First of all, let's get all our talk about the frequent crashing out of the way: Ack! It crashes more often than a drunk playing Grand Theft Auto!

Okay; now that that's out of our system, we can talk about the changes made and how things will be once the servers are again stable. Specifically, I'm going to discuss the much-discussed and much-misunderstood changes to eight-man queues.

Uh-oh... It sounds like I don't think the 4-3-2-2 is the worst idea in the history of mankind! I must work for Wizards of the Coast. Well, I think the 4-3-2-2 queues are a good idea, I don't work for Wizards, and I have good reasons for my opinion. There are several aspects to consider in order to evaluate the current system against the old one - but before I can explain why 4-3-2-2 is a good idea, I have to make sure we are all on the same page as far as understanding ratings.

The Elo rating system, which is used for Magic, uses a formula to approximate the chances of victory as a function of the difference in rating between two players. The system then attempts to adjust the relevant ratings after every match is played, in order to make its estimate as close as possible to reality. The win rate formula is:

Your Win Rate = 1/(1 + 10^((Opponent's Rating - Your Rating)/400))

Thus, for example, if you have a rating of 2000 and your opponent has a rating of 1600, you should win ten out of every eleven matches.

Now, let's talk about earning prizes. Many people apparently were"going infinite" in the old 8-4 queues, but are supposedly no longer able to do so. If we suppose that one of these players was getting an average of ~3.33 packs per draft, we can say that his approximate win rate in rounds 1 and 2 was 75%. We can then calculate his new expected payout, based on the same win rate, and instead of 3.375 packs per draft, he is now getting only 2.34375 packs per draft.

(Note: I assumed that he will win 50% of the time in the third round, where the competition should be significantly steeper than in the early rounds, on average).

So yes, this player is short an entire pack as compared with the previous system. Of course, he is still likely to be able to"go infinite" - except now he will need to sell some of the cards he opens.

One thing we are not taking into account is the difference in competition, though. This is where rating comes in. Before, there were two queues (at least for Onslaught Block), and the 8-4 had a generally higher level of competition. I can't say for sure what the difference was, with no data to back it up, but I will hazard a guess and say that the average rating of players in the 8-4 was a hundred points higher than in the 4-3-2-2. In this case, a player who can beat the 8-4 at 75% win rate would have had an 84% win rate in the 4-3-2-2. Of course, now the 4-3-2-2 is played by all players, so the actual average rating is higher - let's say seventy-five points below the old 8-4. That's an 82% win rate, which corresponds to a payout of 2.65 packs per draft.

Now, let's look at someone who was not a top player before, but who enjoyed the occasional success and wanted to play for higher prizes at higher stakes. That player was probably doing no better than 50% in any round of play, though there were obviously many who did considerably worse. This player was getting approximately 1.375 packs per draft before, and now gets 1.76. So clearly, the conclusion is that the payout change is only a problem for people who were constantly winning 8-4 drafts before. (Actually, anyone who used to play 4-3-2-2 is worse off with the stronger competition from the 8-4 players, but there are many fewer complaints about that.) You can calculate the level at which the change would be bad, assuming the seventy-five point difference in competition. In order for 8-4 to be better than 4-3-2-2, you'd have to be able to win 59% of the time in 8-4 (which is 69% in 4-3-2-2). Of course, the extent to which the change hurts is considerably lower near the turning point - there is not even a third of a pack's difference until you reach a 67% win rate in 8-4.

Thus, we can conclude that only a very small percentage of those who dislike the changes are actually hurt by the loss of pack - an assumption backed up by Charles Mousseau's assertion that he would prefer to play 8-4 despite not being good enough to go infinite in it. In that article (which is misguided, but worth a read), he touches on the real issue, but never comes out and states it. The real issue for players angry with the change is that the prizes no longer seem enticing, regardless of whether or not the actual amount they take home is better. The difference between the two types of players Charles mentions is that one type is pleased to win for its own sake (or the ratings gain), but doesn't want to spend a fortune for the pleasure, while the other type is there to gamble and is pleased with victory only if it means a big payoff.

There is also a third type of player, though - one who doesn't currently draft much, if at all. This type of player wants to play, but doesn't have enough experience to do well and isn't willing to spend a lot to acquire that experience. This player is completely turned off by the"pay $12, play one game" aspect of the current system.

So how can this all be fixed? Well, first of all, what is our goal? Wizards has several goals, all of which they addressed with the current set of changes, but all of which could be addressed better. These goals are:

1) Get the maximum number of people drafting. This means encouraging people who are afraid of sharks, people who are wary of the price, and people who are sharks themselves.

2) Have the draft queues fire frequently enough that people are able to enter the one they want and play in a reasonable time period.

3) Give some meaning to online ratings so that there is an incentive not to constantly change accounts.

4) Stimulate Constructed play.

Let's talk about each of these goals individually.

Goal #1 is met in the current system by reducing and flattening payouts in the main queues, while offering a higher-stakes queue for good players. In addition, the reduced payout results in no split in the finals, which reduces the number of drafts per day that a player can participate in. While the replacement of the 8-4 queues with 9-5 queues that very rarely fire has definitely hurt the"shark" population, the ability to play 8th/8th/8th draft with a 4-3-2-2 payout instead of the old 8-4 7th/7th/7th draft has given many new players a venue to learn drafting which is both more forgiving to play and more forgiving in payout. Overall, this change has been positive - and before the most recent spate of crashes, the 4-3-2-2 queues for both OLS and triple-8th were firing quite regularly.

