Do we really need initial stats settings for MC? I'd prefer to chose class (in the future some facts from prehistory as well).
It'll do for the @, what we do after that is open to discussion. You sent me some decent example of a game where character was built of pre-game events for example or there are plenty other ways to go.
As much as I like our forest (easy to make too), the number of forest backgrounds is limited enough (unless we'll use real ones).
Forest is not that much easier to make than other options. Backgrounds don't all have to be different either and there are plenty to go around. Workload would mostly consist of scripting events, designing tiles in json (backgrounds, events, possibilities, encounters, mobs, drops, terrain, teleportation and etc.) and looking for more backgrounds. Not mentioning that we also have mountains, beach, other cities, villages, roads and who knows what else. It's not really limited to any specific terrain type. We could have some form of teleportation setup so hero can run the brothels while exploring or just leave matron to run business as well.
Pros:
- Uses our BE for encounters with all resources and any further work that we put in it for the Arena and any other on scene encounters
- Very easy to randomize
- Easy to explain
- Can use insane amount of VN functionality
- Possible to build the whole world around it
Cons:
- No "real" tiled dungeon exploration
- Fairly simple design
- Relatively large workload for something simple
- Maps would have to be built in json as well or a custom software would have to be built for this
Alkion version = Kamidori. Not too original, but clean concept.
Yeap, uses Python code to cut a map from auto-tiles... attacks are animations built on a separate pop-up screen.
Pros:
- Randomization is not very hard to achieve
Cons:
- I'll have to port our resources made for Jake's BE into pop-up animated screen
- Maps are built in text (like literally txt) files.
Your version is basically a trpg, for example like Final Fantasy Tactic Advance or Fire Emblem.
Pros:
- Uses the same resources as the Arena/Encounter BE
- Works off mature mechanics compared to Alkion's BE allowing to create maps from different tile layouts and even 3D tiles
Cons:
- Much harder to randomize, especially will be weird if we use pretty maps for custom and simple for random
- Those mature mechanics come at a price of more complicated code than Alkions version
I guess we'll use tiles for EE in one way or another anyway.
As for BE, if we'll use your/Alkion approach, shouldn't we teach AI how to move on the map, avoid obstacles, approach at the correct distance? Whether there are ready-made algorithms, or it means 2 additional years of coding? 
There are plenty of ways to go, not just the three above.
There are EE options without tiles. For example an overhead map/chain of maps of large areas like in Alkion. Basically we have made a decision for BE already I think... Now we need to figure out:
E = Exploration
D = Dungeons
and make it all seem like parts of one whole and intuitive design. It'll take a long while I think, especially with all the normal Sim functionality we have planned that has priority and the code review.
Another important topic is Map building if we decide to go with something that uses tiles:
1) Map layout:
- Tile layout, often built from different tilesets
- Logic layout, more info for the tiles (walkable, trap, special and so on)
- Unit layout, enemy, items, event placement
2) Tilesets:
- There are many different ways of building maps from tiles. There are many tile layouts as well...
- Auto-tiles (RPG Maker type) that use very simple sprite that is then cut in tiny bits to make large number of complex patterns
- Scenery
3) Data:
- Software like Tiled is used to create a map
- Maps (and their layers) are created in plain text in txt/json/xml formats
- Some form of software that can write into those files is used to create maps from pre-determined types
========================
Logic for BE is another topic for discussion... I don't know how long will all take since I've never coded something like that before. AI logic for some reason does not seem to complicated to me but maybe I just don't see all the variables yet.
