Originally Posted by DouglasCole

The value of that extra level of stat/skill very much depends on what kind of challenges the PC will be facing.

I don't have any comment on attribute cost vs skill cost (not here and now, anyway), but on flat vs rising costs, DC's wise remark above inspires the following note:

I have an old, oddball article online that asks what would happen if skills and attributes (stats below) both used perpetually-rising costs.
http://www.gamesdiner.com/gurps/escargo.htm

Doing so offers all kinds of potential benefits. Yet as noted somewhere in the text, it's only a theoretical "what if" discussion, and I've never put it into play.

Why not actually use it? Because despite any nifty benefits in terms of mechanics and GM happiness, I can't actually justify increasing stat costs from the standpoint of utility to the PC. (And if there's a Cult of "Point Costs Should Measure Utility and Only Utility", where do I sign up?)

One can argue that the cost of a +1 in a stat should increase as the stat goes up, and one can argue that the cost of the +1 should decrease as the stat goes up:

1) Why stat costs should increase with level: Buying that jump from DX 10 to DX 11 does not make the character a big deal. Lots of people have DX 11. But DX 15 to 16 – now that's something! Each +1 up at that altitude winnows the PC out an already-rare breed, placing him closer to the top of the heap. Go a few levels higher up, and each +1 moves the PC another effective step into godhood.

Each +1 at high levels expands a range of tasks that only the PC (and his shrinking group of peers) can attempt: combat shots against impossible targets, success despite overwhelming stacked penalties, etc. Clearly, the cost of each +1 should increase as the stat does!

2) Why stat costs should decrease with level: Sure, each +1 lets the PC do more and more things that others can't. But as the stat goes up, each additional +1 does nothing for many of the tasks that the PC actually will perform.

While standing on a rolling ship deck and poking a pint-sized foe in the eye is an impressive use of Saber-25, most combat attacks have far fewer penalties attached. Once you can quickly drop the typical foe dead in one shot, do extra skill levels drop him any deader? How often will those oddly-high levels of Mechanic come in handy, once you can reliably tackle most repair jobs (or just keep trying until you succeed)? Once HT is high enough that you'll almost always make a roll, isn't it much smarter to divert extra CP into other stats that can keep you out of danger in the first place, instead of a few extra percentage points on your chances of HT roll success?

High, one-trick-pony stats make for a poor return on CP invested, compared to a more diverse range of abilities. Clearly, the cost of each +1 should decrease as the stat does!

-----

I don't see either of the above arguments as demolishing the other; both make good points. Like DC says, the value of a +1 in a high stat will really depend on what tasks the player and GM push the PC toward.

So not knowing whether increasing or decreasing stat costs make the stronger argument, I vote for the easy middle ground: flat costs.