Miscellanea (I)

[trollson]

I'll start here with a few simple comments which could be applied server-side without affecting the client.

1.  Dropping Stacks

Rather than applying a drop-chance to each stack in an inventory, simulate applying that chance to each item in the stack.

Obviously, if you use a simple random number generator (RNG), then this would be a ridiculous expense - a stack of 50,000 gc would take 50,000 RNG calls to split!

Fortunately, you can use a binomial distribution function to do this for you, at a cost of ~2 RNG calls internally, rather than one call per item in the stack.

For example, the GNU Scientific Library provides an implementation of a Binomial distribution:

Code:

const gsl_rng* rng; // RNG object
double prob; // Probability of dropping each item, 0..1
unsigned int quantity; // Number of items in the stack.
// ...
unsigned int number_dropped = gsl_ran_binomial (rng, prob, quantity);

Note:  You may say "why not drop 60% of the stack instead?"; this is quite different.  While the two approaches are much the same for large quantities, they have quite different results for small stacks.  Dropping 60% of a stack of 4 will always drop two, and maybe (40%) three, depending on your implementation; whereas the binomial system could drop any number, 0 to 4.

Once you are using binomial distributions to split stacks, you can play interesting games; such as dropping 40% and destroying 10% of the items.  Then ask, was your Sword of Gory destroyed or is your Valet being sly?

You can also use this approach to split a monster's treasure between a party, as was being discussed in another thread.


2.  Monster Treasure Tables

In EL each monster's drops are defined by a list unique to that creature; item, quantity range, and chance.

This seems to me to be too predictable and somewhat limited; a retrograde model compared with the old D&D "treasure tables" approach.

Some creatures may yield unique items (typically body parts), but any intelligent creature with opposing thumbs is likely to drop a wider range of items.

Create a series of tables of treasure (item, quantity-range), each table represents a set of items which may be found on some type of entity (creature, monster, professional).  Each creature is then given a number of rolls on one or more tables.

If you want more control, then make the indexing rolls use gaussian distributions, with a mean and offset; place items in the tables in the appropriate order, and you can have some influence over the subset of a table that is predominately found on a creature.

Then adding a new item to the treasure tables is easier (single point) than adding to to a host of creatures' drop lists.

Some creatures may still have a roll on a table unique to that creature, for specific (anatomical) drops.  However, I'd suggest anatomical drops are better obtained by a task on the corpse rather than randomly being dropped.

3.  Experience for success or failure?

In most of these games you get experience, and thus improvement, for success.  The argument (not that it really matters) is that you've done well, so you've learned something.

This doesn't really stand up to scrutiny; it just encourages grind.  After you've made your 10,000th leather helmet you are not really learning anything new, and how does stitching leather teach you to forge swords anyway?...

It is worth considering the alternatives.

Firstly, experience on non-critical failure, not on success; you learn by your mistakes.

This is useful in an open ended RPG model, as you cannot advance very far through repetitive grind.  Any task of a certain difficulty generates you less and less experience as you get better.

In combat, you get little experience for picking on opponents much weaker than yourself.

Alternatively, add a separate experience roll following successfully executing a task; the experience roll has the same probability as the previous roll, except you must fail in order to have learned something new.  This means that you get best experience from tasks with ~50% success rate.

Some may recall this is the system used by the Runequest RPG

Personally, I wouldn't give any experience for just succeeding in tasks; rather I'd employ a system of reward for achieving quests/goals/jobs, and let the players spend the reward increasing any skill they have.  But that requires much more work in other areas...


4.  Task Success Rolls

Any experience-earning action performed by a character is a task, which compares some skill or attribute against some difficulty level.   For character-character interactions, the difficulty comes from the other character's co-skill or co-attribute.

If a difficulty is equal to the skill (or attribute...), then the chance of success is 50%.  As skill and difficulty diverge, the chance tends towards 0 or 100%.

But don't let anything get impossible or certain, regardless of the difference.

Simplest application:  Chance of success is (skill - difficulty), and compare to a roll from a Gaussian distribution (again, look to the GSL); if roll is less, then success.  The Gaussian produces values weighted to zero, range (-infinity .. +infinity), so no task is certain or impossible.

The Gaussian will need to be scaled, such that a chance (difference) of +1 has, say, 55% success rate (this can be calculated).

