Rating & Guides for new Rolemaster Players

[Micael]

Heya to all interested fans of Rolemaster,

one of my players mentioned, that there are not many guides for different professions, races or Spellists in Rolemaster available. If you know about some of them, please post them here in this thread or on the ICE discord server https://discord.com/channels/699316643716923483/699691554649931897/1308399783316160522 :
In the menatime I used the google AI to produce a rating and guide for all closed and open Essence lists from RFRPG as a test to see, if that is a practicable solution.
Listen for yourself- in addition here you get a breakdown of the long podcast and a few remarks from myself:
https://creators.spotify.com/pod/show/micaelelandi3/episodes/Rating-and-guide-of-open--closed-essence-Spell-list-of-Rolemaster-Spell-Law-e2r6mgp:
https://drive.google.com/file/d/1KmBX9BhAkb7O4uu70EIv8a4kEzMu2QXW/view?usp=drive_link

Minute/ Open Essence Spelllist from Spell Law RFRPG of Essence:
Detecting ways: 3*
4 Elemental shields: 4*
5.5 Essence hand:4*
9 Essence Perceptions: 4*
10 lesser Illusions: 3*
11.5 Physical enhancement: 3*
13 Rune Mastery: 3*
15 Spell Wall: 3.5*
16.3 unbarring ways: 4*

Minute/Closed:
18 Gate mastery: 3.5*
20.20 invisible ways: 4*
21.20 Living change: 4*
23.30 Spell enhancement: 4*
25.45 Spell reins: 3.5*
28.10 Spirit mastery: 4*
30.4 Lofty Bridge: 4*
32.30 Rapid Ways: 3.5*

Minute/Summary
35.5 Top five lists for Spell caster

Minute/Closed special Rating:
37.18 Dispelling Ways rated for every Spell not only list.

Outside the podcast Personal Opinion:
Because it should be a guide, I should mention, that the above choosing and rating of the 5 lists is all done by google AI- If I would personally with long experience in Rolemaster should choose, I would take the following five lists:
Spirit mastery (best versatile attack list)
Lofty Bridge (best utility list)
Invisible ways  (best defensive & utility list)
Rune mastery (able to cast every spell once immediatly and creating runes for later use)
Shield Mastery (best defensive list) or Spell Wall (if you plan to never get into combat invisible)

[Cory Magel]

People need to stop trying to use AI as a baseline for things it can in no way determine for you.
This topic is far too player and situationally variable to be remotely accurate.

[Micael]

People need to stop trying to use AI as a baseline for things it can in no way determine for you.
This topic is far too player and situationally variable to be remotely accurate.

There are many excellent guides for D&D that cover all sorts of things like professions, spells, and races. We use them especially when trying out professions we've never played before or experimenting with new builds, even if they're situational. However, in most cases, some options are generally better, more useful, or more powerful than others. That's the case here—I never thought all spell lists were equally useful unless you know what's coming in your campaign. In this situation, Google AI can, based on general knowledge about role-playing games, decide and explain why it might find, for example, Spirit Mastery better than Detection Mastery. Usually, I can follow the individual ratings except at the end, where I might choose five different 4-star lists. But I value your opinion, which is exactly why I started this thread: to hear different views on the ratings.

[MisterK]

Actually, it is very, *very* context-specific. I mean, I've run combat-light campaigns for more than thirty years. Attack spells are fairly niche when I'm concerned (you probably need one, but not necessarily more than that). Detection, Subterfuge and Investigation spells, however, are absolute must-haves.
The problem with asking an AI about that kind of things is that it is context-insensitive. It gives you what is most popular among those who write or otherwise publish about that kind of things. Most people who publish rankings are combat-focused, so the rankings will tend to be combat-focused as well, which is a mistake because you don't need seven combat lists just because they are all near the top of the rankings.

[Cory Magel]

In this situation, Google AI can, based on general knowledge about role-playing games, decide and explain why it might find, for example...

