Quote from: daniel_ream;563070The Traveling Salesman says this is really, really hard.

Can you expand some more on that, by any chance?

Quote from: The Traveller;563024Is there really any way to CharOp in 2e? You pretty much get what you get.

Dart-specialist fighters come to mind.

When WoW players started crossing over into 3e.

Quote from: Fifth Element;563142Dart-specialist fighters come to mind.

Certainly an abuse of a poorly designed rule which may have gone back to 1E Unearthed Arcana.

You really can't put a definitive starting point. There were always players who looked to have the 'optimal' PC choice. Certainly from the introduction of variable weapon damage you had people choosing the 'best weapon'. Even before then you had people playing what they perceived to be mechanically the strongest class.

Likewise with the introduction of a rudimentary skills system you had people cherry picking the best ones. All you can really say is that there has progressively been more and more scope for charop since the start of the hobby.

I guess key stages in its expansion came with the following milestones:

• Supplements I - IV
• 1E
• Unearthed Arcana
• 2E with its skill subsystem amongst other things
• Complete Series
• 2.5E (skills and powers)
• 3E
• Every splat book under the sun

Quote from: Bloody Stupid Johnson;563141Can you expand some more on that, by any chance?

The Traveling Salesman problem is a famous problem in optimization.

Don't read the Wikipedia page; I've studied graph theory and their description is so jargon heavy it makes my brain hurt.

In short, the time it takes to find the "best" set of choices in a point-buy system (each choice presumably contributing some arbitrary amount of "power" to the final character) increases exponentially with the number of those choices.

As a point-buy system gets more and more complex, the likelihood that any reasonable number of designers and playtesters will be able to find the magic combination(s) of choices that sum up to an "I win" button approaches zero.  They'll find and squash some, but the only to ensure they get all of them is test all the possible combinations, which isn't feasible.  Also, any changes they make to nerf the "I win" button changes the state of the graph, which means they really should go back and test all the possible optimization paths again.

So expecting designers to be able to balance an open-choice point-buy system with any real number of choices  is a grave underestimation of just how hard and time-consuming this actually is.

If you can find a copy of the article, "Ogre and the Fuzzy-Wuzzy Fallacy" describes a real-world example of the customer base finding a major optimization path almost immediately in a published game that completely broke the gameplay.

Quote from: Fiasco;563150Certainly an abuse of a poorly designed rule

The essence of CharOp!

Quote from: Fiasco;563150All you can really say is that there has progressively been more and more scope for charop since the start of the hobby.

This is certainly fair; I would however say that 3E introduced more - and more mechanically important - fiddly bits, though, which gave that much more to optimize with. I'd say there was a big uptick there.

Since the early 80s when I won a AD&D 1st elimination contest by making a 7th level Druid with selected magic items.

I remember some of the 2E Character Kits had awesome attached, but it also had some serious restrictions at the same time (Wu Jen, I'm looking at you!).

They weren't always "fair and balanced", but that mattered more on play style IME, than anything mechanical.

Quote from: daniel_ream;563153The Traveling Salesman problem is a famous problem in optimization.

Don't read the Wikipedia page; I've studied graph theory and their description is so jargon heavy it makes my brain hurt.

In short, the time it takes to find the "best" set of choices in a point-buy system (each choice presumably contributing some arbitrary amount of "power" to the final character) increases exponentially with the number of those choices.

As a point-buy system gets more and more complex, the likelihood that any reasonable number of designers and playtesters will be able to find the magic combination(s) of choices that sum up to an "I win" button approaches zero.  They'll find and squash some, but the only to ensure they get all of them is test all the possible combinations, which isn't feasible.  Also, any changes they make to nerf the "I win" button changes the state of the graph, which means they really should go back and test all the possible optimization paths again.

So expecting designers to be able to balance an open-choice point-buy system with any real number of choices  is a grave underestimation of just how hard and time-consuming this actually is.

If you can find a copy of the article, "Ogre and the Fuzzy-Wuzzy Fallacy" describes a real-world example of the customer base finding a major optimization path almost immediately in a published game that completely broke the gameplay.

Thanks for clarifying. Hmm.
I'm wondering if some point-buy systems - such as skills & powers for 2E - where point-costs of abilities themselves could vary depending on other choices such as class and race - could be considered to be a variant of the 'Travelling Purchaser' problem even, rather than 'Travelling Salesman'.

Quote from: Bloody Stupid Johnson;563178Thanks for clarifying. Hmm.
I'm wondering if some point-buy systems - such as skills & powers for 2E - where point-costs of abilities themselves could vary depending on other choices such as class and race - could be considered to be a variant of the 'Travelling Purchaser' problem even, rather than 'Travelling Salesman'.

Technically to be the TSP, you have to be able to model the problem as a graph where you move from point and point, and finish up where you start, visiting each point only once.  Character build systems don't normally work like that; the TSP is mostly just a shorthand way of pointing out that problems that sound simple on paper ("what's the cheapest round trip around all these cities" "what's the cheapest set of skills & powers that will give me the most bonuses") are often computationally hard in practice.

TL;DR: balancing a point-buy system with more than a handful of choices is a lot harder than it sounds.

Quote from: Bloody Stupid Johnson;563178Thanks for clarifying. Hmm.
I'm wondering if some point-buy systems - such as skills & powers for 2E - where point-costs of abilities themselves could vary depending on other choices such as class and race - could be considered to be a variant of the 'Travelling Purchaser' problem even, rather than 'Travelling Salesman'.