Also, you can use this to generate a degree of success, rather than simple pass/fail:

Code:

success = skill - difficulty + Gaussian

For example, you may choose to treat successes greater than a certain magnitude as critical results; though there are more interesting approaches than this.

trollson

[Suneidesis]

You put some time into this NICE

[krayon]

  OK...
  As always, you've posted quite a detailed list of ideas here

Thanks a lot trollson!  I have read many of your suggestions on the EL forums and really like your ideas.  I hope you don't mind if I impliment (probably quite) a few?

[trollson]

5.  Experience Tables

EL uses exponential experience tables, in common with many other RPGs, and originating from D&D.  Perhaps we should question the merits of this choice?

With an exponential system, the experience required to reach the next level is some multiple of that required for the previous level.  For example, 1000, 2000, 4000, 8000, 16,000, ...

It was recognised long ago that this works well for lower levels, but rapidly becomes disheartening, when you need millions of experience and the next level seems an eternity away; especially in games where there are few other marks of progress.

Even D&D recognised this; at higher levels the experience tables were levelled out to a constant amount, and EL does something similar.

Why use an exponential table?  They keep levels in check, as progress is slowed (sometimes quite drastically) at the higher levels.  In systems were each level brings new abilities, this saves a lot of game development.

But in CRPGs like this one and EL, levels refer to skill ability, which is compared against the difficulty of a task (or opponent's skill level) to determine an outcome.  There is no new game design required for the higher levels; and where new features become available depending on level, that level is just a chosen number.

So CRPGs cling to D&D level tables even though they are a product of game mechanics which no longer apply?

Of course, levels should be progressively harder to achieve, but perhaps the exponential increase is too harsh;  I would suggest a geometric progression instead.

With a geometric progression, the experience needed to obtain the next level is proportional to the level; if level 10 cost 10,000 xp, then level 11 will cost another 11,000 xp, level 12 another 12,000 xp, ...  You only face a gradual increase in effort each level, just a (constant) bit more than last time, and not a disheartening multiple.

This does mean that more levels will be obtained.  But, the rest of the game system controls what a level actually means; the distinct advantage is that the players are presented with a more continual progress.

If using the skill level vs. task difficulty model, then using the gaussian rolls described in (4) above will limit how "overwhelming" large skill differences are; it is the different in skill levels, not their ratio, which is being compared.

For skills, the geometric tables still work with regard to EL-like systems.  However, more consideration is needed when considering "Pick Points" generated by OA levels;  geometric progression means more levels, which give more pick points.

Personally, I'd have experience earned centrally, then "spent" on increasing skills -- and attributes (as skills).  Advantage -- more control over your character development, and the game only needs to record one experience count, as players' would buy full levels in the skills.  But that is another system...


6.  Current and Permanent Attribute and Skill Values

Notice that all skills and attributes in EL are recorded as "A/B"?  The first number is the current value of that attribute, the second is your "normal" value.  Drinking potions and the like can change "A", while "B" is permanent until increased by a level change.

Similar to health and other character parameters, each minute the "A" value moves towards the "B" value by one point.

Really, EL has made very little use of this mechanism.  If attributes are more significant in the game -- and I would definitely put attributes, nexus, and skills all on an equal footing -- then changes to attributes should be used more.

Attribute and skill reduction could come with failed tasks, and could replace cool-down, without making players wait before they can do something else (and making players wait isn't a good idea).

Each time an attribute or skill is used, decrease the current value by one.  This affects performance, and gives those attribute/skill boosting potions a second use.

It also stops the "click-fail-click-fail-...-click-fail-click-success!" play.  If you cannot pick a lock, then if you try again immediately, you have a lower chance.

If a high level fighter is engaged in a drawn out battle becomes more vulnerable (tired), which could lead to subsequent attacks by lower-level opportunists.  Keep them on their toes!   

One idea I have toyed with is requiring a successful attack roll to initiate combat.  This was to support the "law level" concept, to prevent low-levelled newbies accidentally clicking on a passing high level character and starting a (very brief) fight.  If the "initiate" roll fails, then they fail to initiate and the target can continue on their way.

Of course, in that system attempting to initiate a combat increased you current "outlaw" level, making it easier for other characters to attack you...

If experience is calculated as a function of level, then that should be based on the permanent value rather than the current, to avoid exploitation of negative adjustments.  However, if you design the game right, there should be no need to adjust experience for level -- all that is handled by the experience tables!

