Yeah, yeah - there have been plenty of discussions on this in the past... however rather than talk about how to do it, I thought I would tell the story about the development process.
In the next couple of weeks I plan on finally starting the servers up for my first Roguelike, Rogue Element RPG. Shameless plug - you can see some screenshots on the Facebook page: https://www.facebook.com/pages/Rogue-Element-RPG/443112972508604
It all started in about 1998 - 17 years ago now. I wanted to create a turn based multi player roguelike. And so did every other RL player. I read through many of the original emails and find that quite often people have great ideas about what THEY want developed in a game, but few actually have the coding skill to actually write the code required (let alone the time to dedicate to such a project).
Often you come across a huge number of challenges when coming up with something new or innovative in terms of a game. Rogue Element RPG has been through about two complete re-writes from scratch. It is a hard decision to make, but sometimes you make fundamental assumptions in earlier code that can not be fixed. Some people would say you can avoid this with good planning, however when writing something so different there are no guides on what the algorithms look like.
Sometimes you need to take a break from programming. In fact often the solution to a complex problem is very simple, you just have to give yourself time to think about it. Initially the movement in Rogue Element RPG mean the use of real time when players were physically close on the map. It took quite a number of years thinking about how to come up with a solution that allows players to remain turn based the whole time.
About a week ago I came across a problem in gameplay that could have resulted in major problems whichever way I looked at it. So I took a break and worked on some other things. The solution was ultimately very simple - but it was only by taking a break that it was solved.
Be ambitious, but not too much. I have heaps of grand ideas for the future development of Rogue Element RPG. I have added support for multiple languages, 3D graphics and a variety of other things. However many of these things will take time to develop. If you are too ambitious you will spend all your time working toward something that will never be ready for the big time.
But take a risk. Do I think my game is ready for release? No. There are bugs. The server still crashes from time to time. There are things that annoy me. The monster movement code is not fantastic. But sometimes you need to put your game out there and see what happens.
People will not like some of my decisions. For instance I am going to charge a small amount for access to the server(s). But in return, players will be able to be rewarded in the game as well. Permadeath is still enabled.
Use another game as a basis for what you want, but when you are comfortable, do not worry about drifting away from that game. Rogue Element RPG is based originally on Nethack 3.4.3. However in the last 12 months I have removed anything that strictly ties my game to the original (mainly for copyright reasons). Your game can be as unique as you want, but if players are familiar with what your game is based on, they will likely find it easier to adapt.
And finally - don't give up. Rogue Element RPG is my 17 year roguelike. I have spent countless hours coding and trawling through code dumps to track down bugs. It is very possible to have a independent turn-based multi-player roguelike. No co-op, no real-time.