Arena For Profit and Power leveling, a growing trend? - Page 17

Opioid · 08/23/07, 10:44 PM

Originally Posted by Aphyrax

That would still allow a 2500-skilled player to dump himself to 1700, team up with a 1900-skilled player and get the 1900 skilled player to 2100 (and Gladiator title) by repeatedly beating 1900 rated teams. In other words, smurfing would be alive and well and the whole point of the change would be lost.

A 1900 player "cheating" his way to 2100 doesn't strike me as a horrible game breaking flaw that has to be stopped at all costs. The market would be much smaller, as well, for smurfing for an extra 150 arena points for a week, and the labor necessary to pull off such a stunt is greater, which requires planning and not just taking the initiative during the Monday/Tuesday rush. In terms of net utilitarian gain, my system still beats the pants off the way it is.

mikebro · 08/24/07, 12:36 AM

No matter what they do, as long as you only need to rely 1 bracket for points I'm pretty sure power leveling in arenas will be a pretty common thing. All they can do is try to limit it to a degree.

One problem with the 1 week lockout system is that even if you are relying on 1 bracket for points, nothing is preventing you from creating a new team at the start of a week, getting up to 2000-2200 points, collecting your points on tuesday, selling that team for 10 times what you paid for the charter then rinse and repeat.

Komatose · 08/24/07, 12:13 PM

Originally Posted by Mugorim

which is why I'm asking for others such as yourself to point out potential flaws.

One potential flaw I see in individual ratings is it may widen the gap of class representation in high-end arena (something Blizzard is apparently trying to reduce). A high-rated 5v5 would have much less incentive to bring on a rogue for example, unless it's meant to temporarily deflate the team rating.

Example 1 (artificial deflation): Team consists of five 2100 rated players and two 1500 rated players. As long as four 2100 rated players cycle a low ranked player for the 5th spot, they'd be playing rated games around 1980 rating. According to your example, rating-gains from wins are evenly distributed while losses are pro-rated according to actual rank. Unfortunately, it's the losses that bring the mean average in line. But if the "Four 2100, One 1500" group can constantly win, then the mean-average rating would always be 120 points below the higher players' actual rating. Additionally, if and when this team loses a match, the only people suffering would be the high rated players. A 1500 player ought to lose -0- rating if he lost to a team at 1980 (pro-rated for point loss). While inflating the risk/reward of selling team spots, this could hinder classes who traditionally are under-represented in arena (and therefore who potentially have lower individual ratings -- an artificial rating-ceiling, if you will).

The individual ratings would also lead to an interesting paradigm. What happens to the 2300 rated player teamed with four 2500 people. Let's say that in order to attain the rank of gladiator/netherdrake you need to hit 2550. So what does the 2300 player wish for? Losses in order to let him climb that team ladder a bit quicker? Or the highly unlikely win-streak increasing his teammate's to 2750? Granted, at that rating, wins and losses may be more evenly distributed, solving the problem itself. But it is still something to consider.

All in all, putting the risk on the higher rated individual players could solve the smurfing dilemma, but may limit the player-base for high-rated competitive teams when they need to recruit additional members. Additionally, classes already underrepresented in high-end arena could be further under-represented if the risk/reward of bringing them on (assuming they were able to amass a respectable individual rating somehow) is inflated.

Don't get me wrong, these flaws are arguably minimal when compared to the current system where smurfing is the biggest problem. Something ought to happen. Although individual ratings may be a bit too radical for Blizzard's liking.

Question: Would your system of individual ratings incorporate minimum participation requirements for total "team" games? And if so, how would you propose to have that effect the individual rating (if at all) when the member doesn't make the minimum required per week?

Originally Posted by mikebro

One problem with the 1 week lockout system is that even if you are relying on 1 bracket for points, nothing is preventing you from creating a new team at the start of a week, getting up to 2000-2200 points, collecting your points on tuesday, selling that team for 10 times what you paid for the charter then rinse and repeat.

If the participation-requirement is still in effect, the auctioned teams would deflate in rating fairly fast before the new owners would be eligible for any arena points.

The Arena-ID lockout would be the simplest way to minimize smurf teams (although it wouldn't solve it the way individual ratings could). Smurfs would still be around, but would force the Ringers to choose between maximizing their gold, or maximizing their arena points for the week(s). It would also reduce the number of smurf teams-per-Ringer. If the Ringer had 2 smurfs, they'd only be able to contribute to their main point-maximizing team once every three weeks. The effect of smurf teams would be felt much less.