[trollson]

7.  EL Grips

7.1  "Its fantasy, we don't have to be like the real world..."

_{rant alert!}

This is an excuse rolled out whenever something blatantly daft is added to the game, typically to cover up some other fault.  A good fantasy world follows the real one closely, except in a few places; it is the familiarity that makes the fantasy stand out.  Overriding common sense and common knowledge just produces Bizzaro World.

Examples of this:

(a)  Mining with Swords -- a cure for the mountains of useless swords that had accumulated, and to add cost.

(b)  Making bronzer requires advanced iron metallurgy -- right, so the bronze age came after the iron age here?  Again, no reason for this other than bronze came after iron in the game development, and anything new has to be higher level right?..

(c)  Hyperspace Keys -- err... buried treasure would describe the effects far better and without the random fantasy components.


7.2  Economy Matters

EL has never got to grips with the problem of the economic system in an MMORPG; although on his blog Entropy feels confident enough to write articles on the subject.

Value is determined by supply and demand:
(i) Harvestables are available in unlimited quantities, so in general have zero value.
(ii) Animal parts are limited by spawn rates, so could hold some value.
(iii)  Exotics (stones, enhanced, modables) are extremely rare, and so have a strong value.

Since EL is a grind-for-experience game, characters are rewarded for actions, harvesting and creating items.  This reward deducts from the value of items -- in effect, harvestables and basic goods (no exotics or animal parts) have negative value in the economy.

Get rid of grind-for-experience throughout the game, and have all resources in limited supply, and there may be an economy.

*  Experience on fail (discussed in (3)) should reduce items produced through grind, if not the grind itself.
*  Separate learning tasks (producing experience) and creation tasks (producing items) would do the same; but I don't like this unless the same rules could be applied to all tasks (training fighting vs bloody combat, etc).
*  Giving experience for achieving goals, not for tasks, would be my preference, but requires a lot of work elsewhere to provide characters with measurable goals.


7.3  Cash Flow

Closely related to the economic problem is that of cash flow.  It seems that every innovation in the game has as a strong driver the desire to add another "gold sink".

This is another sign of a failed economic model.

Most of the "gold sinks" are aimed at high-value/high-level tasks.  This affects a small portion of the character population, and can be ignored if you don't want to participate.

In addition, with free unlimited resources and a handy NPC with limitless funds, any gold sink can be countered by the lilac bush.

A more effective gold sink would be a very small but pervasive one, such as charging a few coins for each boat trip.  Count up how many boat trips are made each hour in the game, and multiply that by 1-2gc!

But gold sinks are band-aids for the problem; which is that gold is another unlimited resource in the game, via NPCs and their unlimited pockets and unlimited demand.

So if you want a stable currency, limit those NPCs.  Model NPCs more after the trader bots or "Merchant NPCs", who have finite resources and demands.

But, it must be kept fair for all players; a high level "truck" shouldn't arrive early and soak up all the NPC's gold for the day, leaving the NPC unresponsive when newbies arrive with their first bag of harvestables.

(i)  NPCs obay the rules of supply and demand, and have finite resources (though can go into debt).
(ii) NPCs trade with a "out-of-game" entity, at some background rate.
(iii)  The price offered to characters depends on their level and quantity being traded.
(iv)  Track how much characters have made from NPCs during the day and factor that in somehow.

Making a stable supply-and-demand economy were prices vary accordingly is not difficult.  Making one which is fair to all players is.

curtailed due to time

update 2009-01-06

7.4  Anachronisms

• Rectangular hay bails - where are the combine harvesters which produce these?

[EndorVaeros]

Closely related to the economic problem is that of cash flow.  It seems that every innovation in the game has as a strong driver the desire to add another "gold sink".

My husband and I have talked at some length about this. Gold sinks do not do wonders to fix a broken economy. They usually encourage more gold to come into the game to pay for these items. A much smaller percentage of gold already in the game (compared to the sources coming in to pay for this item) actually goes out.

A much better solution would be to lower the price that NPCs sell for. Making NPCs sell for 10 times (sometimes more) what they bought it for does not encourage the high level people to buy the items. It encourages them to go out and get it for themselves, thus creating a larger profit margin for those people, and more money into the game. If it were possible on some items, just to break even (or possibly loose a small some) I know a lot of people that would sink their money into buying the ingredients simply because this is faster. [sarcasm] But that would increase how fast people level up, and we wouldn't want to do that, now would we? [/sarcasm]

(i)  NPCs obay the rules of supply and demand, and have finite resources (though can go into debt).
(ii) NPCs trade with a "out-of-game" entity, at some background rate.
(iii)  The price offered to characters depends on their level and quantity being traded.
(iv)  Track how much characters have made from NPCs during the day and factor that in somehow.

