Charles, zzo38, thanks for the comments. Each points to a design feature I
found essential for controlling the playability of the game. First, I'll
speak to zzo38's googolplex of piece types. The Warlord scenarios use 5
piece types, just like the original game Chieftain Chess. To achieve
playability with very large numbers of pieces requires, I think,
simplification of the piece types. As variant designers, we often add a
"new" piece or pieces for each extra square or pair of them we use to
place the additional pieces on. The very large shogis, I think, represent
the futility of that approach. Proliferation of piece-types clogs the
analysis of moves unduly. 

For my truly large games, those approaching the 100x100 "goal", I'm
offering an optional 6th piece. But unlike the other 5, it is not a unique
piece, but a combination of existing pieces that fills a need in that size
game. And still, no piece moves more than 3 squares in a turn. How you
manage playing a game of this size in a "reasonable" time - which, for
the sake of argument, is a couple hours for the smallest sizes to a couple
weeks for the largest - answers Charles' comment about "huge" games
being those larger than 150 squares, more or less. 

The trick, of course, is organization, and on 2 levels, one being
operational and the other conceptual. Leader rules allow organization of 5
- 10 individual pieces into one "superpiece" which moves and fights as a
cohesive whole. The individual units within the superpiece essentially act
as hit points representing the offensive and defensive strength of the
piece. The specific combination of units within one superpiece dictate the
tactics used both by and against the formation. 

Seeing the game as a diceless wargame allows the players to organize their
thoughts in a highly useful way as they compete for victory. The Battle of
Macysburg uses 84 pieces/side To fight a "3-day" battle for control of
the center of the board, where the city of Macysburg is located. Each
side's pieces are organized into 4 corps of 3 divisions each. Each
division has its own leader which can activate the entire division each
turn, provided all the (surviving) units start in command control each turn
or can be brought into command range by leader movement during the turn.
Each division also has a detachable self-activating unit which can freely
move anywhere on the board. The corps come onto the edge of the board 1 at
a time, the first at start, the 2nd on turn 5, the 3rd on 15, and the last
on turn 20. (Turns are organized into 12 move days, with an optional night
turn allowing rally of casualties.) Each corps is 20-22 pieces in size. 

I was shooting for something that felt a little like Gettysburg in the US
Civil War. So the game is organized as a meeting engagement where forces
come on board over the course of time, and from all different directions.
Each player's divisions come on spaced several squares apart, in shifting
locations that encompass half the board for each player. Day 2's units
enter on an axis almost perpendicular to day 1's units. And there are 3
levels of victory. Tactical victory is complete control of Macysburg at the
end of the game. Operational victory is chasing the opponent's army off
the board. Strategic victory is destroying the opponent's army. Grin, and
I can tell you from experience in this one that, after you've massacred
half of each army in about 3-4 turns late on Day 2, you want more pieces.
You find yourself wishing you had another corps showing up real early on
Day 3. 

The rules and components have been through a large number of playtests -
for a chess variant. For a wargame, not so many. That explains why the
pieces work so well together tactically - they actually show the value of
combined arms on both offense and defense - but the scenario quality is
somewhat spotty. All the scenarios I've posted are fun the first time
played. But not all of them have high replay value, in my estimation. Some
of them display a sameness in the games that starts to show up after a few
plays. The game was always fun, but after ~3 plays, you knew how the basic
flow of the next one would go. There were no real possibilities for
strategic surprise, it was always start on opposite sides and meet in the
middle. The game uses variable terrain, and the specific terrain
arrangements affected the details of the flow, but not the strategic
possibilities in these scenarios. 

As I pushed pieces, both by myself and with my developer, Dave, certain
patterns and alignments of space, time, terrain and pieces jumped out at
me. The game is playable at 2 scales, one with full activation of pieces,
and one with limited activation - specifically, the number of activations
is set to ~1/4th of the on-board army size at start, and does not change
during the game. No completed playtest game has gone 40 turns, in any of
the scenarios. In limited-activation games, it takes about 12 turns to get
6-8 squares from your territory into your opponent's territory in a
successful campaign. 

This info let me put together the introductory scenario A Tale of Two
Countries, a 12x24 square game with 8 activations per player turn, 4 home
cities per player, 4 leaders and 32 other units/side at start. It is meant
to play in the 90-180 minute range. A "teaching" game takes 4 hours,
though, based on experience. The games have lasted between 15 and 35 turns,
with victory determined by a player having a friendly unit in an enemy city
at the beginning of the friendly turn. Reinforcements may arrive, on home
cities, on turns 13, 25, 37..., but no one's made it to turn 37.
Replacements, at the rate of 1 per 4 casualties, also arrive on home
cities, 2 turns after they become available. This is a nice, tight little
wargame with very good replay value, even solo, although the game plays
differently solo.