Mugorim · 08/24/07, 2:41 PM

Originally Posted by Komatose

One potential flaw I see in individual ratings is it may widen the gap of class representation in high-end arena (something Blizzard is apparently trying to reduce).

Blizz says they're working on 5v5 balance, so hopefully class improvements to poorly represented classes will deal with that. But unfortunately, if a class is better or worse at one particular arena, their ratings "should" reflect that. I mean, it's not fair, but the system is supposed to rank players based upon how well they do. If the class limits that, it's either blizzard's fault in poor design of class or equiptment, or player fault in difficulty meeting class potential.

Example 1 (artificial deflation): Team consists of five 2100 rated players and two 1500 rated players. As long as four 2100 rated players cycle a low ranked player for the 5th spot, they'd be playing rated games around 1980 rating. According to your example, rating-gains from wins are evenly distributed while losses are pro-rated according to actual rank. Unfortunately, it's the losses that bring the mean average in line. But if the "Four 2100, One 1500" group can constantly win, then the mean-average rating would always be 120 points below the higher players' actual rating. Additionally, if and when this team loses a match, the only people suffering would be the high rated players. A 1500 player ought to lose -0- rating if he lost to a team at 1980 (pro-rated for point loss). While inflating the risk/reward of selling team spots, this could hinder classes who traditionally are under-represented in arena (and therefore who potentially have lower individual ratings -- an artificial rating-ceiling, if you will).

The individual ratings would also lead to an interesting paradigm. What happens to the 2300 rated player teamed with four 2500 people. Let's say that in order to attain the rank of gladiator/netherdrake you need to hit 2550. So what does the 2300 player wish for? Losses in order to let him climb that team ladder a bit quicker? Or the highly unlikely win-streak increasing his teammate's to 2750? Granted, at that rating, wins and losses may be more evenly distributed, solving the problem itself. But it is still something to consider.

No, I'm sorry if the example wasn't clear, i was trying to reduce typing. In the idea proposed, wins and losses are both pro-rated according to rank. It was just that everyone on the team had the same rank. Since all points accumulated or lost are pro-rated according to rank, the team should converge upon a single rank the longer they play together.

Also, the maximum point spread is supposed to specifically deal with artificial deflation. For the example of 4 2100 & 1 1500, if the maximum point spread is set at 100, then the 1500 players will all be treated as 2000 point players while they play, so the team rating isn't 1980, but instead be 2080, so they don't really gain a whole lot. It does pose risk to the 1500 player, as he's playing in the 2100 bracket, and if he loses, he loses points as if he were a 2000 player (but it still comes off the 1500). This seems fair to me, as, before he joins a 2100 point team, they should all have faith in his ability to play at that level.

The maximum point spread allows high players the ability to bring in 1500s if they wish without significant risk to themselves. If you didn't have the point spread, and they were playing at 1980, then if that new player didn't work out, every loss would be that much more of a significant hit to the ratings of the 2100 players. This will soften the blow a bit, encouraging high-rated teams to experiment with lower end players if they think there's potential there.

Question: Would your system of individual ratings incorporate minimum participation requirements for total "team" games? And if so, how would you propose to have that effect the individual rating (if at all) when the member doesn't make the minimum required per week?

Minimum number of games per week for the individual only (regardless of team) for arena points. Minimum number of games per week on a single team to count toward being an active member of a team (for end of season tournaments only).

If the participation-requirement is still in effect, the auctioned teams would deflate in rating fairly fast before the new owners would be eligible for any arena points.

The Arena-ID lockout would be the simplest way to minimize smurf teams (although it wouldn't solve it the way individual ratings could). Smurfs would still be around, but would force the Ringers to choose between maximizing their gold, or maximizing their arena points for the week(s). It would also reduce the number of smurf teams-per-Ringer. If the Ringer had 2 smurfs, they'd only be able to contribute to their main point-maximizing team once every three weeks. The effect of smurf teams would be felt much less.

