27 replies to this topic

[Deathlike]

Related thread:
http://mwomercs.com/...-new-algorithm/

In the previous thread, I've explained how the CW MM can be exploited... but this thread is actually a more-indepth issue with CW MM, as it relates to the upcoming event this weekend.

PGI's new algo is as simple as this:
All matches will trade between Attacks and Counter Attacks (includes Holds, but for the sake of the discussion, consider it all as Counter Attacks).

This algorithm assumes that there is always a piece to Counter Attack on.... there are two situations:

1) If the first match on the planet (which is @ 0%) has failed, the next match is an Attack, regardless.***

2) If the first match on the planet has succeeded, the next match is a Counter Attack, and another Attack following that is allowed.


The problem with the algorithm is when the assault on a particular planet "has just started", the teams queuing behind the initial assault MUST WAIT for the initial match to complete. It is IMPOSSIBLE to initiate a Counter Attack when a piece is not available to play that particular match.

So, if the first match continues to result in a loss for the Attackers, BOTH SIDES will have to wait for the Attackers to gain a piece before Counter Attacks and a following Attack results.


Assume that both sides have as many 12-mans groups available on hand.. (like the upcoming event this weekend)

Here's what it would look like queue-wise:

#start - (aka no pieces)
Attacking faction launches (initial) assault with group.
Defending faction shows up with their own group

Attacking faction AND Defending faction queue up.
They must wait for the completion of the first match before attempting Counter Attack.

If the Attack fails for the Attacking faction, go to #start.
Continue to #have1piece

#have1piece
Counter Attack is launched. Followup Attack is launched.
People further down the queues must wait for both matches to be completed.

If both Attacking faction groups fail, goto #start
If and only if 1 Attacking faction group succeeds/fails, goto #have1piece.
If both Attacking faction groups succeeds, goto #have2piece

#have2pieces
1 Counter Attack is launched, 1 Attack is launched.
1 more Counter Attack is launched, 1 more Attack is launched.

If all 4 Attacking faction groups fail, goto #start
If 3 Attacking faction groups fail, goto #have1piece
If 2 Attacking faction groups fail, goto #have2pieces
If 1 Attacking faction group fail, goto #have3pieces
If 0 Attacking faction groups fail, goto #have4pieces

Etc.

I could expand on this, but I hope you get the idea. The problem with the matches when they are @ the extremes (0% or 100%) is that a certain match type cannot be initiated unless a particular result has occurred.


The biggest error in the algo is "enforcing" 50-50 regardless of the number of pieces obtained. The optimal situation in which the current algo is favorable is IF IT STARTS @ approximately 50%.

The old algorithm was borked, but was easier to fix in that it shouldn't automatically give the first-in group the mode of their choice... but in the CURRENT ALGORITHM, it FORCES the ALTERNATING MODE @ the extremes where is running under sub-optimal conditions (waiting for a piece to be made available for a Counter Attack when starting @ 0%) aka "you can't Counter Attack on a piece the Attackers don't own".


TL;DR - Why does it matter?

When the event happens, EVERYONE will be trying to queue up..but have to WAIT for the Attacking faction (I believe that's the Clans?) to complete their match with a win... and if they don't, they try again... with another group... while everyone else is held up in the process.

It's not going to work as intended, and unless PGI recognizes this (it requires queue theory to understand), it's doomed to fail.

When the event occurs this weekend, everyone will be waiting in line to get a match in the first few hours... and it's not going to be pretty.

[Arc Viper]

I'm no expert in interpreting algorithms, but from what I gather clans are at a disadvantage in this system, mainly because with this in place it will take much longer to flip planets. Also, if both teams have evenly skilled players the clans still lose because they would only achieve a stalemate. I can only hope a lot of merc units decide to join the clans for this event.

[Sirius Drake]

Are you sure that not just everybody in the initial queue get a "first match"?

[Deathlike]

Sirius Drake, on 22 April 2015 - 11:47 PM, said:

Are you sure that not just everybody in the initial queue get a "first match"?

Only Attacking Factions get the initiative (preemptive defense doesn't do a thing - no matches/ghost drops are launched).

Counter Attacks rely on a piece actually being owned by the Attacking Faction, which is why they don't randomly launch on their own.

[Willard Phule]

*shrug* I assumed is was going to be broken beyond any possibility of being fixed simply because PGI did it. I like your explanation better...sounds a whole lot smarter. +10.

[Mystere]

How did you verify that everyone must wait for the first assault to win? I'm just scratching my head that something was designed to run as you just described.

[Apnu]

Mystere, on 23 April 2015 - 07:58 AM, said:

How did you verify that everyone must wait for the first assault to win? I'm just scratching my head that something was designed to run as you just described.

Yeah that went over my head too.

The planet has... what... 63 shards? So in theory 63 attacking groups can happen at once. So if one group attacks and one group defends, and another group shows up to attack, won't the system initiate an attack on the next shard? Maybe defenders have to wait for an attack, but that's about it.

I like that the OP is invested, but he has no access to the algorithm and has only evidence from a low population contest because CW is a ghost town at the moment.

[Deathlike]

Mystere, on 23 April 2015 - 07:58 AM, said:

How did you verify that everyone must wait for the first assault to win? I'm just scratching my head that something was designed to run as you just described.

Let me ask you the simple first question.

When is a Counter Attack launched?

Hint: It doesn't happen when the Attackers do not have a piece of the pie.

Apnu, on 23 April 2015 - 08:12 AM, said:

Yeah that went over my head too.

The planet has... what... 63 shards? So in theory 63 attacking groups can happen at once. So if one group attacks and one group defends, and another group shows up to attack, won't the system initiate an attack on the next shard? Maybe defenders have to wait for an attack, but that's about it.

I like that the OP is invested, but he has no access to the algorithm and has only evidence from a low population contest because CW is a ghost town at the moment.

Again, answer the original question that was posed.

How is a Counter Attack launched when the Attackers have no piece?

While under a limited sample size, the queue behavior is actually very predictable if you even bother watching the queues.

It is a likely scenario that I've spent more time in CW than you have since the last patch actually watching what goes on.

I could have some wrong assertions mind you, but I'm pretty confident in the behavior of the queues as currently constituted.


Edit:
The only other possibility is that initial launches would be only Attacks until one is successful, in which case it switches to 50-50 once a successful Attack (gaining a piece has been acquired).

Under this instance, the initial Attack sequence/overwhelm would be the only way to get the pieces a Faction needs to win.

Edited by Deathlike, 23 April 2015 - 09:59 AM.

[Mystere]

Deathlike, on 23 April 2015 - 09:52 AM, said:

Let me ask you the simple first question.

When is a Counter Attack launched?

Hint: It doesn't happen when the Attackers do not have a piece of the pie.

Obviously, it happens as soon as 1 attack succeeds.

But, that should not in any way prevent attacks from launching, and then prioritizing counterattacks as they become available.

That's a very simple algorithm that runs like a very simplified version of Tetris, where the "Tetriminos" are only single square blocks instead of four, and instead of making a whole line disappear, you make a block disappear by dropping another block on top of it, and with the dropped block disappearing as well.

Edited by Mystere, 23 April 2015 - 10:19 AM.

[Apnu]

Deathlike, on 23 April 2015 - 09:52 AM, said:

Again, answer the original question that was posed.

How is a Counter Attack launched when the Attackers have no piece?

How can I? I'm not an engineer for PGI. If you want your question answered. Take it to feedback@mwomercs.com and/or @russ_bullock on Twitter. Or Post a thread, specifically asking an engineer like Mr. Buckton to answer and maybe he will, like the map maker did in another thread. Or do all three and hope that's its not annoying.

Posing a question here for us players and hypothesizing how you think PGI did it isn't going to answer your question. None of us have enough information to answer it. We're all guessing.

That's my point.

Edited by Apnu, 23 April 2015 - 10:17 AM.

[Fire for Effect]

which algorithm?

we kept getting consistently Ghost Drops ALTHOUGH there were enemies in the queue!

[Davers]

Ok, I think I see what Death is talking about, but won't this be sorted out within the first 30 min of the event? It's only a problem if the attackers never win any sectors, right?

[SovietKoshka]

Davers, on 23 April 2015 - 02:29 PM, said:

Ok, I think I see what Death is talking about, but won't this be sorted out within the first 30 min of the event? It's only a problem if the attackers never win any sectors, right?

indeed, I think that so long as ARKAB is on the front, the clanners should have no problems getting "#REKT"

[Spheroid]

I don't think the waiting for counter attack will be a problem with the projected numbers. You are making some large assumptions based on the planetary attacks this week which were virtual ghost downs population wise. Most planets were lucky to have 24 players on both sides but for Tukayyid there should be atleast several hundred or more on both sides.

