GURPS Power-Ups 10: Skill Trees

[Kallatari]

In my previous post, I had mentioned that I had knee-jerk reaction as to the added cost this would have to templates and how you had to adjust the starting character points. I had even taken a wild guess and said it's probably at least a 50 point increase while admitting I had no evidence to back it up. I've now run it through and redesigned approximately 30 templates and have some actual numbers to share with people as an FYI.

First caveat about this work. These are my templates for my games, not templates of various GURPS products. So obviously, others may get different results based on how they built their own templates.

My templates start off as "ordinary professionals", and then have lenses to improve them to heroic and cinematic levels. I only used the professional levels of skills since it's the upfront cost of reaching normal levels of skills that increases in Skill Trees. The ordinary professional bases mostly all had a standard design in skills.

Primary Skills: One skill at +3 [12] and two skills at +2 [8 each]
Secondary Skills: Four skills at +0 [2 each]
Background: Nine skills at -1 [1 each].

That comes to a total of 45 points. Obviously, because skills vary between Easy, Average, Hard, and Very Hard, the actual levels might vary a bit, but the breakdown of points remained the same in my templates, so we can ignore those.

So, I converted them and redid the skills using the Skill Trees, with Trunks at [7 x level], Branches at [3 x level], etc. If you use different values, the results will be different.

Also, although there have been posts suggesting getting a slight increasing then plateauing costs for Trunks and Branches (e.g., something like [1] for the first level, [3] for the second level, then [7] for the third level, and [+7] for each additional level), I did not try that in any of my tests. I wanted to use the rules as presented in Skill Trees. Although I did vary the "starting level" for my tests just to see the effects of that.

Here are my results so far:


Starting Level of Attribute -5

For one very focused template where all the skills miraculously fit under 1 Trunk, the new cost was 49 points, so only a 4 point increase. I consider that one an outlier and didn't include it in my other results, but sharing that fact with you anyway as an FYI.

Most templates require 2 or 3 Trunks, and those ended up costing 112 points (or a 67 point increase from 45). That's probably a better number to use. However, since everything had been converted into Trunks only, it lacked some of the nuances using Skills did.

On the other hand, when I assumed nothing could fit under a single Trunk and had to purchase each individual skill as a Branch, that ended up costing a total of 234 points (or a 189 point increase from 45). This scenario never actually occurred, as I could use Trunks in my templates, but it was a nice exercise to determine a potential upper limit as to what the costs could be for a template with a wide mix of unrelated skills.

When I tried to min-max Trunks and Branches to more closely match the nuances, the values really fluctuated by the template and its original skills. Right now, my average is 138 points (or a increase of 93 points over 45) , but the width of distribution is wide, so additional entries can cause that to move. Note, many were impossible to get exactly the same, as purchasing a certain level of a Trunk when you had 2 skills at that level automatically increases all the other skills that were lower in that Trunk to match. Still, I made them in a way that I was satisfied that it matched the right feel of the template.

So, in summary, it resulted in a cost increase that was typically between an additional 67 to 189 points. Since I started at 45, that's x2.5 the original points in skills to x5 the original points in skills, with an averaging around x3 (and large distribution of results).

Hopefully those will help for someone who wants to start with Skill Trees exactly as presented, noting that you will need a lot more points in skills for starting characters.


Option 2: Starting Level of Attribute -2

It had been suggested in some posts that instead of starting at Attribute -5 you could start at Attribute -2. So I ran those numbers instead, using the same costs of [7 x level] and [3 x level] for Trunks and Branches.

In my one outlier anomaly of fitting everything under a single Trunk, that ended up costing only 28 points, which is a savings of -17 points, or x0.6 the required number of points. Again, this was excluded from my other calculations.

In most cases, using only Trunks (and thus removing the nuances between similar skills) ended up costing a total of 49 points (or an increase of 4 points from 45). On the other end, using only Branches ended up costing a total of 90 points (or in increase of 45 points from 45).

