MMO vs Decentralization

Ah, good old days of private UO servers with tweaked game mechanics, added items, added skills, added graphics...
More and more games tend to pop-up with a decentralized model where you can open your very own private server and customize it with mods, based on the "Minecraft" model. (CubeWorld, DayZ, Star citizen...)

Is the MMO/centralized server model dead?
What's the advantage and drawbacks of both?

Decentralized:
+Reduced costs for game editor
+No monthly fee
+Ability to mod/customize the game

-Servers can close anytime
-Less people on server
-Can't freely hop from server 1 with mod A to server 2 with mod B
-Can't monitor servers admins (What if 10 overstuffed people hop into your server, destroy everything and lead to the next one like crickets?)
-If you monitor servers, tendency to have less "official/compatible" servers because people feel like they're volunteering/paying without any benefit. People want to play god.

Centralized:
+Server stability
+More people on server
+Fair to everyone

-No mods
-If servers close, it's done (unless someone shows up with an emulator)
-Costs for the game editor
-Possible fee for the players (got to cover the costs of employees + infrastructure)

In the end, the difference lies in persistence.
Do you play multiple "games" of that game? (That's how I used to play DayZ)
Then go for decentralized.

Do you want long term cooperation and teamplay achievements?
Then go for centralized.

Fuck this gay earth.

Bullet Physics is a great lib poorly documented.
I was working with a wiki example of ContactResultCallback derivated struct and I copy/pasted code from the wiki, that works for bullet 2.76.
Fun fact: doesn't work for bullet 2.81 because they added a Wrapper for the btCollisionObject, which is quite simple to update since all it requires is to add ->m_collisionObject to the example.
Yet, in a tutorial, nobody thought of updating that.
So I logged into the wiki to update that annnnnd:

The connection was reset
          The connection to the server was reset while the page was loading.

Fuck this shit.

Think big, start small, grow fast.

Stopped working on Zem, started working on a small platform game project.
Why is that? Simply because it's smaller and helps me to understand how to integrate bullet physics gracefully, how to make some nice shaders, animate one mesh,  organize my objects in classes (and maybe write headers one day and stop that "everything is public/everything goes in the main.cpp" habit).

Every bit of code is thought to be copy/pasted independently of the rest.
So far, character moves: http://www.youtube.com/watch?v=xos2MnVxe-c

Integrating bullet physics made me re-think the whole thing and now I think less "mesh" and more "physics" oriented.

Fun dev fact: you lose more time trying to integrate something badly documented or not adapted to your system rather than developing your homebrew solution.

Some cities/rivers/lakes generated:

Current GUI progress:

Currently working on the terrain shader.

Main menu GUI, first try. I don't like the logo, will probably rework it.

GUI/Inventory progress

There will probably be 3 sizes of items: small, medium and large.

Example of damage indicator:

Finished an alpha version of the buildings builder.

It's not bug-free but it works well for now!
Finishing this paradoxically increased the to-do list:
Texture atlas, X and Z rotation, more shapes...

Download it here: http://triscales.com/zem_builder.rar

Map customize menu alpha (lacking amount of trees, rivers, lakes, mountains slope/inclination, loots...)

Made a logo.

GUI.

We all remember this:

Since there are some irrelevant parts, I modified it a bit, but left the watermark:

 Added the stats/skill tab:

 Added a belt:

Face merging, significantly improving frame rate. From 350 to 2300 FPS.
In a 15x2x15 map: from 20 to 100-160 FPS.

Work still in progress.

Update: Fixed bugs, works like a charm.

Mountain/Lake generation algorithm. Two parameters shapes the mountain's slope.

There is something messy  here.

City generation algorithm

Update: New road algorithm

Update: new building algorithm.
50% 1x1 ; 16% 1x2 ; 16% 2x1 ; 16% 2x2

Test with a 4500 score:

 Test with a 500 score: