Last year, qz and I were on a couple of threads talking about building a REST api for the game that would make rebuilding the UI and app-ifying the game easier. We haven't got much done beyond mapping out the spec for it, but I spent a little time over the last month or two starting on a better UI framework and wanted to share it for feedback as I move beyond the POC phase and into the design phase.

Here's some screenshots of the desktop, tablet and phone experiences. It's got an example Build page, Public Market page and some screenshots of the War Room attacking flow.
http://imgur.com/a/K1sKt

So under the hood, it's built with React and uses Redux to control the state. Using React-Native, we should be able to run the web and apps under a single codebase and single API, which was a big part of what made me want to actually put time into developing this. :)

Design-wise, we've used Google's Material Design as the jumping off point for this design. It's still a little bland right now but the general concept is an improvement over our current UI and a major improvement over the mobile UI, IMO. The tablet/phone size menus have worked pretty well on touch devices from my limited personal testing. It's pretty neat to flip through pages -- the static screenshots don't do it justice, but it's all I have for now :p

Anyway, if anyone has an feedback or comments feel free to reach out or post them here.

Thanks!

looking good pang, thanks for putting some effort into it

It lacks polish but it's very clean and a significant step up from the current UI, especially the mobile iteration. I'm actually kind of excited.

Ubuntu 12.04 is out of support and you should setup TLS, like LetsEncrypt or sth.

As for the topic and the design, is the intent to diverge from the classic color theme or to try and work it in there? I'm like 80% confident you are diverging given the new icons.

Direction-wise, obviously a great step that you're taking. By basing onto a REST API for the game, where are you planning to draw the line of supporting non-EE clients?

And while on the mic, is qz's https://github.com/s4lt3d/EER the closes thing one can come to completely offline-simulate the game for ML purposes? (like, GPU-scale parallellism and iterations)

k thx bye

looks awesome, feedback:
quickbuild, should have a option when holding the buildxx to change to build 2 turns,3,4,5,6,7,8,9,10 and so on.

change icon from a car to a tank for war?

plx delete

Does this means ranks will be real time updated?

Looks brilliant

That looks pretty cool Pang.


Turning it into a phone app (if I understand it right) will be a great step and could probably spark quite a few players back into playing. (That's a guess lol not something I know for fact).

Definitely an improvement!

Thanks for the feedback, I'll address a couple of issues for discussion ->



I'm trying to keep some of the classic look and feel + colours but would making it more modern. It's a Material-looking app right now more than anything else, we'll massage it a little once we start getting feedback and put more energy into polish.

What actually drives me to stay somewhat engaged and put new effort into development is the opportunity to create a robust API in order to be the basis for university/college computer science classes. Specially, for group projects related to AI, design or building front-end systems integrated with a remote API.

With that in mind, we'd probably open bot competition servers (API rate limits apply, so you have to code efficiently :D) and get some crowd sources bots to help play ghost accounts/add fodder for certain servers. But those would be run by us in concert with the developer.

I'm not sure what ML stands for in your post, but I'm not familiar with the EER project, is that qz's or something someone else wrote on the bot API we released a few years ago? Right now, there is no way to get a real offline EE experience that I'm aware of. Part of the work I want to do with this project is work on containerizing and setting up the whole system to be more scalable (not that we need it, I just want to improve my skills in that area) so we could spin up custom servers for people if they ever wanted that feature. Is that something that is a request right now?





That's a good idea re: # of turns to build with. Maybe a slider or a dropdown control.

I need to commission some icons from a friend of mine, lots will be changing :p





Based on the way we coded rank that wouldn't change. Computing rank is an expensive operation computationally so doing it in real-time would bog down the system and make every request that needs it slower. Sorry!



Any other feedback is welcome -- I'll post some more when I get more worth sharing. :)

(I'm not sure how to quote a quote post so a non-quoted reply)
Pang, thanks for your response!

With ML I am referring to Machine Learning. When using GPU's to run the number crunching, you need very short latency for the ops or the simulation will take a very long wall time and you won't see learning happening as fast as possible. So the most efficient way to implement machine learning is to essentially encode the simulation rules into the code that runs on the GPU itself. You can think of it in a way that it can run like a million games/countries/whatever in parallel when it's training. If you would attempt to set this up but have the backend be your online-servers, your bandwidth bills and CPU utilization wouldn't be happy. So clearly online has to be rate-limited since you have a limited operational budget. Thus, the only truly viable way to do the ML thing efficiently is to have local simulation.
This depends on building an efficient implementation (C, CUDA) of the game, either single-country-only and lose some aspects, or full-game with market/war/other countries. Regardless, it depends on having access to a complete set of up to date game formulas.

You asked if this is a request. Well, no? :D I have toyed with the idea of using this as a sort of challenge when I'm now learning AI and ML. But there is a huge barrier to entry in well, needing the above mentioned infrastructure before even getting started. But it does sound not too far off your containerization idea, which I like and that could serve as a basis for someone to possibly make a instruction-optimized implementation (to train faster). This all relies on you being able to actually completely open source the game engine. I think you should, and I'll posit one argument in favor: the strength of earth empires is not in the secrecy of the code, but the (unfortunately currently small and shrinking) community, and that can't be copied or replicated. I suppose it is possible to release it under an "educational only" license or similar.
So an outspoken request? No. But barrier-to-entry to further ML research? Definitely. So something in that direction could help enable your idea of AI class room game challenge.

The EER project, well, it's something I found. It looks a bit dated, R code for simulation, but seemed to have a not too shabby implementation of the game logic. *shrug* I've seen the PHP bot too, but that's only client side logic essentially, it has no replication of game logic for offline simulation/training.

A well-trained AI deep-learning neural network bot would probably beat a human 1:1 with no effort at all, FWIW (Terminator?). :]

Food for thought?

The online AI servers role is in the above context where the different student project AIs could meet to face off each other.

Another feedback for added optional depth to the game: Let market be transparent, i.e. show all buy/sell orders (and for-sale goods). This allows for deeper market analysis and gaming. Not sure if anyone else is would like that though. :)

This sounds like a great step in good direction. I am excited, I would definitely be interested in a bot competition server :D.

Oh, machine learning. Of course :p

I doubt we'll allow colocation or the downloading of a simulation but if someone wanted to do a research project or something we could see about pricing out instances in a datacenter which would get around some of those issues. At that point, API calls could be made more quickly without since they'd be local to each other while being separate instances.

Related to market transparency, keeping it opaque is actually a a feature IMO. A lot of the limitations on mechanics are meant to increase the challenge :)

ttt

white text on grey background might not be best choice unless text font is bigger.

The only thing I dont like is the menu bar on the left I think it looks to bla.

Nice.

Don't forget my 'Attack Again' button. I donno what I'd without it...

Really, it looks light and easy. I digs very much. Good work!

Attack 5 times could be a thing ;-p

for PS only ;)

Look at this game and see how the graphics are done - Diplomata The Game. Maybe you'll find some ideas because the games are similar. The game still does not have a mobile version.

digging the mobile update, especially the bulk build. I think you should keep the black background though.

I would love to see us keep a 56k modem no graphics version. Its great for me to be able to use it while im flying on a plane

Yea, I toyed with the idea of graphical components years ago and decided that it's mainly fluff for this style of game.

I'm not opposed to making things more graphical but it would need to help tell a story or indicate things -- just tossing in graphics for the sake of making things graphical doesn't improve the gaming experience.

Things like a better advisor graphic and better use of iconography are high runners, though!

how about titties! if you don't get DH on hits the nipples get perkier ;) and if you do DH the titties get saggy and baggy :(

this is supposed to be pg-13 game (yeah right).