(i) Though there would have been a time when I would have agreed with you that having NPCs obey the rules of supply and demand would be a great thing, this still leaves a problem. The more people you put into game, the higher you have to increase what the NPC can buy. In some ways this would make a lot of sense, but there are things that wouldn't. Yes, with more people in game, supply and demand are bound to go up, but not as much as people think they would. This is because the common gamer is driven by the urge to make more money. This means buying as little as possible. Though this is corrected slightly here by having classes and not being able to level everything up, this is only going to fix things slightly. A decent number of people will just have second accounts to supply the things for themselves. If you want a good example of this, check out the PK server.

(ii) Having the NPCs trade with an "out-of-game" entity would be good, but again, if you are trying to go with supply and demand, this "out-of-game" entity would have to follow the same rules. After all, why would someone want to buy the same item every day?

(iii) This is a great idea, for some harvestables. And it would do a wonderful job of curbing the amount of money that suppliers/Artisans can get. But what about the fighter type classes? Ya know, the people who would stab out their eye if they tried to use a pick. If we have relatively common drops, such as bones and raw meat, the higher level mobs, as well as the lower level mobs would drop these items. (How does it make sense that a rabbit drops a piece of raw meat, but a dragon has none to speak of?) How do you balance this? This is not the best idea to implement until you can find a way to make it fair across all classes, and not just single one or two out.

(iv) If you are trying to make it have some semblance of sense, this would not make sense. If an NPC needs it, they are not going to start paying you less for the items they need, while paying the person standing next to you more. This encourages the use of multiple accounts, and not a more player-to-player driven economy. Some of the things that could be done to fix it would take a lot more coding than most developers are willing to do. This is because it usually just postpones the inevitable breaking of the economy, keeps it from happening.

[Side note] It would be great if people stop trying to clog threads with posts that do nothing to further the thread. Saying that it is a good idea isn't nesicarily a bad thing as it helps encourage more people to post ideas, but saying something completely unrelated gets very old. This is usually an attempt to up post count, or because someone feels the need to make themselves sound more important. It would be great if it stopped. But that is a discussion for another thread, please do not clog up this one with responces to this side note. [/Side note]

[trollson]

Gold sinks ... usually encourage more gold to come into the game to pay for these items. ... A much better solution would be to lower the price that NPCs sell for...

EL's NPCs, with fixed prices and infinite pockets, should not be considered as participants in the game economy, but as limits on it; they cap the price ranges in the game economy.  When NPC prices become too realistic, the player economy is backed into a corner and can suffer.

For example:  Medallions were a major source of income for our Merchant.  To help support crafters, the NPC buy price was raised to something more "realistic" - more than we were selling for!  Therefore our prices had to rise to match, but at that price point sales dried up.

(i) ... The more people you put into game, the higher you have to increase what the NPC can buy. ...
(ii) ... if you are trying to go with supply and demand, this "out-of-game" entity would have to follow the same rules. ...
(iii) This is a great idea, for some harvestables ... But what about the fighter type classes? ...
(iv) ... this would not make sense ...

Making a dynamic, self-balancing economy from the EL game model is not an easy task.

(i) As the game grows, this is represented by the NPC's business booming.  Therefore the NPC's resources increase, and prices are adjusted.

Clarification:  I would expect the system to self-adjust, so these changes would be automatic within the system.  I don't envisage any manual mucking around with buy-sell prices etc.

(ii) Assume that the game is set within a relatively small part of a much larger world, then we have an "intra-game" economy, which we are discussing, and a much larger "extra-game" economy.  The latter is large enough not to be affected by changes in the former.  So, when I talk of an "out-of-game" entity, I am implying trade with the "rest of the world".  This can also provide a stabilising force in the game economy, though less rigid than the fixed NPC prices (and of course the transport costs make relying on it less economical).

(iii) This was a "fudge-factor" suggestion.  If it is modelling anything realistic, it would be social standing or rank related.  If harvestables are a "spawned" resource, like creature drops, then its less relevant - though spawning has plenty of fair-use problems itself.

