Wouldn't surprise me if this didn't happen, given low participation last year and the proximity of SC2, but I chanced by and saw this thread, so I thought I'd offer an idea...
Last year, organizers decided to make the Mapathon into a collaborative project, to try to get a large and fun map rather than bunch of incomplete minigames. Some people seemed to feel that this didn't leave participants enough creative control, which might have been a reason for low participation, and the logistics of merging multiple entries became a problem almost instantly.
Here's a possible middle-road approach: before the contest (or as the first contest or two, like last year), develop and release a reusable framework. Contests involve creating a game using the framework, but each contest entry is a stand-alone game, so you can change anything you want and you don't need to worry about merging or compatibility with other entrants. We just use the framework as a bootstrapping tool to help people produce deeper and more polished maps more quickly.
For example, perhaps the framework consists of a system for two teams of players to select and customize heroes.
The first contest is to make a game involving territory control, so one entrant makes a king-of-the-hill game, one has heroes running flags from their base to various command points to seize control of them, and a third has heroes skirmishing over resurrection points similar to
Runes of Power. Next week, the contest is to make a game featuring a dangerous environment, so one entrant has the two teams of heroes racing through a valley fortified by undead, another puts the teams in an arena match filled with deadly traps, and a third features a survival contest against waves of monsters.
All the maps feature essentially the same set of heroes and the same rules for selecting and customizing them (possibly with some changes necessary to support/balance the particular game), which means that the map makers can just focus on the core game rather than the large set of options and abilities that make it interesting and replayable, and players who've tried out one map can more easily get into all the others. However, each designer retains complete creative control, can change anything they want, and need not worry about integrating their work with anyone else's.
As another example, the framework might consist of a tech-tree suited for constructing fortresses: walls, traps, sentries, and the like. Different mappers in various contests might use this to create a game where each team builds a fortress and then races to destroy the enemy fortress, a tower defense-style game, an empire-building game where you defend your cities and trade routes against barbarians and bandits, a game where you design a prison to try and contain dangerous creatures for as long as possible, etc.
TLDR Version: Produce lots of complete, polished maps quickly by giving people reusable components and letting them focus just on the part of their game that makes it unique.