and that the results don't necessarily match your claimed intuition about the impact of larger squad sizes has on the proportion of "good" squad partitions. In fact, my proof was incorrect for a number of reasons, though I've finally turned up what looks to be a paper by an undergraduate or graduate student pair that can be used to put the question to rest for a few basic examples. https://www.rose-hulman.edu/mathjournal/archives/2008/vol9-n2/paper5/v9n2-5pd.pdf ; I'm checking it and then I'll have to futz around with some CAS stuff.
- All of the discussion of proportions of valid partitions is completely irrelevant to the problem at hand. We have absolutely no clue as to how frequently groupings of particular sizes typically arrive in the queue. If solo players turn up excessively often then the entire discussion is meaningless. This isn't a matter of optimization, it's a matter of queuing theory, and a complicated matter at that even if we had the information that we're missing.
There are so many better points of contention in this discussion space than trying to make an argument based on mathematics that none of us have the time to synthesize. The statement is entirely vacuous without more information and more analysis. Even further, it completely ignores the psychological aspect of how the queuing environment will probably change as players respond to new conditions.
I'm not entirely caught up on this thread because it has run so long. Perhaps you were suggesting that we should just restrict groups to sizes of 4, 8, and 16 wholesale in FW. That might not be a terrible idea for a number of psychological and gameplay reasons, but in terms of math? I'm not so sure.
Have a pony
|
Leither Yiltron
Molon Labe.
1
|
Posted - 2015.07.05 22:03:00 -
[8] - Quote
deezy dabest wrote:Leither Yiltron wrote:Adipem Nothi wrote:Kain Spero wrote:At it's core though the indication is that issues with the team builder breaking down are not based in any kind of real math. This is a textbook optimization problem, and optimization is a very real kind of math, Kain Spero. Restricting input sizes to 4, 8 or 16 would absolutely result in less process time and - more importantly - substantially lesser odds of oddly shaped inputs getting hung up in queue while awaiting complementary pairs. If you think I'm trying to fool you, ask Demens Grimwulff. Or get Leither in here to defend his theory. Leither's math is fun and appears to be accurate, but for these purposes, it is incomplete and assumptive. Edit: You've got a billion Isk, right? Care to make a wager? You're completely right that the problem is irrelevant, but you kinda say: "The problem is irrelevant" and then "The problem is actually relevant" in two subsequent sentences. Your statement "substantially lesser odds of oddly shaped inputs getting hung up in queue while awaiting complementary pairs" is unclear for two competing reasons:
- It is precisely a statement which is equivalent to the counting problem I attempted to detail in my discussion. It's clear that the problem is difficult, and that the results don't necessarily match your claimed intuition about the impact of larger squad sizes has on the proportion of "good" squad partitions. In fact, my proof was incorrect for a number of reasons, though I've finally turned up what looks to be a paper by an undergraduate or graduate student pair that can be used to put the question to rest for a few basic examples. https://www.rose-hulman.edu/mathjournal/archives/2008/vol9-n2/paper5/v9n2-5pd.pdf ; I'm checking it and then I'll have to futz around with some CAS stuff.
- All of the discussion of proportions of valid partitions is completely irrelevant to the problem at hand. We have absolutely no clue as to how frequently groupings of particular sizes typically arrive in the queue. If solo players turn up excessively often then the entire discussion is meaningless. This isn't a matter of optimization, it's a matter of queuing theory, and a complicated matter at that even if we had the information that we're missing.
There are so many better points of contention in this discussion space than trying to make an argument based on mathematics that none of us have the time to synthesize. The statement is entirely vacuous without more information and more analysis. Even further, it completely ignores the psychological aspect of how the queuing environment will probably change as players respond to new conditions. I'm not entirely caught up on this thread because it has run so long. Perhaps you were suggesting that we should just restrict groups to sizes of 4, 8, and 16 wholesale in FW. That might not be a terrible idea for a number of psychological and gameplay reasons, but in terms of math? I'm not so sure. Back to the math stuff. The simple issue here is that you are trying to calculate based on a clean slate which is simply not the case. We already have squad prioritization in effect which means that would have to be reworked as well before 16 man can be added.
No, I've never purported that the conclusion here has any bearing on the actual situation from a mathematical standpoint. What I've set about to show is that yours and Adipem's intuitions are absolutely not useful, and that's patently the case. It should be clear to you and Adipem now that the existence of "bad" partitioning configurations is only a single, tiny observation in what would be a detailed investigation. For the length of the history of this discussion, I've been emphasizing that the fact of the matter is that there are no useful conclusions to draw on a mathematical basis without substantial further analysis of a type none of us are going to do.
"There will be odd parts!" is a concern that is neither here nor there without information that we don't have. The problem is not simple, and both of your continuing posting patterns make it clear that you don't really understand the level of analysis necessary to draw even cursory conclusions.
I'm not really going to sit around arguing this point much further, since I'm pretty sure only you and Adipem insist on trying to draw mathematical conclusions without any particular rigorous basis in fact. From the outset, the only conclusion any reader should take away is this: Without proof it's completely impossible to say whether a mathematical statement is true or false. Claiming that a statement is true because of a few observations that make you "feel that way" is a great place to start investigating proof, but it's not a valid place from which to draw a conclusion.
Have a pony
|
|
|
| |
|