(iv) No, it doesn't.  Again its a fudge factor, and there should be better ways to share NPC's finite-resources around.


So problems we have:

(a)   There is no economy unless resources are finite (a sweeping statement to make a point!).
(b)   Finite resources need to be shared fairly between characters of different capabilities.

Part of this is that character progression is only measured by levels, from experience, from consuming resources and/or producing items.  ie, the grind.  This is more a feature of CRPG than RPG - on the tabletop we had many other ways to advance our characters other than level gain.

Can we:
(a)   Get rid of the grind?  Drive progress through achieving goals rather than repeating arbitrary actions?
(b)   Provide other achievements for characters, beyond level gains, which are meaningful in-game?

I have plans for (a), but there is a lot more work involved.

For (b), consider:
(i)   Ranks within groups - guilds, cities, institutions.  These need to be earned (not bought), and could grant access to "things".
(ii)   Positions within groups; unlike Ranks they are a limited resource, you cannot have two Captains of the Town Watch for instance.  So there is a game in holding, maintaining, and acquiring positions.

Both ranks and positions can be lost if neglected, and holding some may prevent access to others (cf. EL's churches).  But acquiring them should be (mostly) independent of levels, so they are not another by-product of the grind.

[trollson]

8. Law Level vs PK-Zones

This is an alternative to the black-white PK-Zones in games like EL, and it has been discussed on the EL Forums.

I will try to provide a brief summary here, some will need to be changed to reflect the QEL system:

• Each map tile has an associated "LAW LEVEL" value.  A high value means the area is very safe, while a low or negative value means it is dangerous.
• Each character has an OUTLAW Level, which may make them easer to be attacked in lawful areas.  This should be treated as an attribute (1/min decay back to base level).
• To initiate combat, the aggressor's ATTACK value plus LAWLEVEL must be less than the target's DEFENSE plus OUTLAW.
  
  
• To prevent newbies "accidentally" (but briefly) harassing more developed characters, combat is only initated if the attacker makes a successful attack role.  Failure costs 1 attack point, recovered at 1/min as usual (see section 6 above).
  
  OUTLAW can be modified by events:
  
  
• Equiping overt armour and weapons raises OUTLAW.
• Initiating combat temporarily raises OUTLAW.
• Failing a thief task temporarily raises OUTLAW if seen.
• Politics and inter-group dynamics will also modify the odds (cf. Guild Wars).
• Raising the "hue and cry" raises both your OUTLAW and that of the target.
  
  Local LAW LEVEL can change over time, in response to game-world events, or transitory events (spells, invasions).
  
  There could be chapters written on gaining and losing OUTLAW points.  OUTLAW is probably the flip-side of KARMA.
  
  The nice thing with this system is that:
  
• You can have safe routes through hostile territory -- stay on the path little hobbit!
  
• Big character's cannot easily pick on little characters; this was the biggest discouragement from getting involved in PK in EL, even in level-limited Arenas.
• It should integrate with concepts of outlaws and open warfare.

[trollson]

9.  Variable Spawn Strength

A very brief suggestion...

I'm sure that there has been talk of spawn monsters increasing in level like player characters.  The problem with this is that monsters "play" the game almost as much as an unemployed teen, and would eventually become overwhelming if they followed the same process as players.

Consider instead a simple self-balancing approach to the level of a spawn:

• Each time the monster is killed, its "spawn level" is increased.
• Each time the monster kills, its "spawn level" is decreased.

In this example, a creature would have to kill twice during its lifetime in order to decrease its level on the next spawning (+2 for two kills, -1 for being killed).

Constraints would have to be applied to avoid "unrealistic" results:

• Only certain types of creature should level; at least semi-intelligent or mythical, rabbits are unlikely to change.
• How "spawn level" translates into creature abilities may depend on the creature; we shouldn't need to use caps here, just decreasing returns as the values become more extreme.

So, consequences?

• An individual cannot camp on a spawn indefinitely, as the creature gets stronger far faster than the character.
• A spawn monster which develops too high a level for its area will be a notable risk for a while, but will eventually fade.

Exploits?

Camping a weak spawn to leave a powerful creature for "fun".

• Constraints should make this hard work, but not impossible.
• If the creature is too high level of the area, then it will score kills and drop back towards its "normal" level.
• The world shouldn't be a static background, keeps players on their toes etc.