Trying to obtain as close a match as possible mixing and matching Trunks and Branches is currently averaging about 60 points (or 15 points above 45).

In summary, it resulted in a cost increase that was typically between an additional 4 to 45 points. Since I started at 45, that's approximately x1 the original points in skills to x2 the original points in skills, with an averaging around x1.3 (and large distribution margin).

This is actually fairly decent with respect to keeping the point costs between Skill Trees and the current system the same. And you can apply Skill Trees as written other than the starting point (Attribute -2 instead of Attribute -5).



Option 3: Starting Level of Attribute -0

For the fun of it, I tried again, this time with a starting level of Attribute.

When using only Trunks, I ended up with a cost of 42 points (so a savings of -3 points compared to 45). When using only Branches, it came to about 66 points (so an increase of 21 points from 45). So you end up with a point variance between x1 to x1.5 Mixing both Trunks and Branches together averaged out to about 48 points, so that still rounds to x1 (with a wider distribution of results, of course).

This had a drawback, however. My original templates included Secondary skills at +0, and Background skills at -1. You can't achieve that if you start at Attribute, as adding just one level of a Trunk or Branch brings you to +1.

In summary, using a Starting Level of Attribute keeps the point total pretty much the same with you getting higher skill levels in the secondary and background skills, so you're actually more powerful. In the end, I personally didn't like this option because all the skills ended up with a higher level than the attributes.




Note that Skill Trees does start to get less expensive overall once you reach skills of +5 or higher everywhere. I intentionally did not do my tests to that point. In my games, I found it rare for characters to have that high a skill level in more than just a couple skills, so I didn't think it would end up being that much of a savings compared to what they needed to spend on all their secondary and background skills at much lower level. Still, it is something one should consider in addition to what I presented above. If your templates are very cinematic with very high skills levels, it might actually be cheaper for you even with a starting level of Attribute -5.

And I just want to conclude by once more reminding people that this was the result of converting my own personal templates that had standardize skill costs/distribution. You will probably get different results for your own templates. This is therefore just one data set, not an absolute conclusion. But it hopefully gives you a good idea as to approximately where such a conversion may lead for your game.

[Celjabba]

Originally Posted by Kallatari

In my previous post, I had mentioned that I had knee-jerk reaction as to the added cost this would have to templates and how you had to adjust the starting character points. I had even taken a wild guess and said it's probably at least a 50 point increase while admitting I had no evidence to back it up. I've now run it through and redesigned approximately 30 templates and have some actual numbers to share with people as an FYI.

First caveat about this work. These are my templates for my games, not templates of various GURPS products. So obviously, others may get different results based on how they built their own templates.

My templates start off as "ordinary professionals", and then have lenses to improve them to heroic and cinematic levels. I only used the professional levels of skills since it's the upfront cost of reaching normal levels of skills that increases in Skill Trees. The ordinary professional bases mostly all had a standard design in skills.

Primary Skills: One skill at +3 [12] and two skills at +2 [8 each]
Secondary Skills: Four skills at +0 [2 each]
Background: Nine skills at -1 [1 each].

That comes to a total of 45 points. Obviously, because skills vary between Easy, Average, Hard, and Very Hard, the actual levels might vary a bit, but the breakdown of points remained the same in my templates, so we can ignore those.

So, I converted them and redid the skills using the Skill Trees, with Trunks at [7 x level], Branches at [3 x level], etc. If you use different values, the results will be different.

...

Here are my results so far:


Starting Level of Attribute -5

For one very focused template where all the skills miraculously fit under 1 Trunk, the new cost was 49 points, so only a 4 point increase. I consider that one an outlier and didn't include it in my other results, but sharing that fact with you anyway as an FYI.

Most templates require 2 or 3 Trunks, and those ended up costing 112 points (or a 67 point increase from 45). That's probably a better number to use. However, since everything had been converted into Trunks only, it lacked some of the nuances using Skills did.

