Our characters all have attributes, a staple of roleplaying games since Gary Gygax (peace be upon him). This stands to reason - some people are by nature smarter, some more charismatic and so on, and it seems an appealingly straightforward way of making characters seem different and distinctive. But attributes are supposed to be more or less fixed - so CCP in their wisdom decided that they would affect the speed we learn new skills. They also added learning skills (later removed) and implants, so we could learn faster - and remaps, so we can change our attributes if they no longer suit us. This, of course, removes a lot of the point of having attributes in the first place; they don't make my character feel distinctive, they're just a means to a rather abstract end.
All in all, this a mildly horrible piece of game design. Learning implants (which increase attributes, accelerating learning) have a few redeeming features, but they too are badly flawed in execution and impact. I've learned recently that this is not a universally-held view, so let's unpack this for a moment.
1. Remapping attributes offers no interesting gameplay choices, and requires zero ingenuity. There is an optimal choice, and a range of sub-optimal choices; putting a wish list into EVEMon and pressing "optimise" involves neither mastery nor fun. Even without third-party tools, anyone willing to do the busywork of checking the learning attributes for every skill in their plan can easily create a near-optimal training plan for themselves; it's not clever, it's just a hassle. Compare this with post-Crius manufacturing, for example, where the interplay of job costs, logistics costs, risk exposure and market prices creates a complex system, where dozens of variables have to be juggled to find a constantly-changing range of differently-optimal solutions.
2. The optimal learning plan always comes at the expense of usability and usefulness for active characters. For optimal results, I have to neglect whole categories of skills for long periods; this generally means that I can't take full advantage of the skills in my plan for months at a time. I sacrifice a gradual, incremental growth in capabilities in order to get to my end goal sooner. In other words: in order to progress optimally, I have to have less fun.
This is also the problem with learning implants: they punish risk-taking, as the optimal learning strategy is never to undock. Subcap PvP with a head full of +5s isn't a bold and audacious choice which can pay off for a skilled or lucky player; it's simply a mistake. Again, this mechanism rewards inactivity at the expense of active players; inactive alts can be left with +5s in-station to train, whereas players will hopefully want to undock and do things on their main, and rightly recognise the foolhardiness of doing so with high-end implants.
3. The consequence of all this is that attributes and learning implants offer a large advantage to (veteran) players training inactive alts in comparison with active (new) players training main characters. If I spend PLEX on an alt account or multi-character training, I can train my alt up to fly a freighter or whatever, and do nothing with that character until their training is complete. This is not an option for an active newbie, who will want to increase their capabilities gradually and as soon as possible, necessarily sacrificing efficiency in order to increase short- and medium-term utility.
Sacrificing fun for success was bad enough, but the last of these, to me, is the killer. EVE has long been described as a gerontocracy, and the accumulated resources and skills of veterans will always dwarf the capabilities of newcomers, but the game's design should aim to mitigate that problem, not exacerbate it. Veterans will always be able to take maximum advantage of any game system, but Malcanis' law notwithstanding, the systems should be designed to make it possible for a resourceful new player to gain the maximum benefit without having to sacrifice their own gameplay or fun.
With me so far? Ok. So here's the thing: the idea of being able to customise my character is nonetheless really appealing. Why shouldn't Voth be better at some things and worse at others? The problem is not with the idea of attributes, but with the way they function in the game. Changing the function of attributes could make them more meaningful and interesting; rather than simply removing them from the game, let's use them to make different characters feel slightly, but recognisably, different.