The skill of a clan is given by individual skill of the players and the level of team-play or cooperation they can put together. The cooperation can be also divided into routine cooperation and adaptive cooperation.

Routine cooperation is something that a clan learns during the practicing and uses without any change during the game. Known approaches are used in known conditions.

Adaptive cooperation is the ability of a clan to react on a new, unprecedented condition in the game with the most appropriate action quickly and efficiently. The players don't have any established routine on the current situation, but are able to quickly adapt cooperatively - mostly by a quick and efficient communication, combined with common sense.

The most interesting games are those where all the three factors of the clan skill come in play and depend on it. That is when the game isn't dominated solely by the individual skills of the players, perfected routine cooperation cannot replace the lack of ability to adapt and vice-versa - a clan which doesn't have any routine knowledge about the map should be affected by it.

If the performance of a clan doesn't depend on some of the three clan skills, that particular skill becomes irrelevant in the game and the game gets less exciting - both for the players and the audience.

Small map pools
Very small map pools (for example 1 map only) extinguish the need for both adaptive cooperation and routine cooperation. With the map getting played long enough, a clan gets something close to "absolute knowledge" of the map - that is they know what to do in every possible scenario - ability to adapt to something new becomes irrelevant as there never (or very rarely) is something new to react to.

Also with high enough amount of practice games the number of clans that reach the "absolute knowledge" increases.
Then the only thing that differences top clans is the remaining factor - individual skill of the players.

Large map pools
Too large (infinite) map pools emphasize the routine cooperation too much. A clan needs only a small amount of time to get "at least some" knowledge of a map totally unknown for the opponent to gain domination on that map. The other two skills become very insignificant, or at least players spent most of the time getting trivial knowledge of many maps - something that is not very interesting in a tournament environment.

Adaptive cooperation comes in play only when clans are not able to surprise their opponents with unknown maps and it's not possible for too many clans to get something close to "absolute knowledge" of majority of the maps. That is - when map pool is both not too small and not too large.

Most of the opposers of the "TB5" map pool don't want the adaptive cooperation to be a significant skill of a clan. In their opinion the game should be decided only by the knowledge of the map (= known team-play situations) + individual skills.

I think that is the keystone in the current "TB3 vs TB5" debate. There are many other aspects for and against certain maps or for or against TB3, but the core of the argument I think always will be in the three skills.
The knowledge of the TB3 maps nowadays got very high and is removing the adaptive cooperation skill significance.

In duels in QuakeWorld the trend is to emphasize individual skills together with routine knowledge of a map (by a single player). The adaptive skill was more or less removed from the "standard tournaments" and appears nowadays in the "Kenya" tournaments, where only new maps are played - players don't have enough time to get superior knowledge of the map, as the map name gets announced only a few days before the tournament. What mostly decides the game there is the ability to adapt. (That's why I recommend KTK to everyone - it's a great new experience.)

In my opinion for 4on4 the course in QuakeWorld should be different. Games should not depend only on individual skills and routine cooperation, but the adaptive cooperation factor should matter too. It's the core of the "team-play" - the ability to cooperate in many ways - both by known routines and in an adaptive (dynamic) fashion. It should be made sure that getting near to absolute knowledge of majority of maps (2 out of 3, 3 out of 5, ...) becomes an unreachable goal.

---

Note: The procedure to select these maps should be to let the players vote for top 5 popular maps. That is because everyone should have a fair chance to assume which maps have a potential to be played in (future) tournaments and have a chance to practice most or all of them in advance. Such selection also brings the most "fun factor" as most of the players are playing their most popular maps.

Therefore it is not important if it's gonna be TB3+cmt1b+cmt3 or any other combination of maps. What is the "most popular five" is always good.

P.S.: When increasing standard map pool size (like now in EQL10 from 3 to 5 maps), typically two problems occur:
1) some clans don't practice new maps at all
2) the level of the game-play on the added maps is significantly lower.

The first problem is solely the individual problem of the given clan. Every clan has the same amount of time available to practice and get better. This approach is absolutely fair and the problem is in my opinion irrelevant for the debate.
Second problem is relevant and I think cannot be avoided anyhow. In the long term, however, clans get more and more time to practice and if the map pool is stable, the level of game-play can get very close to the "TB3".