Okay so this thread will be a brief interjection of my idea, whetting the appetite to stir interest in something I have wanted to do even before Skyrim's release.
A few years ago there was a nice modder who released an Oblivion x2 mod. It essentially doubled the size of the terrain (of Cyrodiil).
Play testing this mod made me realize something very interesting (and perhaps it was just my own perceptive predisposition), having realistically proportionate terrain made the game feel better and more immersive. Suddenly forests felt like forests, miles like miles and the distances between cities made sense. (He also made a x4 version that I adored but it didn't load the objects like the other version did.)
I am wondering who else would be interested in doubling, tripling or quadrupling the size of Skyrim to realistic proportions. It would be great for players who love immersion. And most importantly, if we can make this a project early on there will be more mod support for this later if it was released. You could potentially have a lot more area to 'plugin' content from other modders. And it would, in my mind, be a big hit to most PC gamers I would imagine. I anticipate the Skyrim modding community to outclass Oblivion's.
I said I would make this post brief and will discuss further details after I find out how much interest there is in such a mod.
UPDATE: Here is a post that "nice modder" made in this thread.
1. As SevenDragon says, the Cyrodiilx2 and Cyrodiilx3 (and there was even a downloadable 4x) weren't meant to be replacement mods, just successful proof of concept experiments (done about 3-4 years ago). I was curious whether the game could handle such a thing, how would it look, how would it feel, then threw the playable world out there to everyone in case anyone was serious about taking it on - and giving them a monumental headstart in doing so. It was entirely programmatic upscaling - nothing was hand placed:
http://www.projectmanager.f2s.com/morrowind/Tamrielx2/index.html
http://www.projectmanager.f2s.com/morrowind/Tamrielx3/index.html
There was a (now also dead) 2x Vvardenfell (Morrowind in Oblivion) project that came first and was scaled and converted the same way (there were no crash problems with an upscaled Red Mountain btw - there would have been a minor issue with a 2x TES3: Morrowind though as it would not have been able to display the peak of Red Mountain - but it was visual, not enough to cause a crash). Oblivion can render land miles high, 1x Red Mountain is only 1/3 the height of Oblivion's tallest mountain:
http://www.projectmanager.f2s.com/morrowind/2x2/index.html
2. For those questioning why the static meshes were also upscaled; it'd look awful not to start that way; nothing will join - buildings would be made of fragmented walls, pieces float up in the air, clutter would be airborn, tree roots would be the wrong height and so mostly underground - some things wouldn't even show up above ground level, likewise rocks etc. When scaled up, LOD statics are in their correct places. And the wilderness looks fine - with the FOV distortion of the game 2x scale trees don't seem out of place at all. Landscape textures But 2x scale towns, ruins etc are easily replaced by copying and pasting the render from the original worldspace; then they can be improved upon. 1x scale meshes was always intended to be the final solution (for settlements and ruins at the very least) - scaled was the visually correct starting point because you could instantly appreciate the overall world better.
3. Someone said just half the scale of the player; this achieves nothing but changes the visual size of your character when in 3rd person mode, the game still plays the same, your field of view remains the same and the same amount of land still visible. Doubling the size of your character might solve the visual height problem in a rescaled world, but creating a much larger area was always the goal.
4. Oblivion didn't have a problem with 3x scaled landscapes. It was just that the CS had problems generating the DistantLand LOD meshes. But someone now has a solution for that game with a mesh generator which produces better LOD meshes for the vanilla engine, and doesn't have the height clipping issue of the CS.
5. Hundreds of miles of upscaling; please forget doing that with the current TES game engines; they're too limited. Whilst you can load such a heightmap and play on it, it takes many minutes between loading each game (4x Cyrodiil which is 1500 sq miles took 1.5 mins to load up), you'll have dreadful animation jitters the deeper you stray in to the wilderness (bug has been there since Morrowind), and trying to fill it, region generate it and fix bugs, the rest of the world would be playing TES14 (Morrowind III


6. Expanding the world map instead and doing the other regions; Bethesda will always be doing the same, and probably beating the amateur teams to the deadline. Whatever wonders the modding team creates, the Bethesda will be the standard one the vast majority play. e.g. I wonder how the teams of modders who were working on a Skyrim feel atm, and those doing Cyrodiil (for Morrowind) just as Oblivion came out?
7. "7000 steps". When you scale the meshes at the same time as the landscape, then the steps already look perfect in their original positions. If the steps are too steep for the player to walk up (or you actually want to get more steps in) remember that on a scaled landscape the gradient is still the same - so it's easy to snap 2 or 3 of the 1x scale meshes together and have extended length steps.
Enlarging Skyrim
It's possible, and best done soon if there's a chance to sway the culture of modders towards an enlarged land rather than vanilla Skyrim. I don't want to project lead, but I can probably provide a technical starting point, a completely scaled exterior world, maybe more.
Remember that fractional scales are also possible (e.g. 2.5x) - but whole numbers are psychologically nicer.

Such a project must be started with scaled statics as a reference point: Most mountain profiles have are mostly static rock meshes; without them appearing in the right place and size then LOD mountains will be reduced to the more curved style of oblivion. Waterfalls and distant trees the same. Just as it was if anyone had taken up the mantle with a 2x / 3x Cyrodiil, they'll have to replace the 2x/3x scaled building render with that from the original Skyrim - but that's a trivial, though tedious, copy and past job. Then interior<->exterior doors will have to be relinked with the original Skyrim interiors. There's a world of AI and scripts that need doing too; that is if the plan is to scale up everything in the existing world (and not just start with new quests in redesigned towns).
My current version of TESAnnwyn should be able to rescale a textured Skyrim landscape already. But I don't have a working version of TES4Scale anywhere atm, I haven't worked on it for nearly 3 years; I was turning it in to an all singing all dancing thing to upscale scripts, re-link interiors to exteriors, scale regions etc; to all intents and purposes upscaling the entire Oblivion.esm and any TES4 mod file programmatically. Then my virtual PC (I developed on Linux) went belly up and I couldn't work on it anymore. And I never had a working version to go back to. Shame because it would probably have been simple to apply it to Skyrim too. I'll have to dig it out (will have to wait until next week as I'm throwing a big party this weekend) and see what progress I can make saluaging what's left.
Here's some sort of initial plan, which is all quite doable:
1. I'll scale up Skyrim in to a new worldspace using TESAnnwyn at 2x and 3x.
2. When I can get TES4Scale going, I'll shift the static co-ordinates (including scale) of meshes to fit on to the new landscapes like I did for Cyrodiil.
3. Hope to hell the Construction Set (when it comes) has a decent LOD landscape generator, and not that appalling slow (takes days or weeks) one that was provided with the the GECK (Fallout3). Otherwise hope that the new improved LOD generator someone's written for OB can be extrended to Skyrim's system (I've had some contact with the author and he is curious).
4. People will have to playtest the 2x and 3x landscapes. Naturally there'll be disagreement of which to go for.

After that some sort of project could be formed based on it, and agreements on how far to go; whether people want to re-link all the existing Skyrim interiors to exteriors. Preserve existing quests (in which case exterior AI and scripts need recalculating for new positions). path grids should be redone. And how far to go - reproduce Bethesda's Skyrim game (i.e. quests) on a larger landscape, or start with a fresher world?