| Uleaum |
I have always wanted to play on online version of a points based game like GURPS or the old Champions based system from Hero Games. I will assume by “skills based” game system you mean something like those. The alternative would be system where the click fest of a weapon maxes you out in ability and everyone ends up the same. The points form the basis for the choice of your advantages and limitations, so I will refer to a points based system for clarity.
The problem with a points based system is a Gamemaster has no way to identify the weakness of any character without examining the build of that character. Level based systems reduce the effort of the Gamemaster by assigning strengths and weaknesses based on class. In an online world the Gamemaster is a machine, so a points based system must be an algorithm. If you are lucky and use lots of points to smooth out the relationships between advantages and disadvantages, you will be able to balance your system just using Calculus, differential equations, and linear algebra. You should probably expect to use Tensor mechanics and Graph Theory to get good results. If you don't use enough points, and the system is chunky instead of smooth, you will need Combinatorics and other discrete mathematical methods to balance the system.
The points based system ideal of pen and paper RPG's has never been realized online because it's almost an artificial intelligence problem. I was in the Champions Online beta just long enough to realize Cryptic Studios got one of the most advanced point based system ever designed and discarded it for a City of Heroes part 2, A la cart system. I gave up when having a very high CON could determine how hard you hit, so players wouldn't be penalized for designing bad characters. They're still running into the problem of a point based system being so complex that players also need a more structured advancement system because they can't figure out the min/max of the builds Cryptic offered.
I really hope someone succeeds in developing a good points based system for online play, and it would be wrong to mock you for trying such bold and inspiring goal, even though you are small start up developer and no one has ever really pulled it off. On the other I find it ironic that the Piazo business model is based on Wizards of the Coast discarding one of the most advanced and play tested level based systems on the bet that they could invent a better system, so HAHAHAHAHAHAHAAAAAAHAHAHAHA! (It was going to happen, I just wanted to be first.) I believe the technical term for this is the Fourth Edition Mistake.
The feat based system of the 3.5 variants gives the ability to minimize class weaknesses but still remains balanced. It's a system many people are familiar with. Once you build and balance a new combat system and figure out to train players to use it, then try and figure out how to implement in an online game the parts of 3.5 no one has yet, the nontraditional combat and conflict system. If you're going to do stuff like making sneaking, stealing, fasting talking around some boss combats, and professional crafting skills, you might as well do the one thing that no one else has done-the Wizard.
The Wizard's special advantage is the ability to change advantages and disadvantages. The class weakness are large numbers of opponents (cavalier followers), immunities (monks and paladins), and prepared opponents(special gear). Immunities should be confined to 1 or 2 types of gear, like a helmet so a Wizard can win so long as they have the right spells. Some people just can't get past the fact that a prepared Wizard should always win. The time of preparation is really not a lot different from a warrior grinding out for superior gear or a thief taking extra time to sneak. Catching the Wizard unprepared means tracking him to his lair or sending large group after him. If you can play balance this you will have a new innovation in online gaming. You would need to add all the spells in the book, not just the combat ones, like divination (to figure out where a wizard is or who a murder is) If you can play balance a Wizard, you can play balance anything. That could be the core of your game.
With the sandbox idea, players would be responsible for developing your content. If you had an algorithm for play balance, then the design of quests and dungeons could be relatively automatic. Instead of having to craft every item a player could design a piece of gear they wanted and then build a instance related to the gear. Area knowledge or Archeology might be used to hunt down relevant items. Sorcerer based gear might require going through a series of monsters that each had one or two of the elemental immunities. Under OGL you get to keep plots and creatures. A plot generation mechanic should fall under this, so there really is no absolute reason to abandon the OGL system. If you don't do this, someone else will.
Flight, water, mount, and vehicle(ships and wagons) based combat should be allowable. Use engineering and architecture skills to develop fortification and bridges for trade and mass combat situations. Make languages relevant. Find a specific way to use every skill in the Pathfinder Rulebooks. Use in game social constructs like bounty hunters to regulate economics and player behavior. Make noble titles and social institutions for players, and part of the game is to attack, topple and assume the powers of the realms, through force, deception, technologies, logistics, and inspiration, while having a mechanism to keep the realm borders stable.
If your end game is the social game, then total free play is possible in end game content. Being a noble, or playing an advanced class might still require some type of payment, but a fighter could play as the member of an order without ever having to pay anything. Make sure that in game gold can be used buy good things, especially for premium players, but the subscription fee is always cheaper than the cost of gold for V.I.P. Subscribers Two premium maxed out character slots should cost the same to maintain for a year as a subscriber pays to maintain all characters in year. Subscribers should have their own server and be able to transfer characters easily. Top level leaders should be Subscribers. There should be some form of player elected goverments.
If you're going to create a gamified social experience you should practice with your development model. Crowd source your research. Ask players to compete for providing the most or best information about a topic. Raise money by running adds on your forums and give away prizes for finding the best exploits in your game ideas.
You should consider designing every visual aspect for multiple resolution levels. Players with high video cards should be able to have photo realistic experience, but lower end cards should be allowed for play. If you're going to do mass combat, at some point past a clip plane, the video rendering might even be just 2D sprites so you can do mass combats for everyone.
That's my 2 cents.