Clans won 52% of all battles last CW event, so extrapolating forward a good chunk of the planetary zones should be available the phase immediately after the initial attack.

Edited by Spheroid, 23 April 2015 - 05:27 PM.

[odiemoncrew]

Spheroid, on 23 April 2015 - 05:24 PM, said:

I don't think the waiting for counter attack will be a problem with the projected numbers. You are making some large assumptions based on the planetary attacks this week which were virtual ghost downs population wise. Most planets were lucky to have 24 players on both sides but for Tukayyid there should be atleast several hundred or more on both sides.

Clans won 52% of all battles last CW event, so extrapolating forward a good chunk of the planetary zones should be available the phase immediately after the initial attack.

I agree with this guy, You are making Large assumptions based on planetary attacks from when CW was a ghost town, without factoring in the fact that there will be 63 sectors and a larger amount of people.

If we go by the Sheer number of zones & possible number of people play, it will be quite possible that they might have
changed the algorithm to allow a few sectors, 5, 10, maybe even all 63 of them to be fought over at the same time depending on who is queued up, before going to attack/counter attack

or it could be a cliusterF and do exactly as you laid out. We don't know til the event starts, since they haven't
exactly spelled out how the first round of attacks will play out before switching to their attack then counter attack

[Chemie]

1. Clans field 63 teams. Here clans win in less than 10 hours due to the 6% advantage. IS can field 100 teams but it won't matter; all they get is long queue times.

2. Clans field a larger number (say 40) but PGI limits ghost drops to 1 count-down at a time (ie one ghost drop out of the 63 tiles). There will always be a clan team coming out of games to reset the counter. Here, only 40 games are played and the side with fewer teams dictates the number of games. IS can have 100 teams and only get long wait times. Clans win again with 6% advantage

3. Maybe PGI always 4 ghost drop counters (like 4x15) like 4 continents. As long as clans have ~15 teams, they can reset the 4 counters before 10 min is reached. Something like 15 teams can locked this down. IS can have 100 teams but only 15 games at a time are played.

4. PGI allows unlimited ghost drop counters. Here, IS numbers win. They need 6% ghost drops to counter clans advanatage.

But, PGI has said they don't want ghost drops but IS can only win if ghost drops wins is >6% to counter clans higher win percentage.

Conclusions:
* Clans win with 6% advantage unless IS get enough ghost drops which is determined by GAME MECHANIC and not numbers.
* Side with LOWER number of teams determines the number of games. The side with more just see longer queue times.

The above happens because this is ONE planet vs last time where there were something like 20 fronts.

Prediction. Clans have all tiles within 24 hours (say half by end of Friday night and the rest sometime Sat).

Remember, 6% is huge if there are lots of games.

The only way this does not come true is if IS can get to 50/50 wins which only happens is all the best units go IS leaving clans with pugs

IS teams beat clan pug but with so many pugs, IS clans beat IS pugs and win the planet.

[gloowa]

Chemie, on 24 April 2015 - 03:48 AM, said:

snipped for brevity

This post is so full of assumptions stated as facts, biases stated as facts and opinions stated as facts, it was just a waste of time to read. Sorry man.

I love how the 6% is carved in stone in your mind. Because clans had 6% more wins last event (someone had to) this means that it is physically impossible for IS to perform better in this event. Because in MWO entropy and randomness do not exist and not a single player changed faction and not a single player started playing CW and not a single player left CW and quirks have not been changed and new mechs have not appeared... wait...

Edited by gloowa, 24 April 2015 - 04:14 AM.

[Mangonel TwoSix]

I have to say I am excited to see how this event turns out. Without the IS being able to light rush we are going to get to see (or PGI is) what the win percentage is for strait up IS v Clan in the 'combat' game modes of CW.

Since the IS will start with all of the sections of the planet, the IS will never see a generator to rush. I'm sure in the last event the numbers for IS wins were inflated by being able to rush the generators and get a fast win against teams that were not prepared to defend against it.

This time the matches will be decided by combat for the most part.

[Dracol]

Small chance I am mistaken, but I am 99% sure I saw a planet last night that had all blue territories and yet two battles were in progress.

In effect, the algorithm is more complex then the OP claims.

[AlmightyAeng]

Perhaps an easy fix is to turn on the "switches every match function" once a planet hits 50%?