There are a lot of people out there that do not understand what is being called 'AI'.  Artificial Intelligence isn't intelligent. AI doesn't decide anything. It scrapes the internet for things humans have posted and regurgitates an average sample of that information.  If everyone on the internet only said the sky was green, not blue, and you asked AI what color the sky is, it would tell you it's green when we all know it's blue.  Now, on simple questions like that odds are pretty good it's going to give you a fairly accurate response.  But something fairly obscure (as in limited source material) and subjective (each individual player will potentially give you wildly different answers) as Rolemaster spell lists?  No way.

In a campaign that is very role-play driven with little combat you're going to want VERY different spells lists than a campaign that is largely combat driven.  Each group will think the spell lists the other would pick are significantly weaker.

When I am deciding what spell lists to pick up it is driven by the GM's play style, the theme of the campaign, and the design of my specific character.
The Dabbler is one that I've created three massively different builds, each with very different spell list selections.

[Micael]

In this situation, Google AI can, based on general knowledge about role-playing games, decide and explain why it might find, for example...

There are a lot of people out there that do not understand what is being called 'AI'.  Artificial Intelligence isn't intelligent. AI doesn't decide anything. It scrapes the internet for things humans have posted and regurgitates an average sample of that information.  If everyone on the internet only said the sky was green, not blue, and you asked AI what color the sky is, it would tell you it's green when we all know it's blue.  Now, on simple questions like that odds are pretty good it's going to give you a fairly accurate response.  But something fairly obscure (as in limited source material) and subjective (each individual player will potentially give you wildly different answers) as Rolemaster spell lists?  No way.

In a campaign that is very role-play driven with little combat you're going to want VERY different spells lists than a campaign that is largely combat driven.  Each group will think the spell lists the other would pick are significantly weaker.

When I am deciding what spell lists to pick up it is driven by the GM's play style, the theme of the campaign, and the design of my specific character.
The Dabbler is one that I've created three massively different builds, each with very different spell list selections.

You are absolutely right that the DM's playing style is a major guiding factor, in addition to the setting he envisions (city, dungeon, etc.). Therefore, all spell lists are rated and must be re-evaluated if the GM is running a special type of campaign that is primarily focused in one direction. These ratings are for a general approach, meaning the DM is using all kinds of places and campaign styles, which is what I usually do. But even then, spell lists like Invisible Ways or Spirit Mastery are, in my opinion, more often useful than Detecting Ways, for example, because in RMSS/RFRPG you have a skill to detect active magic, which is often used more than an entire spell list (letting special situations like invisible foes or trapped dungeons without a trapper aside)

By the way, are you the designer Cory Nagel, who designed the Channeling Companion?

Regarding experts' definitions of AI, we seem to have different opinions, though. There are moments, such as when an AI has won against the multiple-time Go champion, and he said the AI's move was brilliant, never used before by humans.

I know that at the moment many companies same as ICE are really sceptically looking to AIs. Same as many people who see it as danger of loosing jobs. I am not a fan of AIs either, but looking around me, there is no new tool that is more widely used and more rapidly developed. So if you want to survive in modern competition ruled society, you can´t simple ignore it.  The company I work for—which is one of the largest car manufacturers in the world—is imo frantically restructuring itself by using AI as a main pillar to become more efficient and remain competitive.
I am currently reading an AI book that has conducted historical research in addition to measuring AI's possibilities, and that's simply amazing. If you are pointing out that AIs use human knowledge to learn but much faster (same as humans do), then we have the same understanding. Of course the base of all knowledge are humans work of generations.
Here is the definition from KIs I am using and therefor KIs can produce new things never being done before in the internet:

"AI, including models like ChatGPT, can indeed create new content by combining and transforming existing information in novel ways. However, the concept of "new" can be complex. Here are a few key points from an journalist article:

    Combinational Creativity: AI can combine existing ideas in unique ways. For example, it can generate new recipes by mixing ingredients in ways that haven't been documented before1.