On the other hand, when I assumed nothing could fit under a single Trunk and had to purchase each individual skill as a Branch, that ended up costing a total of 234 points (or a 189 point increase from 45). This scenario never actually occurred, as I could use Trunks in my templates, but it was a nice exercise to determine a potential upper limit as to what the costs could be for a template with a wide mix of unrelated skills.

When I tried to min-max Trunks and Branches to more closely match the nuances, the values really fluctuated by the template and its original skills. Right now, my average is 138 points (or a increase of 93 points over 45) , but the width of distribution is wide, so additional entries can cause that to move. Note, many were impossible to get exactly the same, as purchasing a certain level of a Trunk when you had 2 skills at that level automatically increases all the other skills that were lower in that Trunk to match. Still, I made them in a way that I was satisfied that it matched the right feel of the template.

So, in summary, it resulted in a cost increase that was typically between an additional 67 to 189 points. Since I started at 45, that's x2.5 the original points in skills to x5 the original points in skills, with an averaging around x3 (and large distribution of results).

Hopefully those will help for someone who wants to start with Skill Trees exactly as presented, noting that you will need a lot more points in skills for starting characters.


Option 2: Starting Level of Attribute -2

It had been suggested in some posts that instead of starting at Attribute -5 you could start at Attribute -2. So I ran those numbers instead, using the same costs of [7 x level] and [3 x level] for Trunks and Branches.

In my one outlier anomaly of fitting everything under a single Trunk, that ended up costing only 28 points, which is a savings of -17 points, or x0.6 the required number of points. Again, this was excluded from my other calculations.

In most cases, using only Trunks (and thus removing the nuances between similar skills) ended up costing a total of 49 points (or an increase of 4 points from 45). On the other end, using only Branches ended up costing a total of 90 points (or in increase of 45 points from 45).

Trying to obtain as close a match as possible mixing and matching Trunks and Branches is currently averaging about 60 points (or 15 points above 45).

In summary, it resulted in a cost increase that was typically between an additional 4 to 45 points. Since I started at 45, that's approximately x1 the original points in skills to x2 the original points in skills, with an averaging around x1.3 (and large distribution margin).

This is actually fairly decent with respect to keeping the point costs between Skill Trees and the current system the same. And you can apply Skill Trees as written other than the starting point (Attribute -2 instead of Attribute -5).



Option 3: Starting Level of Attribute -0

For the fun of it, I tried again, this time with a starting level of Attribute.

When using only Trunks, I ended up with a cost of 42 points (so a savings of -3 points compared to 45). When using only Branches, it came to about 66 points (so an increase of 21 points from 45). So you end up with a point variance between x1 to x1.5 Mixing both Trunks and Branches together averaged out to about 48 points, so that still rounds to x1 (with a wider distribution of results, of course).

This had a drawback, however. My original templates included Secondary skills at +0, and Background skills at -1. You can't achieve that if you start at Attribute, as adding just one level of a Trunk or Branch brings you to +1.

In summary, using a Starting Level of Attribute keeps the point total pretty much the same with you getting higher skill levels in the secondary and background skills, so you're actually more powerful. In the end, I personally didn't like this option because all the skills ended up with a higher level than the attributes.




Note that Skill Trees does start to get less expensive overall once you reach skills of +5 or higher everywhere. I intentionally did not do my tests to that point. In my games, I found it rare for characters to have that high a skill level in more than just a couple skills, so I didn't think it would end up being that much of a savings compared to what they needed to spend on all their secondary and background skills at much lower level. Still, it is something one should consider in addition to what I presented above. If your templates are very cinematic with very high skills levels, it might actually be cheaper for you even with a starting level of Attribute -5.

And I just want to conclude by once more reminding people that this was the result of converting my own personal templates that had standardize skill costs/distribution. You will probably get different results for your own templates. This is therefore just one data set, not an absolute conclusion. But it hopefully gives you a good idea as to approximately where such a conversion may lead for your game.