The current changes, though, do not address all the aspects of Goal #1. Not only are the"sharks" hurt by the essential non-existence of the 9-5, but the gamblers are also less interested in playing due to the lack of any top-heavy prize structure available to them. In addition, the third type of player - the one who wants more experience for their entry fee and is less concerned with big prizes - is far from satisfied.

Goal #2 is met by lumping everyone into what is essentially two queues: 4-3-2-2 triple-8th booster and 4-3-2-2 OLS booster. This has been quite successful as far as those two queues are concerned, but is a dismal failure as far as the 9-5 queues and other formats are concerned. It is important to keep this goal in mind, though, because it tells us that we cannot expect too many regular queues to be able to thrive. At most, I think we can expect three or four regular queues across all formats to fire at a reasonable rate.

Goal #3 is met by the creation of the 1800 room, and is somewhat successful. On the one hand, people now feel challenged to reach an 1800 rating, which spurs them to care about rating points. On the other hand, most people realize that there is really no benefit to accessing the 1800 room, since the queues intended to draw them there never fire anyway. It is clear that the popularity of the 9-5 queues is linked to their usefulness as a drawing card for the 1800 room; thus, we need to find a way to make 9-5 more popular while keeping its exclusivity intact.

Goal #4 is admirably met by the flattened prize structure. The more Constructed queues that we can get firing, the more demand there will be for secondary market cards, thus making the resale value of drafted cards higher and allowing more people to draft more frequently. Properly set up, the Constructed and draft queues complement each other, with more of one making it cheaper to do the other. Also, the more queues that fire, the more money for Wizards, which makes them happier with the system and more likely to continue it.

So what do we do to address these issues? Well, several ideas have been presented, most of which you have probably heard in one form or another. With all due respect to the Alpha Lackey, Charles' suggestion that many different payouts with ticket-only entry fees be created is not a good one. Not only is there not enough room for three or more different queues in each format, but there would then be no secondary market value for the prizes from winning a tournament. Under the system he proposed, a booster pack would be worthless to a tournament player without two other booster packs to trade in for the ten tix.

Swiss draft, on the other hand, is a great idea. First proposed by PapaLorax on the Magic Online message boards, a one-pack-per-win Swiss draft is the key missing ingredient in the current system.

Here are the changes I believe need to be instituted once the server is stabilized:

1) Swiss draft, three rounds, prizes of one pack for each win should be instituted in place of the 4-3-2-2 queue for triple-8th draft.

2) The rating requirement for entry into the 1800 room should be lowered to 1700. This would probably require changing the name of the room to the 1700 room.

3) A new eight-man Swiss queue with prizes of one pack per win should be created for Standard Constructed. This would probably be best done as a trial run to see if people prefer it to the 4-3-2-2. I know there are people who would like to play three rounds of Swiss, but don't have time to play five to eight rounds in a Premier Event. I'm just not sure if the player population can handle three different payouts for Standard Constructed.

Obviously, the main thrust of the proposal is the first two items, though the third item might be an additional way to spur interest in Constructed play. Here's a breakdown of how I think this will improve upon the goals mentioned above:

Goal #1 would be much better satisfied by these changes for several reasons. First, the lowering of the requirement for the 9-5 queues means that the player population would be able to support them. Despite Randy Buehler's optimism, the 1800 room will never contain a significant portion of the population without major ratings inflation. Lowering the requirements allows the good players and the gamblers to have a place to play, without sacrificing Goal #3.

At the same time, switching the 4-3-2-2 queue for triple-8th to a Swiss draft gives players who are learning to draft a chance to play three matches with their deck. This gives them the opportunity to see which cards work well and which ones work poorly, to get the thrill of Limited play without committing to a league or a day-long Premier Event, and to hopefully take home a pack or two in the process. A Swiss draft format would do wonders for addicting - I mean, recruiting - players who are currently wary of drafting.

But as great as it is for that purpose, it is not a good format for the mid-level competitive player. With only Swiss draft and 9-5, the gamblers would have a format to aspire to (but nothing to play while they get there), the new players and casual players would have a format to relax a little and fun with, and the good players would have a format to clean up in... But the mid-level competitive players would have no format to improve in. For this reason, I don't recommend replacing the other 4-3-2-2 queues with Swiss queues. While some people may complain they want to play their format with their prize structure, it is necessary to minimize the number of offered queues for the sake of Goal #2, and the most sensible way to divide the formats is to make the core set the new player format and the expert sets the competitive formats. Thus, everyone has a place to draft, with the 4-3-2-2 providing the perfect training ground for those who wish to play with the big boys in 9-5.

Goal #2 would be maintained by these changes, since there would be only a slight drain on the current (read: before they shut the queues off) player pool. That drain would probably end up comparable to the drain that 8-4 had, but the rest of the changes will hopefully have drawn a significant number of new players into drafting, making everything fire faster.

Goal #3 would also be maintained by continuing to require a certain rating to access the top-level queues. I think this goal could be furthered even more by offering invitations and byes in championship events on the basis of rating and perhaps instituting more events of this nature, a la Grand Prix and Pro Tour events. But that's another article.

Goal #4 is already well addressed by the version 2.0 changes, but the proposed change #3 would allow a trial to see if regular Swiss events could spur even more interest.

In conclusion, I think that the decision makers behind the MODO v2.0 changes to eight-man queues were thinking along the correct lines, but missed a key component of the system they were trying to improve, and underestimated the size of a player pool required to support a draft queue. With the slight changes proposed - all of which would require minimal programming time - I think Wizards could ensure a significant improvement in the quality of the game and keep a much higher percentage of their players coming back for more. They need to do something to combat the negative publicity of the current instability. The beta is a start; these changes would be the finishing touch.

Lenny Eusebi
ravenight on MODO