Both of these are concerned with the most optimal path, however.  There are many other CharOp combinations that are not the absolute optimal, but still cause massive problems at the table.  I don't think it requires solving either travelling problem to catch most of the worst abuses.

And even so, a modern personal computer can solve a 10,000 node Travelling Salesman Problem in a matter of minutes.  Well, probably closer to 30mins.  Regardless, there are no RPGs that have 10,000 nodes (decisions) when generating a character.  It would take far longer to decide on the weighting for the paths than to solve the (maybe) thousand nodes most games would have.  Plus, there are a few tricks that can speed up computation.

Quote from: daniel_ream;563153In short, the time it takes to find the "best" set of choices in a point-buy system (each choice presumably contributing some arbitrary amount of "power" to the final character) increases exponentially with the number of those choices.

As a point-buy system gets more and more complex, the likelihood that any reasonable number of designers and playtesters will be able to find the magic combination(s) of choices that sum up to an "I win" button approaches zero.  They'll find and squash some, but the only to ensure they get all of them is test all the possible combinations, which isn't feasible.  Also, any changes they make to nerf the "I win" button changes the state of the graph, which means they really should go back and test all the possible optimization paths again.

Yeah, this is the crux of the issue. Well put. Personally, I think the only really effective way to prevent CharOp-based rule abuse is through the social side of gaming.

As a piece of anecdotal evidence, when we started with 3.0 D&D in 2000, we allowed any formally legit character build into the game. This resulted in some game-breaking combinations, namely "I trip him" guy (he could trip anyone and anything, but little else), a multiclassed horror (Paladin/Monk/Sorcerer/Fighter/etc.; lousy combat ability but insanely high AC and saving throws), and a spiked chain specialist. After a while, it turned out this was causing problems, not to mention the characters were damn silly in play. The spiked chain guy had already retired on level 2, and the trip specialist was eaten by a purple worm on level 3 of Rappan Athuk (couldn't trip that, heh ). I gave the remaining character an in-setting opportunity to convert to a straight Paladin, and he took it after we discussed the problems the build was causing).

After that, we just agreed we would be sticking to more sensible class combinations, and focus on playing characters instead of building them, and after that, we encountered no more problems. We could have turned the whole game into an escalation of system mastery, which would have sucked for the people who weren't into that (including me), but I think we made the sensible choice. One guy was slightly unhappy about losing all those character-building opportunities, but then settled into playing a gnome Wizard/Tech Smith (with a shotgun and a mechanical servant), and had fun with the character.

Quote from: Melan;563252Yeah, this is the crux of the issue. Well put. Personally, I think the only really effective way to prevent CharOp-based rule abuse is through the social side of gaming.

I concur with this, and with a reasonable explanation of the fact that the GM will be fair to the players, they can even ignore gaping exploitation opportunities.

I made a point based character creation system for CP2020, and one of the primary loopholes in the initial concept of it was that it would (obviously) cost less to raise your stat a single point  than it would to raise all the skills governed by it during character creation.  So, for those looking to make the best of the best, a '10' in REF, INT, EMP, and BOD were common.

I had two solutions to this, the first was to sit everyone down and explain to them that the point based system was designed to allow them to play characters that were conceptually what they wanted to play, since the baseline CP2020 characters felt like idiot savants within their given field, not to allow them to make those same IS's with higher numbers and even more specialization.  They agreed, and made characters with a little more depth (for the most part...) For the first campaign this worked fine.

For the second campaign I adjusted the point costs so that they were on a sliding scale based on how high you wanted the stat to be... a 2 could be had for 1 point, but a 10 cost 40.  This skewed stats towards the average from then on, and people were more likely to wind up with 7-9 in their specific area of focus and average stats in the others.

Munchkin-ry has been around from the start I imagine. Even AD&D was fairly optimizable, depending on the DM.

http://www.rogermwilcox.com/ADnD/IUDC1.html  Some might find these stories amusing, or terribad.

CharOp has always been there

In 1e when that kid turned up with the 1/2 Orc Figther/assasin who wore a girlde of giant Strength was ambidextrous and weielded 2 magic short swords and used dust of disappearance that was CharOp. (suprise attack to assasinate or to minimally get a billion points of damage)

In fact you could argue the Druid, Ranger, Paladin were all CharOp.

In the end you could play a 'Paladin' using just he Fighter class but you get no magic powers.

Dragon, Behonder, White Dwarf all churned out munchkin classes. Fanzines were doing it for OD&D. The tale of the Vampire PC that inspired the Cleric is just a CharOp build.

An increase in options, growth in feats, skills etc and the Internet to discuss it all adds to it but a munchkin player is always a munchkin player.

MY cousins character Balthazar is a case in point.

1e - Start as a figther, get to 2nd level, dual class as a MU. Now you have 2d10 HP  Now keep your head down but you rapidly catch the party becuase the XP levels are low so soon you are a 3rd level wizard and you can start using all your Figther skills again.

Now you are a 3rd level wizard with 2d10 + 3d4 HP, you can use swords, wear armour when you run out of spells etc etc .

When we went 2e he retrofitted the character to have double specialisation as a 1st level fighter and single weapon style spec which gives him an addtional AC bonus.