I started converting a DFRPG character at 3 levels (62 pts, 125 pts and 250 pts) and I get similar results.

I will try to post fully done characters later on, but at first glance if you convert so that you have at least the same skills at the same relative levels (*), there is an huge increase in cp for characters with many diverse skills in the -2 to +2 attributes range.
You can mitigate a little by playing with cross-trunk skills, twigs, and reducing IQ! and removing talents may gain you some points, but you are still looking at a +100% to +200% increase in skills cp budget on average. (to be confirmed once I finished my test characters, but it match @Kallatari findings.
Especially for super-optimized characters like DFRPG ones...

Of course, you do get some serious benefits : lots of other skills if you used trunks and the possibility of using 10+trunk outside of combat.

Once you look at characters with many skills at the +5 to +10 relative level, both method prices start to converge.

I think that for low-level campaign, the variant "default is attribute -5, once you have at least 1 level in a skill, you jump to attribute -2 instead of -4" would be a good alternative. You "save" a base cost of 21 points per trunk, 9 for branch (assuming no cross-pollination). Still more expensive than the classic way, but I feel it would get balanced results if you care about staying in teh same budget ballpark than the traditional way.

Your suggestion of moving the default to -2 would get even closer prices, but given the side benefits, it may be too generous, especially with trunks... YMMV.

Either variants will make very skilled characters in many skills cheaper, which is something to watch for if the characters have the budget for it.

* : Obviously, the 2 systems are based on a different approach, with the tree costs being linear but starting with lower skills, so characters build from scratch will likely end up different depending on the system, as each will point toward a different build approach.
If you are not converting templates made for the traditional system, it is probably as easy to increase the cp budget until you get the characters where you want them (possibly using points buckets for that) and ignore the fine tuning above :)

[WingedKagouti]

All this talk about costs make it seem like the system was primarly designed for characters with most of their points in closely related skills.

[Kallatari]

Originally Posted by WingedKagouti

All this talk about costs make it seem like the system was primarly designed for characters with most of their points in closely related skills.

To be clear, this talk about points is for the purpose of "converting" (for lack of a better word) between the two systems in order to get similar final skill levels. If your skills are all closely related, you "win" at the conversion game and it's cheaper. If your skills are completely unrelated, you "lose" at the conversion game, and it's more expensive.

GURPS Powers-Ups 10 - Skill Trees is merely an alternate way of using skills, and doesn't really care about "conversion" with respect to points. It assumes you start the campaign with it, not that you're converting characters from the old system to this one. If you start that way, then it's fair for everyone; no one wins or losses, other than by their personal choice of what skills they purchase, much like in the current system (e.g., spending points on skills where you have a good attribute or talent is more effective than those who don't). Optimizing for best use of points exists in either system; the optimization is just done differently, and that makes for widely different point costs if you "convert".

Also, GURPS Powers-Ups 10 - Skill Trees states in its explanation as to its underlying assumptions, that skills are the most important part of a character, more so than attributes, with regards to defining who they are. From there, in my opinion (not something written down in the text), does that therefore not mean that if skills are the most important part then you should be spending most of your character points on skills? In that, the Skill Trees actually backs that up by working in that manner - you will have to invest a lot of points in skill to make a good rounded character. It's not in any way stated that way in the document, but personally I find that to be a happy coincidence.

I'm only playing around with the point costs because I like what's in Skill Trees, and I intend to use it in future campaigns. I've been playing with the current skill structure for over 20 years, so I understand it perfectly and can easily give advice to my players as to how many points they should spend on skills for a given concept. Skill Trees is brand new, and changes the point assumptions, and I just want to get a good wrap around as to what exactly it implies so that I can give the same sort of advice to my players. I'm trying to get that sense by "converting" what is there, and it's starting to show me what I'm going to need. So, to be clear, you shouldn't be converting anything, but testing it out by converting gives you a better idea how to implement it.

