Reliability is key
When building a tournament deck, above all else, reliability is key. After all, good ideas won't work if you cannot get them to appear reliably. And the first step to building a reliable deck is mana distribution.
Now this goes beyond just how many lands to play with, but also how many sources of a given color one should play with, as well as whether one should use "alternate" sources of mana, such as creatures or artifacts.
Now working in the current TYPE II format, the considerations are as follows:
Most decks, in order to be reliable, will require the ability to play a source of mana every turn for the first three to four turns. This means that on average, you want four mana producing lands (we will consider alternate sources later) in your first eleven cards. This adds up to around 22 mana producing lands in a 60 card deck. Now with the key word being ``reliability'', we may want to up this to around 24 or 25. The reasoning for this is that it is much better to have too much mana at the beginning of the game as opposed to too little mana.
Now for people who have not played much outside of casual groups, this may come as a shock. The traditional ratio of mana cards to non-mana cards has usually been 1 to 2. After all, if one plays with 25 lands, then there is only space for 35 other cards. And as well, what happens in the mid-game, when you're cursing under your breath because all you are drawing is land? Without getting into actual deck building strategies, my answer is that in the mid game, one should either be putting all this mana to good use (e.g. Jayemdae Tome, Disrupting Scepter), or be winning outright. Getting mana screwed is much worse than having an abundant source of mana, as one situation will have you drawing cards which you cannot use (becoming a target for Black Vise), while the other lets you either increase your hand size (to counter The Rack or Disrupting Scepter) or play some extra mana for future use. The point is, you have more options, and when the cards you need are drawn, you can actually play them.
Having alternate mana sources can also help in the mid-game, especially if your opponent is playing with Armageddon. As well, it forces the land destruction player to deal with non-land cards.
As a last point, alternate mana sources help to achieve a certain amount of mana more quickly in the beginning of the game. Even if the sources are disposable (Tinder Wall, Dark Ritual), the advantage gained by an early boost of mana may make it very difficult for the opponent to come back.
For a single color deck, it is usually straight-forward. In fact, single color decks can usually get away with playing with very few mana sources, since the type of mana they need will consistently show up. Note that this only applies to weenie decks (and perhaps black due to Dark Ritual), since their mana dependencies are very low. All-green decks with big creatures tend to play with many mana- producing elves and sometimes even with Nature's Lore and such, and as a result tend to play with very few lands, but overall play with many mana producers. For decks using two or more colors, it becomes very important to determine the color dependencies of cards, and when you are likely to play them. A card is considered very color dependend if it requires a lot of mana of a particular color to play. Cards which are very color dependend tend to be less reliable, since the mana required to play them may not always be available. This is one of the main reasons why artifacts are good, and this is also a big part of why Erhnam Djinns are very dominant in TYPE II play.
Another factor in deciding how much mana of a given color to play is when you are likely to play the cards of that color. If you throw in one Fireball into your deck, and this is your only red card, then it may not be such a bad idea to play with, say, 4 or 5 sources of red mana, the reason being you've only included it in your deck for the end-game, at which point you will very likely see at least 1 source of red mana. On the other hand, if you play with Balance as your only white card, then you may wish to play with quite a few white sources, since Balance is not necessarily an end-game type of card. Keep in mind that a white source of mana need not give white mana exclusively. In fact, it would be very prudent to make those white mana sources give up some other color as well. Being stuck with a plains could be very bad if Balance is indeed your only white card.
Brasslands are very flexible in that they provide colorless mana painlessly. If your deck contains many basic lands, this can be a great boon because you will only take damage in tight situations. Having that extra reliability is always good, though, because bad draws will always happen and it is up to you as the deckbuilder to minimize those bad draws.
As for the City of Brass, it is usually only necessary in large quantities (3 or 4) if you are either REALLY pressed for reliability (e.g. permission decks REALLY want blue mana in abundance) or your deck requires a lot of differently colored mana. Two color decks with a touch of a third color should consider playing with City of Brass, depending on what cards are of the third color. Any deck which uses three colors extensively will probably require the reliability of City of Brass.
To build a successful tournament-level deck, the most important thing to consider is reliability. And the first step to reliability is good mana distribution. If you can ensure that you will never be mana short, then you will actually be able to play your deck and test out your new ideas.
Seyil Yoon wings@io.org