"I like large numbers" isn't a good reason to start things above 0, IMO, because for every one who likes them you probably find someone who dislikes them. I know I do.
...
This is why 3e had 4x skill points at first level, and why PF gives you a +3 bonus on trained class skills. A +4 is big when it's the maximum you can get from ability modifiers.
Find me someone that is using the reason "I like large numbers", and they can defend that position for themselves. No one here as of yet has taken that position.
As for the skill points variants, that's great at 1st level. It does almost make the 2 point guys work--though the scope of what they can do is awfully lousy for their supposed archetypes. But it doesn't scale well. The reason it doesn't scale well is multiplication again. Let's take a reasonably bright fighter and a rogue that went with more Cha as a second focus, just to be reasonable (i.e. leave out the dumb fighter and briliant rogue for this comparison, which would make it even worse). Both have a 14 Int. I'll use 3E numbers, though the point doesn't substantially change in any of the variants. We'll make them both human, too, and assume that neither particularly wants cross-class skils.
The Fighter has 5x4=20 skill points at 1st level. The rogue has 11x4=44 skill points at 1st level. That's a ratio of 5/11. Assuming no change in Int, that
skill point ratio will stay constant. Let's say that they want to maximize 5 and 11 skills, respectively. This will also stay constant. I believe this was the simple view of the math when the system was designed. A little better than 2:1 for smart rogue versus smart fighter doesn't sound that bad, especially with the rogue having a few extra things he needs to do his job.
However, when examined more closely, it has two major problems:
1. The above is pretty much the best case scenario. If you are, say, a dwarf cleric of average Int, you get your 2 skill points per level, and you'll like it.
Better make that extra 6 at first level count.
2. But more important, the ratio of skill points is the math, but
the ratio of "useful skills" is the thing that matters. Let's say that our rogue above wants to branch out in a couple of things. Well, if he wants to sacrifice one measly level in all of his skills, he can get two extra skills at 5 and 6 ranks (+ ability mod). Then he can maintain those skills at 11/13th effectiveness compared to his level (overall). Meanwhile, our "good case" fighter can sacrifice that same level worth of skills to get one at 5 ranks. Then he can maintain those skills at 5/6th effective compared to his level (overall). That's not bad, which shows what happens when you start at around 4 or 5 with one of these numbers.
Meanwhile, our dwarf cleric can get another skill at rank 2, under the same circumstances, and he can maintain his now three skills at 2/3--or in other words, he can just about nullify all his skill effectiveness over time by doing so. That is,
he can't really afford to add a single skill.
Now of course cross-class complicated that (but got cleaned up some in later variants). And certainly feats and magic item bonuses can help (and the fighter certainly has those.) And the way an Int penalty interacts is downright criminal--and 8 Int is a skill pariah--and as bad as an Int 3 in a lot of classes, because you can't get worse than 1 skill point per level.
It has other drawbacks without other changes (and dealing with those are not always easy), but you can largely handle this particular issue by simply giving every 3E class +2 skill points per level. Now the worst case, average Int base is 4, which means that such a character can add a single skill while maintaining skills at 80% (4/5) effectiveness, which isn't enough off of the rogue's (new) 13/15 for adding 2 skills, to quibble over.
This is an example of hidden multiplication in effectiveness that is not seen by a cursory examination of the numbers. As I said before, unlike the Hero System SPD example, the effects in D&D are often subtle.