Since I've done a lot of work to get to that conclusion, I'm just sharing that work it so that others can benefit from it rather than doing it all themselves.

[Kallatari]

Originally Posted by Celjabba

Your suggestion of moving the default to -2 would get even closer prices, but given the side benefits, it may be too generous, especially with trunks... YMMV.

I didn't actually make any reference to default in that post, as I was simply looking at point costs if you "converted" templates.

To further clarify what I meant by "starting level of Attribute -2", I merely meant that you add you level of trunks, branches, twigs, and leaves to Attribute -2 to calculate your skill level. It wasn't intended to imply a default of Attribute -2. I probably should have been more clear in that post.

I would still leave Default at Attribute -5. So in effect, I'd also have a "jump" once you purchase at least 1 level of skill (at any tier).

And that's only if I want to keep points "roughly" equivalent between systems for "it makes me feel happier" reasons. I might just go all the way and give a lot of extra points to players and use the full -5 starting point because there is technically nothing wrong with that per se. You just need to adjust how you build your character. (Although -5 is still in the options for me, I do admit to leaning to skills start at -2 at the moment.)

Originally Posted by Celjabba

If you are not converting templates made for the traditional system, it is probably as easy to increase the cp budget until you get the characters where you want them (possibly using points buckets for that) and ignore the fine tuning above :)

I'm also looking a point buckets more closely now, and possibly further breakdowns (e.g., points buckets for primary skills and point bucket for secondary skills). It's one thing to say "I'm going to give them an additional +50 points (to pick a number at random) so that they can use them to get reasonable skill levels", but what happens if they spend it on attributes or powers instead? What if you're doing +100 or +150 points if you want to start at default -5? I do think Skill Trees really needs to tell players that they want to invest points in secondary skills to be real characters, and no just focus on their core concepts because it's significantly cheaper now to ignore secondary skills.

[Celjabba]

Originally Posted by Kallatari

I didn't actually make any reference to default in that post, as I was simply looking at point costs if you "converted" templates.

To further clarify what I meant by "starting level of Attribute -2", I merely meant that you add you level of trunks, branches, twigs, and leaves to Attribute -2 to calculate your skill level. It wasn't intended to imply a default of Attribute -2. I probably should have been more clear in that post.

I would still leave Default at Attribute -5. So in effect, I'd also have a "jump" once you purchase at least 1 level of skill (at any tier).

And that's only if I want to keep points "roughly" equivalent between systems for "it makes me feel happier" reasons. I might just go all the way and give a lot of extra points to players and use the full -5 starting point because there is technically nothing wrong with that per se. You just need to adjust how you build your character. (Although -5 is still in the options for me, I do admit to leaning to skills start at -2 at the moment.)

I misread your post and made a wrong generalization about default - sorry.

[Kromm]

The system is pretty clear about its goals:

As good character concepts focus on narrow sets of related things, you can simplify your character sheet by raising only the highest-tier items that define your concept, plus a few unrelated bits that add color and personality.

and:

. . . you'll get sharper character concepts for subject-matter experts.

If the proposed Trunks don't fit your sense of "related," see Template Trunks (p. 16). You can define "everything on a template" as a Trunk if that's important to you for your genre or setting treatment.

Also take note that "you do not have to spend character points on a Trunk to have or improve its Branches, a Branch to have or improve its Twigs, or a Twig to have or improve its Leaves." That is, you can just up and buy Twigs or even Leaves; see Missing Pieces (p. 4). This is useful if all you want to do is be the best in the world at one specific thing; take 15 levels of a Leaf for 15 points and you can be operating at skill 20 . . .

Finally, there's a strong implication that people are buying Trunks instead of attributes:

Skill Trees favors PCs who are well suited to a role over generalists with high attributes and skills chosen willy-nilly.

and:

. . .skills are more expensive in Skill Trees, in part to encourage players to buy Trunks rather than attributes.

