Re: Another idea for the old skill VS Attribute conundrum
 

Sounds like an article for Alternate GURPS IV... :)

Re: Another idea for the old skill VS Attribute conundrum
 

You can't fix this without drastically changing the skill system so that there is a difference between DX10+skill at DX+4 and DX14+skill at DX.

Re: Another idea for the old skill VS Attribute conundrum
 

To be fair, there already are some differences. For unarmed combat skills, I think all of them see better damage above skill at DX. With Weapon Master, higher relative skill also gives better damage (and it isn't much of a change to require skill at DX+1 or DX+2 to get the full damage on the weapon table anyway). Applying the Trained ST progression from Technical Grappling can also make high relative skill important. If enforcing Perk limits, higher relative skill means more points invested, thus allowing for more Perks related to that skill. More points also means more Familiarities - the higher your relative skill, the more likely you'll be able to use a new piece of equipment at full skill without penalties. Finally, there's always the possibility of needing to "float" a skill over to another attribute - when a DX 15, IQ 10 Swashbuckler with Lockpicking at DX comes across a puzzle lock (requiring a roll against IQ-based Lockpicking), he's going to be at a relative -5.

Re: Another idea for the old skill VS Attribute conundrum
 

Floating skills is indeed the Great Overlooked. Not to toot my own horn, but I don't think GURPS content is to blame for this . . . the published oeuvre is chock-a-block with examples. Rather, it's mostly due to a mixture of rules-lawyer players insisting on "getting their points' worth" by always using the skill based on its nominal controlling attribute, and GMs either giving in to that pressure or being uncomfortable about making judgment calls.

Perhaps interestingly, my initial vision of GURPS Fourth Edition had skills bought and recorded strictly as relative levels: -3, -2, -1, 0, +1, +2, etc. It left attribute selection to the GM, guided by examples. However, other people felt that not listing ready-to-roll-against levels was a bad idea, and that requiring the GM to pick an attribute for each use was an even worse idea. Thus, we settled on the current system.

Re: Another idea for the old skill VS Attribute conundrum
 

Well, if you combine Half-Stat Defaulting with the Rule of 20, you get a system where stats stop adding to skill at 20.

In HSD, skill progression is based off Default+X. The highest possible default is 10+Y, with Y being a positive or negative modifier based on how hard the skill is.

So once your stats hit 20, the only way up is to invest in skills, talents and techniques. I consider this a feature.

Re: Another idea for the old skill VS Attribute conundrum
 

Okay - I typed out a lot of logical assumptions and breakdowns, and got another very long post... then lost it all. So I'll sum up with this rewrite:

The problem is not simply that Skills cost too much, specialisation works fine and at low Attribute levels no cracks can be seen. The issue is that raising the cost of many skills costs too much when compared to the cost of simply raising the controlling Attribute.

Equally, the problem is not that Attributes cost too little, they affect many other things than Skills, and there are many other reasons to improve them into 'excessive' levels. The issue is that they're commonly available to everyone and their limits combined with minimal (and cheap) investment into Skills produces insanely illogical expertise in whole swaths of generalist fields.

A solution that is both quick, easy, and directly addresses this is almost already in Basic Set. Defaults cannot exceed a set point regardless of how high the Attribute its based on is, namely, they assume any Attribute higher than 20 is just 20. The solution is simply to apply the same notion to all Skills.
Done.
GM's may wish to push this number as low as 15 or as high as 30, depending on the setting - but simply capping the contribution of an Attribute forces you to invest in Skills once that cap has been met.

A slightly more complex alternative, which might be fairer and more realistic, is that after you reach said cap - every level either grants a free level of a single Talent of your choice, or allows you to more easily train into one (if you use training etc). This lets you specialise into a field of excellence, but prevents you from running amok with insane proficiency in almost half of the skills linked to your amazingly high Attribute. Your GM may have to make a number of new Talents to fit this bill, in which case I suggest grabbing a copy of Power-Ups 3: Talents, its awesome.

Honestly though, its easy enough to fix if a GM discourages it and offers suitable Talents for individuals who wish to have a wider range of 'specialisation'. The tools are already out there to fix this 'issue' - but sticking a cap on your Attributes might incline your players (and the GM) towards using them a little more.

Re: Another idea for the old skill VS Attribute conundrum
 

I take the listed attributes as suggestions. The only time it's really worth noting the "standard" sum is for Parry, I think.

Re: Another idea for the old skill VS Attribute conundrum
 

I have thought about this for many a long year, and the fix I came up with is relatively simple for the plain reason that I am incredibly lazy. It may have its own cracks and weaknesses (e.g. cost of characters increases quite a bit), but it's worked pretty well for me so far.

Trained skills are no longer directly linked to attributes. Untrained default levels for skills are the relevant statistic minus 4 for Easy skills, 6 for Average skills, 8 for Hard skills, and 10 for Very Hard skills, with a normal maximum level of 12 (Easy), 10 (Average), 8 (Hard), or 6 (Very Hard). Attribute modifiers to skills are included in these numbers and are not added separately, but bonuses given by advantages (such as Talent) can raise a person’s default skill level over this maximum. In this way, it is possible for an untrained character with high relevant stats to have a skill ability greater than that of someone just beginning their training, but this ability will include bad habits that will have to be unlearned when formal training begins, and only an extremely rare natural talent will even be able to approach the skill level of someone with a lot of proper training. Default levels in a skill that end up being below three are treated as having no default for that character.

I then have trained skill be based just on the points invested in skills, with arbitrary skills values set for one point (lower for harder skills, higher for easy skills) but have a range of skill modifiers based on the relevant attribute, ranging from -2 to +2. Easy skills get a higher modifier earlier than harder skills. If you float the skill, you use the appropriate new attribute modifier instead.

Re: Another idea for the old skill VS Attribute conundrum
 

If the problem is that a lot of training in many skills is much less point inefficient than extremely high attributes while such training is more common than extremely high attributes in the setting (Kromm gave such an example from the development of GURPS Special Ops), then there is already a solution in the Basic Set.

Giving an Unusual Background cost for high attributes would make the most point efficent builds more realistic in that sense.

The same can be done for when a lot of points are spent on a single skill. That could make it so that the most point efficent build for reaching a very high level in one skill is to have both a high attribute and a lot of training (just like what is the case in the real world for those who are extremely skilled).

Re: Another idea for the old skill VS Attribute conundrum
 

Slight note: "The GURPS Staff" is Sean, Jason, and Steven.

I think it's more accurate to say that "those who write for GURPS see certain common threads on the forums by a vocal few, and have written some articles to address their concerns."

(Edit: Should have figured I'd be totally ninja'd by Steven early on)

I'm actually fine with defaulting mostly as-is, even though I wrote half-stat defaults and have used it in play. It really only has a big impact on characters with very high stats and games where niche protection is very valuable.