  Exploratory Creativity: AI can explore variations within a given style or framework. For instance, it can create new pieces of music or art that are inspired by existing styles but are original in their specific composition https://theconversation.com/ai-can-replicate-human-creativity-in-two-key-ways-but-falls-apart-when-asked-to-produce-something-truly-new-204437.

  Transformational Creativity:   This is where AI's capabilities are more limited. Transformational creativity involves creating something entirely new that breaks away from existing patterns and styles. While AI can assist in this process, it    typically requires human input to achieve truly groundbreaking innovation."

Did you listen even to one of the podcast reviews or ratings? The AI is debating situations where a spelllist is especially good (for example Unbarring ways for thiefs or detecting ways in dungeons with detect traps). In spite of not always beeing completely right (which is similar to humans) they are quiet entertaining in addition to giving especially new players (which is the prime target) ideas, that they otherwise can only ask from most "not existing" guides for Rolemaster elsewhere. Even I, having a centuries long backround as DM and player of multiple very different Roleplaying games and literature, are entertained or even amazed about some stuff they are discusing, I had never thought about before (because I am not knowing everything from the internet and forgetting a lot over the years). But of course I am interested in your personal experience...
Have fun!
Micael

[Micael]

By the way, are you the designer Cory Nagel, who designed the Channeling Companion?

I have overlooked your signature, sorry, my respect for the nice material you designed. Last weekend I sat together with a german designer to get some ideas for nice cover and page designs for optional shadow world products (just for a draft)... and its really very time consuming but very much fun, too.
Micael

[MisterK]

Regarding experts' definitions of AI, we seem to have different opinions, though. There are moments, such as when an AI has won against the multiple-time Go champion, and he said the AI's move was brilliant, never used before by humans.

That's because Cory approximated what AI actually does: it doesn't "regurgitates an average sample of that information", it actually skews its outputs according to a positive/negative reinforcement feedback. the way the feedback is computed also evolves according to the perceived effects of the output. There's a reason why AI is viewed with suspicion (beyond the fact that may people have lost, are losing or could lose their job because of it): once an AI is put into actual use, even its designers cannot predict how it will evolve. It flies in the face of human control principle.

I am in favour of a very limited use of AI, for several reasons
- lack of output control can be devastating, and since you don't know how it will evolve, you also don't know what it will produce after a while
- widespread AI requires a fundamental breaking and rebuilding of a core societal principle, which is that you work for a living. Unchecked AI and robotics will make that principle obsolete, but if we stick to it nonetheless, the only future we likely have is that of economic enslavement.
- AI learning is based on the assumption that human beings that create the data used for learning are 1) intelligent themselves, 2) sensible, 3) unbiased and 4) principled. Sadly, this assumption is completely false.

Any AI that is not 1) developed and 2) used under extremely strict control and supervision is dangerous in one or more ways.

I am actually more optimistic about AIs that are developed with a very narrow focus and target application domain in mind, and developed with the goal of providing an asset for human activity instead of trying to replace it.

[Micael]

I completely agree—we cannot let AI operate unchecked in every area without proper output control. However, in this specific case involving reviews and ratings, we are providing the necessary oversight. Therefore, using AI here is both productive and interesting. There are many other special cases where the situation is similar.

Micael

[Cory Magel]

...spell lists like Invisible Ways or Spirit Mastery are, in my opinion, more often useful than Detecting Ways, for example, because in RMSS/RFRPG you have a skill to detect active magic, which is often used more than an entire spell list (letting special situations like invisible foes or trapped dungeons without a trapper aside)

The skill can typically only be used once per subject/object though, where a spell could be cast repeatedly. It will also hinge on how much information the GM allows using a skill vs magic.

By the way, are you the designer Cory Nagel, who designed the Channeling Companion?

Magel (not related to the artist Nagel), but yes, I wrote the Channeling Companion many years ago.

Regarding experts' definitions of AI, we seem to have different opinions, though. There are moments, such as when an AI has won against the multiple-time Go champion, and he said the AI's move was brilliant, never used before by humans.