Rather than buy IQ 14 [80] and toss 1 point into each of ~10 sciences at skill 11 (H) or 12 (VH), for ~90 points, a generic "scientist" would buy IQ 12 [40] and Natural Sciences 7 [49], for 89 points, and use all sciences at 14. And if the GM decided that a roll should be based on a score other than IQ, well, the second character would usually do better; assuming DX 10 in both cases, a DX-based roll for something tricky in the lab would be 7-8 for the first character, 12 for the second one.

But yes, there's definitely a degree of favoritism toward specialists and a big hit to the efficiency of jack-of-all-trades types. Players would need to pick different optimization goals under this system, while GMs would want to both hand-craft Trunks to suit their campaign (I think templates would be a good starting point) and float rolls to many attributes more often as a matter of course.

[Kromm]

This supplement is tightly written due to price-point constraints, so a lot of things won't be obvious at first. However, it exists to address many, mostly unrelated complaints that I've received about GURPS over the years:

• The overwhelming top-heaviness of attributes. This leads to minimaxing to optimize skills under the assumption that the GM will rarely or never base a roll on anything but a controlling attribute, which in turn leads to cinematic competence in non-cinematic campaigns ("I am a great scientist, so I am also awesome at social skills and art, and am strong willed and hyper-alert."). So, I made attributes less useful than skills for specific competence at a given point cost.
  
  
• The presence of Talents and wildcard skills and grouped skills and specialties and techniques and . . ., seemingly uncoordinated, to cover various degrees of general and specialized competence. So, I unified the lot, and in the process reduced the likely number of things written on a typical character sheet (possibly to something like "Thief 10" in some cases).
  
  
• The insanity of default webs. Those who love it, love it a lot! But everybody else is asking enough questions to merit their own FAQ. After encumbrance, defaults are among the rules most often killed by GMs. So, I made being good at related things innate and advised the GM to be flexible.
  
  
• The very existence of skill difficulties. They seem goofy to many people, and clutter character sheets. Why not just call some tasks harder than others? So, I got rid of difficulties and made modifiers, which need to exist in any event, more important.
  
  
• The 4-level jump from 0 points to 1 point on the Skill Cost Table. For each person who believes that "a little training goes a long way" is realistic and likes that jump, I have complaints from five people who find the "–" entries on the table confusing, unrealistic, or both. So, I got rid of the jump.

. . . plus some minor stuff, like acknowledging Unarmed Grappling and Unarmed Striking as a better way to treat unarmed skills relative to Melee Weapon skills, and getting rid of the especially goofy Average/Hard division for techniques, which is arbitrary anyway if you know how techniques are designed.

Veterans gamers might not agree with any of this! I understand that. But if you want to lower the barrier to comprehension for new players, you might – even as a veteran – consider using this system.

[Kromm]

As a final comment, don't overlook Trunks and Flexibility (pp. 18-19) and Attributes and Fluidity (pp. 19-20). A lot of the time, the GM will just say, "Okay, make a roll against 10 + any Trunk that you can convince me applies." Maybe there will be task modifiers, maybe there won't be. But one could run an entire campaign with just Trunks and extreme flexibility . . . it might feel like Risus, but that isn't a bad thing.

[NocTempre]

Perhaps I missed it on my first read, but I definitely was hoping for more thoughts on controlling skills (prerequisites) which I find interesting and realistic, as well as a generalization of the application of skill context (expert, basic, sport, hobby, etc). I also would have appreciated a worked examples of a style under this new system. As a system hacker with pages of notes this is a welcome addition, but completely the opposite direction of what I expected. Perhaps I could humbly request more designer notes be posted over time?

Edit: I will add emphasis that I really enjoyed it, and it does fantastic groundwork on harmonizing many overlapping and at times downright confusing design elements of 4E and earlier. I it would be excellent to follow this up with a How to Be a GURPS GM volume pointing to all these alternate systems for even the CORE RULES that have been collected over the years with an eye on tuning a game for a certain feel beyond bolt-ons. You could call it Beyond 4E if you want to be cheeky...