Although I consider myself pretty knowledgeable about the series now, I actually only started playing Pokémon proper about three years ago, when, finding myself with an overabundance of time on my hands, I decided to start on Pokémon Ruby, which was then already several years old. What struck me about the game was that it really wasn't very much fun, yet there was a highly addictive quality to the monster collection concept. I then proceeded to play through FireRed, Emerald, and Pearl in rapid succession.
Burnout was inevitable, but I was also disappointed at how little the game really evolved with each installment. For the most part, each generation added just a slew of new Pokémon to catch, while implementing only subtle changes to the technically sound but incredibly old-school turn-based battle engine.
That pattern is inherently catastrophic. As of the fourth generation, we're already at close to 500 Pokémon, not even including multi-form monsters such as Deoxys. Are Nintendo and Game Freak seriously just going to keep on adding? "Gotta catch 'em all" is already an impractical goal at best. What happens when we hit four digits? Will any even among the most diehard fanatics still bother to try?
Personally, I think we've already reached the limit of this design, and it's now time for a Pokémon revolution. (No, not Battle Revolution.)
Aside from the monster collecting, Pokémon has, for me, always been about the battling, so my redesign would concentrate on overhauling the combat engine.
Perhaps the most significant change yet introduced to Pokémon battling was the introduction of two-on-two combat in Ruby/Sapphire. In practice, double battles were rare, and the core games have remained primarily one-on-one, with up to six Pokémon per side tagging in and out. I think that's a shame, because, in my opinion, two-on-two only enhanced the combat by adding new options and encouraging a deeper consideration of the team-building that is so much the essence of the game. As such, before Diamond/Pearl came out, I initially thought going three-on-three would be the logical next step. Now, on second thought, why not just take it all the way to full six-on-six, setting aside the tagging altogether, as team members fill assigned roles within a more organic, simultaneous framework?
Complementing the more crowded battlefield would be a real-time engine to largely eliminate turns. Rather than giving the player direct control of the Pokémon, however, I would prefer to keep to a more strategic approach inspired by the Gambit system from Final Fantasy XII, utilizing preset if-then AI scripts to command them indirectly.
This would mean that the fighting would be mostly automated, but, hey, take a stroll through any online Pokémon forum, and you'll run into heated "theory fighter" debates that make Shoryuken.com seem casual by comparison. Given how much of the fight is determined before it even begins by the setups, we may as well spare players the formality of tapping A to issue commands during combat. That way, instead of mashing A to get through the first several rounds of the stat-boosting "Dragon Dance" with Gyarados, it will just do it automatically according to plan. Unless, of course, it's facing an unfavorable matchup, in which case the if-then preset will have it shifting intelligently to defense. Then again, with so many allies on the field, maybe its teammates could be instructed to help stall, while it powers itself up in the back.
As for the player's role once the fighting begins, the trainer would toss out healing items, issue manual tactical adjustments (i.e. call plays) as necessary, and make substitutions. Yes, in addition to moving to six-on-six on the field, I would also propose adding a reserve of another six backup Pokémon, bringing the experience closer to an actual professional team sport. All of this would occur in real time, so a swift input mechanism (e.g. touch screen) would be needed. The substitutions, in particular, I picture operating fluidly as in a hockey game.
Finally, as a fighting game fan, I've always hated the idea of having RPG-type "levels" and semi-random stats in a competitive game, upsetting the pure contest of skill and strategy. I'll concede that having a team of Pokémon bred and built by yourself can strengthen your attachment to them, but, once in the competitive arena, I would favor the option to normalize the stats and focus more on testing the players' custom move sets and AI routines.
It's foolish of me, I know, to waste my time dreaming about a game that Nintendo will never even consider, but I do believe that something radical must be done with the series to avert a horrible crash within just one more generation. Alas, I suspect it's more likely that they will simply start removing old Pokémon while continuing to add new ones, in order to keep the "Gotta catch 'em all" experience barely manageable according to the same old formula. I suppose, within the fan community of Pokémaniacs who can name all 493 Pokémon, that would be considered "radical," but not in a positive way.