Nice to meet you, guys!

We are going to create a game that is quite similar to Google Ai Challenge (aichallenge.org), Robocode () and RoboWar (statico.github.com/jsrobowar/), CoreWar (), MindRover () and CarnageHearts (video.google.com/videoplay?docid=-5659442866583550320) games, with enhanced tactical/analytical experience.

Unlike Google Challenge, this would be a tactical/analytical 2D challenge for EVERYONE, programmers and non-programmers alike.
The gameplay should be focused on programming the logic for one or more "robots" that make certain actions
during the match on a randomly generated map. Players should be able to test their units beforehand, and then participate in a full-scale championship ("Versus" mode) all alone or in teams. There should be multiple winning strategies, that is really important.
The match itself goes on without the player, he is just a spectator (watches his "robots" fighting).
Preferred setting: Open Space.

What do you think is better: making a game that supports common languages (Python, Java, etc.) OR invent some interesting way to implement coding without... actual coding?

Thanks!

Support common languages, people should know where software comes from but its more accessible if you can implement coding without actually demanding a player to fight with the grind of code.

Meta-game tutorial type gameplay could encourage a player to type out/piece together/fill in the blanks/etc for key parts of that code. Indicating that the player is building the code towards fulfilling the actions, then displaying these "coded actions" with icons (maybe with semi-transparent image of the code on each icon). Each icon could explore new aspects of what is involved in the basics of coding.

Then the player builds the AI by dragging the icons onto the grid map to create AI actions and reactions based on the location the robot moves to on the map. The prowess of the robot could be based on how many action/reaction icons the player can place on a single grid location on the map.

The fun part could be programmers can build new actions/icons and expand the game's library of interaction. These could be submitted, judged and then the winning code each week could be animated and added to the game.

That's all I've got!

Thanks, Mratthew!
Really nice idea with a grid, something like CarnageHearts have.

And yes, sort of modding capabilities would definitely look great.

I agree that it shouldn't involve grinding of code, but would still love
to hear from more people.

Provided the game is to remain accessible to non-programmers, there can't be too many different, complex logical "sentences" pre-built into the game, otherwise things might get so complicated that unless you were there at release, the learning curve will become too much for new players.

I think some kind of subset of some widely used languages might be most appropriate here.

Expanding on the idea of being able to build a library, perhaps it should be user specific, starting with basic logic, and allow users to compile those pieces into miniature "programs" or "functions", which can then be saved (perhaps as Mratthew said as icons), and then be further combined and compiled into more complex ones.

Say I'm playing the game, here's a vague representation of what I have in mind:
Action A is something I built using basic logic, I want all robots to do this.
Action B is some new action that I want certain robots to do.
A+B is saved as some new Action C, which I can drag and drop onto robots of my choosing.
So technically I've programmed it, but once I've written the basic code once, I don't have to go back and do it again or make it more complex through typing, I just have to click on the icons and combine them to get the result I want.

"Functions" will definitely work fine!
Thanks, pixelartist.

I agree that too many complex sentences will scare off casual players,
but that's still a challenge. The idea is to find out who is the best in
logic and analytics. That's why it shouldn't be too simple, too... At least
multiple winning strategies should exist.

I think it will likely be challenging for non-programmers. Somewhat real-world/high level blocks may help, e.g. a block for filtering for nearby enemies has two outputs (a list of enemies, or no enemies present), the branch for enemies can be fed straight into melee attack, the branch for no enemies can feed into a non-combat branch such as move left or search for distant enemies.

Thanks, jefferytitan and Dir3kt!!

I've never heard of Scratch, but it's awesome!=) I'm sure we'll get some nice ideas from it.

jefferytitan, it will likely be challenging for non-programmers, but I'm afraid the only way to attract more
"true coders" is by letting lots of people spread the news! If anyone can do at least smth in the challenge,
then a good coder will likely hear about it from somebody.