I actually agree that the lockout ID (play 1 game on a team an you're locked into it) would be another good solution. It's still not as thorough in my mind, but it will limit smurfing beyond where it is currently, and i'd imagine a bit easier to get in.

oldmandennis · 08/24/07, 4:09 PM

Originally Posted by Aphyrax

"Floor and flatten the win/loss rating gain curve at 0 instead of 1 making winning a broader delta unprofitable"

That would still allow a 2500-skilled player to dump himself to 1700, team up with a 1900-skilled player and get the 1900 skilled player to 2100 (and Gladiator title) by repeatedly beating 1900 rated teams. In other words, smurfing would be alive and well and the whole point of the change would be lost.

That's really a lot of effort. To get 200 points in this way takes 400 games, because your pro has to lose one for every one you win. At 10 games an hour, that's the equivalent of buying a 1-60 powerlevel - except you can't pay the Chinese to do it, it has to be somebody with a low ping who can communicate. Remember at the end of this grind, you are trying to play with one pretty good player and one godlike one, against 2 very good ones. Any loss at that level will set you back about 6 hours, because your scrub will be losing the maximum amount of points. You are talking about going beyond what's really possible for a paid service, and into the realm of crazy things you hear about happening one time.

Also, your pro in this scenario is giving up a huge number of points by staying at 1700 instead of his real rating.

Aphyrax · 08/24/07, 7:59 PM

If I were in charge of fixing this problem I would not tackle it via matchmaking and ratings changes. If you remove smurfs, the matchmaking and point-awarding system works remarkably well for the vast majority of the population. It is a bit iffy at the very top, especially for 2v2 with its rock/paper/scissor nature, but overall it works very well, and Blizzard cannot conjure top-rated opponents out of thin air.

The advantages of using means outside of matchmaking is that it would not affect the average player. You don't have to average out the individual ratings of 10 people and then compare that average against all players, feed that into the points formula and normalize the point gains. Instead, the only people affected by this would be the smurfs and their customers, enabling a very targetted approach where you can use complex, evolving strategies to counter exploits without complicating the system for 99% of the population.

The simplest such measure would be the already proposed cooldown on team switching. But that would restrict people who swtich teams for fun and not to exploit the system (a case could be made that this should still be prevented since distorts the system but that's a seperate debate).

A more complex approach would be to for each week keep track of the point gains and losses of each player on the team. For example, say the ringers boosted the rating by 200 and the scrubs brought it down by 150. Overall the team gained 50 points. Now you compare the average gain of 50 with the individual gain of erach player, and if there is a huge negative gap that player does not get squat. For example, if (average ratings change per game for the week for the entire team) - (average ratings change per game played by player X for the week) is greater than max gain/3 (ie 10 in the current system), the player does not get anything.

In our case, say the ringers played 14 games and the scrubs played 6. The average gain per game would be 2.5 (50/(14+6), while the average gain for the scrubs would be -150/6 = -25. The gap would be 2.5+25 = 27.5. Such a big gap will virtually never occur under normal circumstances.

That system would require your performance to be roughly in line with the rest of the team to get points. It has no impact on matchmaking or on teams that don't have ringers dropping in and out, so it can be more complex and under the covers since the average gamer will never see it. You can make the gap wide enough that even teams with two lineups that vary somewhat in skill will not be caught by this. But a gap of 27.5 as in our example would require one lineup to win pretty much all their games and one to lose pretty much all their games. That is as conclusive proof of fishy behavior as you can get.

Mugorim · 08/24/07, 9:54 PM

Originally Posted by Aphyrax

If I were in charge of fixing this problem I would not tackle it via matchmaking and ratings changes. If you remove smurfs, the matchmaking and point-awarding system works remarkably well for the vast majority of the population. It is a bit iffy at the very top, especially for 2v2 with its rock/paper/scissor nature, but overall it works very well, and Blizzard cannot conjure top-rated opponents out of thin air.

The advantages of using means outside of matchmaking is that it would not affect the average player. You don't have to average out the individual ratings of 10 people and then compare that average against all players, feed that into the points formula and normalize the point gains. Instead, the only people affected by this would be the smurfs and their customers, enabling a very targetted approach where you can use complex, evolving strategies to counter exploits without complicating the system for 99% of the population.

The simplest such measure would be the already proposed cooldown on team switching. But that would restrict people who swtich teams for fun and not to exploit the system (a case could be made that this should still be prevented since distorts the system but that's a seperate debate).