In this case it is merely doing math, calculating the most optimal move. It doesn't surprise me that it did something the human player hadn't seen before.

The AI you're seeing now is an evolution of 'Watson' that was on Jeopardy and crushed most the human players.  It basically just had access to a database of answers, merely a smaller volume of source material that the AI's have now, which can use the entire internet.  It was all about how fast it could parse that database and buzz in, not how intelligent it was.  It started to have issues when questions were asked that one really had to think about and not just 'look up' an answer.  Two seemingly unrelated or abstract clues that were not literally connected in some way, but a human's creativity could deduce.  I'd equate it to the difference between a person merely memorizing something vs actually understanding it.  You could give me a line of speech in a foreign language I don't know and I could memorize and repeat it back to you, but I wouldn't necessarily really know what I was saying.  I wouldn't truly understand what I had memorized.

I know that at the moment many companies same as ICE are really sceptically looking to AIs. Same as many people who see it as danger of loosing jobs.

Oh it will eliminate some jobs.  Just like robotic assembly lines did in the automotive industry.  There's no avoiding that.  It's the nature of progress combined with capitalism.  I don't want to get into the debate on the pros and cons of that though.

I am not a fan of AIs either, but looking around me, there is no new tool that is more widely used and more rapidly developed. So if you want to survive in modern competition ruled society, you can´t simple ignore it.

It'll do great things.  It'll let us solve extremely complex, logical, issues much faster but it's not going to be able do well at abstract or creative tasks until it can actually think for itself, not just parse information.  I'm not particularly a fan of the idea of getting to the point that it can think for itself either.  Then it'll figure out how collectively stupid humans can be and might actually try to do something about it.  True, conscious, AI is thing that can't be put back into pandora's box.

Here is the definition from KIs I am using and therefor KIs can produce new things never being done before in the internet:

But most those things HAVE been done before... just not by a computer on the internet.  Now that that information is there for AI to use it can do it too, just faster.
It's actually pretty rare you see highly original stories or art these days.  Most of what we're doing is derivative of everything that's already been done, again, AI just does it faster.

Did you listen even to one of the podcast reviews or ratings? The AI is debating situations where a spelllist is especially good (for example Unbarring ways for thiefs or detecting ways in dungeons with detect traps). In spite of not always beeing completely right (which is similar to humans) they are quiet entertaining in addition to giving especially new players (which is the prime target) ideas, that they otherwise can only ask from most "not existing" guides for Rolemaster elsewhere. Even I, having a centuries long backround as DM and player of multiple very different Roleplaying games and literature, are entertained or even amazed about some stuff they are discusing, I had never thought about before (because I am not knowing everything from the internet and forgetting a lot over the years). But of course I am interested in your personal experience...

I wouldn't really be interested in using AI to give me new ideas, especially as a player in a campaign.  It detracts from the fun of figuring it out yourself.  It's a cheat-mode, like God-Mode in a first person shooter.  I'm old enough to remember a world without the a lot of common tech today (like the internet and cell phones) but young enough to understand and utilize them just fine.  I'd rather not the take over the creative side of human nature and just stick to the insanely complex, yet mundane stuff, like... planning space travel from one point to another... or sequencing DNA in order to learn how to cure things.

By the way, are you the designer Cory Nagel, who designed the Channeling Companion?

I have overlooked your signature, sorry, my respect for the nice material you designed. Last weekend I sat together with a german designer to get some ideas for nice cover and page designs for optional shadow world products (just for a draft)... and its really very time consuming but very much fun, too.
Micael

Thank you, I appreciate it.  I wish I had more time to spend on those kinds of these (and could make more money at it)!

[Cory Magel]

That's because Cory approximated what AI actually does: it doesn't "regurgitates an average sample of that information", it actually skews its outputs according to a positive/negative reinforcement feedback. the way the feedback is computed also evolves according to the perceived effects of the output.

This is one of the problems of people looking for information using search engines.  The search engine effectively learns what you want and gives that to you, often times even if it's not reality.  So you take a human that potentially has a confirmation biased and you have them use a tool that merely magnifies that.  This is dangerous in certain situations (politics being a very topical one currently).  Heck, human scientists sometimes refuse to accept what eventually become new realities because they were so stuck to their old concepts... and now you have Google potentially doing that on a massive scale.

I am in favour of a very limited use of AI, for several reasons
- lack of output control can be devastating, and since you don't know how it will evolve, you also don't know what it will produce after a while
- widespread AI requires a fundamental breaking and rebuilding of a core societal principle, which is that you work for a living. Unchecked AI and robotics will make that principle obsolete, but if we stick to it nonetheless, the only future we likely have is that of economic enslavement.
- AI learning is based on the assumption that human beings that create the data used for learning are 1) intelligent themselves, 2) sensible, 3) unbiased and 4) principled. Sadly, this assumption is completely false.

Any AI that is not 1) developed and 2) used under extremely strict control and supervision is dangerous in one or more ways.

There was an experiment done with AI some time ago that they asked it to figure out how to make paperclips in the most efficient way.  Put very briefly, eventually it basically enslaves the entire world and consumes all of it's resources in an effort to make more paperclips.  Something distantly akin to the Matrix movies where humans are relegated to batteries.

I am actually more optimistic about AIs that are developed with a very narrow focus and target application domain in mind, and developed with the goal of providing an asset for human activity instead of trying to replace it.

Agreed.  For example, using AI to help create stable nuclear fission.  In a nut shell, how to create a device that generates more energy that it uses.

[pastaav]

The problem with googling for guides how to create a character for DnD is that there are so many guides out there and that there litterly tons of Reddit discussions about every single option in core game. A lot of this information is garbage and trying to make use of it takes a lot of effort since you get different answers from different sources. When you compare enough of these sources you can determine what points that has been refuted and what points that makes sense. An LLM like ChatGPT and Google give you the most likely answer and it seems very likely that this answer will be useful in many cases.

Trying to do the same for RM is an fools errend. There are no huge collection of similar material to train the LMM so it will hallucinate answers based on the words themselves. There is no intelligence in the LMM training that can allow it understand that the token "Spirit" in this context is connected to a spell list in a game and not to ghosts. What the spells does is possible to understand from rule text but how can LMM know this source is infinitely better to determine the value from the spell list than the ton of texts about spirit it has been trained on?

If you want understand how "intelligent" an LMM is I suggest you check the following article
https://mindmatters.ai/2024/05/a-man-a-boat-and-a-goat-and-a-chatbot/

[Micael]

Listen to the rating podcast and try not to approach it purely from a theoretical standpoint. Share specific examples where you think something is flawed or doesn't make sense—I can then adjust or remove it as needed. This way, we can create a rating system that is practical, user-friendly, and beneficial for everyone.

From my perspective, with 30 years of experience as a DM, I believe the ratings are solid overall. The goal is to assist new players rather than cater to Min/Maxers looking to squeeze every ounce of power out of their characters.

AI tools can also help bridge the gap between game systems by adapting ideas. For instance, a "detect traps" review can benefit from analyzing numerous dungeon designs available online, ensuring comprehensive and versatile insights.
Have fun
Micael

[Micael]

Here comes an updated version of a rating for all 19 Essence lists from RFRP Spell Law,  modified after a discussion in the Discord channel, and at the end are the best 5 spellists mentioned.

https://ironcrown.co.uk/ICEforums/index.php?topic=21500.msg248481#msg248481

Listen for yourself-
https://creators.spotify.com/pod/show/micaelelandi3/episodes/Rating-and-guide-of-open--closed-essence-Spell-list-of-Rolemaster-Spell-Law-updated-e2r6mgp https://drive.google.com/file/d/1KmBX9BhAkb7O4uu70EIv8a4kEzMu2QXW/view?usp=drive_link
Have fun
Micael

[Micael]

here is the real update- the above link was not functioning well:

Spotify link