A more complex approach would be to for each week keep track of the point gains and losses of each player on the team. For example, say the ringers boosted the rating by 200 and the scrubs brought it down by 150. Overall the team gained 50 points. Now you compare the average gain of 50 with the individual gain of erach player, and if there is a huge negative gap that player does not get squat. For example, if (average ratings change per game for the week for the entire team) - (average ratings change per game played by player X for the week) is greater than max gain/3 (ie 10 in the current system), the player does not get anything.

In our case, say the ringers played 14 games and the scrubs played 6. The average gain per game would be 2.5 (50/(14+6), while the average gain for the scrubs would be -150/6 = -25. The gap would be 2.5+25 = 27.5. Such a big gap will virtually never occur under normal circumstances.

That system would require your performance to be roughly in line with the rest of the team to get points. It has no impact on matchmaking or on teams that don't have ringers dropping in and out, so it can be more complex and under the covers since the average gamer will never see it. You can make the gap wide enough that even teams with two lineups that vary somewhat in skill will not be caught by this. But a gap of 27.5 as in our example would require one lineup to win pretty much all their games and one to lose pretty much all their games. That is as conclusive proof of fishy behavior as you can get.

not a bad idea, however there are a few flaws with it. You have to be very careful to make sure that no legitimate teams can get hit by this. Especially because it would be behind the scenes, and not give your average player any warning, you'd get a wonderful WTF WHERE'S MY POINTS reaction from anyone who is inadvertantly hit by it, so you can't make the spread too low.

Not to mention you could still smurf around this. Even behind the scenes, people will find where that number is, and will play around that. If you set the number too large, ringers could still just level the team up to a high rating one week, and then play just enough games to stay under the point difference barrier, even if this means losing some rating from week to week. They'll slow the decent a bit, and after they've fallen sufficiently, they'll just boost the team back up in the next week and start smurfing again the next week. And they can still do it with multiple teams/week. In fact it's even easier on them because it essentially gives them a maximum number of games/week, so they would likely have time to run more teams at once.

however it does make smurfing more of a pita to run, which is good.

It still doesn't deal with lose 10 & reform that provides welfare epics to people who never win. Nor provide the potential for a single team to duplicate itself, play up to it's own level, and then queue up alts or friends against their main team to artificially inflate points. Granted, this secondary behavior i've only heard about people trying without much success so far, so not that much to worry about yet.

Killmour · 08/27/07, 5:57 PM

I really hope this gets fixed soon. We played 10 games today, since we have been busy all week and all we came up against was people power leveling teams. We lost 6 games against teams with full t2 glad gear, and some with some t6 off items, and yet against all of those teams we lost 20+ rating a game.

Like I'm supposed to believe that a team with full glad gear and t6 is 1500 rated. We never once played the same team twice either, it was 6 separate teams that were leveling up sell teams.

It almost seems like only 1 out of 3 teams we play are actually in our own bracket, which kind of makes arena stupid. What is the point of playing arena if you can't even play against your own bracket?

Puffypaladin · 08/28/07, 3:54 AM

I think that most of the automatic system will fail, due to the fact, that human beings will always find ways to out smart the machine. At least without super, crazy, complicated system to hold them off.

I like the Gold sellers and AV (future) system. Let the players right click on the teams, and say they are power levelers. If enough complains will pop up for the GM, he will go and check for himself the history of the player and ban/prevent him from arena.

This will use more human power from blizzard, but it will give, in a long term, a much better solution. If they managed to kill some 90% of the gold farmer, they sure can kill this problem. Just issue one GM per arena, per battle group, and you will get, in my humble opinion, a much better results.

Leto · 08/28/07, 10:00 AM

One complication with reporting would be that new teams of arena vets would perform similar to smurf teams, though without the team selling, but I guess those could be dealt with on an individual basis as there shouldn't be a ton of those around.

Puffypaladin · 08/28/07, 10:34 AM

My proposal was to fix the grand sellers. If one person do it twice a season, its not a big deal. But if someone do it every week, he will be found out pretty soon, and wont be able to do it.

The punishment should be, you cant change your team for the coming month, or so. Nothing serious, so even if someone got this punishment by mistake, it won't hurt him too badly.

Mugorim · 08/28/07, 2:20 PM

Originally Posted by Puffypaladin

I think that most of the automatic system will fail, due to the fact, that human beings will always find ways to out smart the machine. At least without super, crazy, complicated system to hold them off.

I like the Gold sellers and AV (future) system. Let the players right click on the teams, and say they are power levelers. If enough complains will pop up for the GM, he will go and check for himself the history of the player and ban/prevent him from arena.

This will use more human power from blizzard, but it will give, in a long term, a much better solution. If they managed to kill some 90% of the gold farmer, they sure can kill this problem. Just issue one GM per arena, per battle group, and you will get, in my humble opinion, a much better results.

It seems like you misunderstand the AV "solution". You report people in AV, if enough report them, then they get a debuff that will prevent them from getting honor. However, if they enter combat (apparently with any NPC), then the debuff is cleared, and the rest of the battle ground must report you again to reapply the debuff. There's no GM involvement whatsoever. It would be great if this fixed the problem, but i severely doubt it will, as it sounds far too easy to circumvent.

This is a horrible system, I would like to play the game, not have to run around policing their stupid game for them. Especially if they can easily clear their "punisment" and go back to what they were doing.

But you are not necessarily correct about "people will always find a way to beat the system". Yes, that may be true to some degree, but you've got to account for the fact that the "system" in this game can and will change to adapt to people who are abusing it. This is an MMO, and will (provided Blizz has the desire) change. Besides, if you leave it up to human judgement, it's not necessarily fair, as the GM gets to make the decision. If the system is designed to solve the problem, then everyone gets treated the same way. Not to mention, the desire isn't necessarily to fix the problem entirely (as a "perfect" fix to all problems is not likely possible), but to limit and reduce the problem so it's not worthwhile to most of the people doing it.

Amera · 08/29/07, 12:31 AM

Would something as simple as adding a gold cost to joining teams reduce the problem? I mean it would be a monkey sink along the lines of respeccing, and maybe enough to raise the cost of team swapping enough to reduce its frequency.

Puffypaladin · 08/29/07, 2:02 AM

Since people are ready to pay 1k and more for this power leveling, and on the other hand, no one will agree to pay 300g just to replace a friendly (and lawful) member in the group, its obvious that this solution will not work.

Sartuk · 08/29/07, 2:15 PM

Originally Posted by Zure

I can't see any way to prevent *some* abuse without changing the current system.

If a player has to play a certain number of games or a certain percentage of a team's games to get points/drake/etc, then the purchaser can just play as many games as they like at a 1500 rating, then sit while 2/3/5 people power level the team to the top of the brackets.

But how it plays out from here differs based on what people want:

Drake/title:

Strategy: I think the strategy is to just hold the team where it, playing as little as possible after the PL. Here, counting only the last 100 games is the proposed solution by Mr. Kaplan. But a 1500 team can be PL'd to a 2400 rating in 70 or so games (100-30 starter games played by the purchaser), so I don't see this working perfectly. Additionally, 1500-1800 (and beyond) could simply be played with the purchaser on the team, just contributing as best he can.

Proposed Solution: Penalize teams for lying dormant. This can be done in many different ways, such as charging 100 points if the team doesn't play the minimum number of games (which could be less than 10). There is collateral damage with this solution though. Legit teams that just can't get all 5 people on for the final weeks of the season could lose their mounts. In the end there may be no fair way to avoid people buying nether drakes. But really, I'm not sure I see a huge problem here both in terms of competitive impact and number of occurrences.

Points/gear:

Strategy: I think the strategy is to get PL'd to a very high level and then put in 3 games a week during peak times when you know you won't lose 30 points in a game, then queue leave for the next 7 games.

Proposed solution: Have the team have to actually *play* a full 10 games a week to get points, which should cost about 150 points a week, and likely more. Seems like an easy fix, am I misunderstanding something?

Basic math: You purchase a 2200 team -- You'd have a week at 2200, then 2050, 1900, 1750. Say you could have sustained 1500 on your own. 1202+1070+894+692+432 - (376*5) = 2410 net points "earned" from purchasing a team, which isn't a game breaking amount in my opinion.

Under the current system you'd earn about 4800 points net in the first 5 weeks of purchase, and about 6000 over the lifetime of the investment.

Pretty sure that the "queue leave" technique does NOT work any longer, nor has it worked for at least around a month.