When I said use a mini periodic table, I also meant the chemistry that went with it, ie, valencies etc, so that for instance, group 1 bonds well with group 5 I think ( the flourine group - it''s been a bit since I did chemistry ).

sorry, my post got cut short because i wanted to leave work more than i wanted to keep typing
but anyways, i think UltimaIV had a great system: the different "reagents" did different things, so you could learn that "black pearl" made a long-range spell, and "nightshade" had a "kill things" effect; sulphurous ash either added a "light" or "firey" effect. so, knowing this, you could make an educated guess as to what to put in a spell to make it have the proper effects.
i am not saying that you should copy this system, but i like the idea that you can learn what some of the individual elements do; in a more complex system, you could have interactions between the elements also (ie. element1 is explosive, element2 is poisonous, but element1+element2 have some chemical reaction that causes healing). this way, after learning some stuff about some of the elements, the players can research and experiment based on their knowledge, but they had best be careful with their mixing. the addition of these special combinations will keep it interesting, so that a player might assume that adding tons of element1 to everything to make it blow up could eventually mix it the wrong way and end up healing an enemy and learn a whole new thing.
as far as the many many combinations, a whole lot of combinations would do nothing useful (either fizzle, turn some color, just sit there and make you wonder if you should drink it or not). this is something i learned when i got a hobby chemsitry set when i was little. there is a lot more to being a mad scientist than mixing random things and hoping for something to happen
damn, that would be neato.
i don''t think you should go as far as valence electrons and stuff; this is just personal taste, but that sounds too much like high school for it to be fun. you can make a believable and fun system without forcing the players to learn annoying details.

--- krez (krezisback@aol.com)

whether or not base elements mix well would be determined in the non-reversible mixing function that produced the characteristics (properties) of the resultant. im shur we would need some in game chemical rules so the player would have some idea what to try and mix to produce a desired material. there also needs to be some results that are unexpected as throughout history chemical-mistakes have been sustantially helpful to the world. i still havent thought of a good way for the player to experiment and develop new chemicals quickly but not just brute force. i think automation is needed because of the high number of possible mixtures.
if the automation process could try 1000 mixtures a second, it would take about a minute to find all 4 element mixtures, so after a minute the player would know all 4 element mixtures, what fun would that be? i dont think slowing it down would work, cuz then the game would simply be reverse engineered and sped up by some user so they could get an edge over everybody else. perhaps the non-reversible mixing function (MF from here on out for "mixing function") would be some sort of hash function that in itself could be designed to take a while to process? the MF could take a string of elements in a specific order, and produce a n-bit output that contains the properties of that mixture. properties could contain explosive power, mass (weight), armor-protectiveness level, PH level (a strong acid might be useful, but then youd need to find a strong base so you could hold it), or perhaps even a new type of sensor for the bot (in my game your entity is a robot you design and build and program the A.I. for, which can run semi-autonomous or autonomous).

i dont know. these are just ideas im throwing out. more heads are better than one. (depending on who the head belongs to i guess).

what do you all think?

- jeremiah
http://fakemind.com

great ideas krez. having most mixtures "either fizzle, turn some color, just sit there and make you wonder if you should drink it or not" (hehe) would make it more realistic and fun. i also really like the idea for the user to know some idea about basic properties like you said but add in some "randomness"...

quote:
the addition of these special combinations will keep it interesting, so that a player might assume that adding tons of element1 to everything to make it blow up could eventually mix it the wrong way and end up healing an enemy and learn a whole new thing.

i completely agree.

i havent really thought much about how to impliment the MF, although its probably the most important part of the whole system. any ideas?

- jeremiah
http://fakemind.com

Edited by - fakemind on November 19, 2001 7:22:19 PM

http://www.foresight.org/Conferences/MNT9/Abstracts/Parker/

These people may know some things that would interest you. Certainly their work is the most relevant application of my own interest in this area that I''ve ever seen.

ld

not to be a stick in the mud, but why do you want the players to have the computer mix batches of chemicals automatically? wouldn't this take the whole point out of doing this system? even if it takes some amount of time, a player can set it up to mix several things in different orders and the go take a nap.
if they know a formula will work, they should be able to have it mix a certain amount (mix 15 pounds of element1 and 10 pounds of element2 to get 25 pounds of, um... well, whatever)... but just telling the computer "mix all possible combinations of elements 1 through six and report back" would take the fun out of it.
that's just my view on it though.

um, just out of curiousity, what type of game is this going to be? the only thing i picked up so far is that it is online and has robots.

--- krez (krezisback@aol.com)

Edited by - krez on November 19, 2001 8:46:22 PM

nobody is a stick in the mud simply for having an opinion. *smiles*

----

the game is one ive been wanting to make for years. its not going to be a multiplayer-only game, just that an online multiplayer play mode will be there. the game is going to take place in a top-down tiled world. first of all you will get to design your robot. its transportation (tread, hover, wheeled, etc), its weapon, its armor and shield if any (armor = material on robot to protect it, shield = energy field .. think star trek), its sensor to locate other objects and robots (heat, movement, etc), and other tangible features. then you will have to program the artificial intelligence of your robot. i guess you dont have to, you could run your bot manually, but it wouldnt be as good as if it had A.I.. anyway, the scripting language for your bot''s A.I. will be similar to C, and will be based off a finite state machine. in semi-autonomous mode you give orders to your bot of when to switch modes, and you can also send coordinate points to your bot with the mouse. (ie. switch your bot to "attack" mode, and then click an area you want him to focus around). in fully autonomous mode your bot will run complely on its A.I. alone. there will be a 1 player mode that will basically have a progressing adventure where you will be able to re-program your bot between stages and change his physical attributes as well, depending on the task at hand. in multiplayer mode your bot will fight against other bots to see who is the best of the best, and this is where the multiplayer mode will come into play. connect to your friend and have a battle. there could even be teams where you program/design a group of, oh, say, 5 bots against your friends 5 bots. imagine a battle of 10 bots each designed and programmed to a particular strategy where each bot can communicate with its teammates or try and jam its opponents radio frequencies. with fully customizable bots, weapons, and other features thanx to things such as in-game chemistry as well as other features im going to add. (ie. another feature is a way to have the players be able to design their weapons, so there will be no "you can only pick these 20 pre-set hard-wired-into-the-game weapons", you''ll get to design your own based upon basic components)

anyway, thats the idea of it, keep in mind ive just started. so far ive got a robot running around a tiled world, so im just to the point of "its working, now i need to add in all the features".

----

now back to in-game chemistry:
well, the imphasis is not going to be so much as the player mixing elements and finding different materials on its own, but rather the vast variety of chemical resultants that will make each players bot unique. kind of like each player will have a different set of elements they have researched themselves, which i think will add in a huge replayability to the game and keep each individual robot special.
i guess, for my game at least, the in game chemistry is simply a way to keep it new, afresh, a high amount of replayability and unique-ness to the players bot, and not necessarily a big part of the game. .. i hope that makes sense, sometimes things make sense in your head, but not when you try to communicate them.

- jeremiah
http://fakemind.com

very interesting...
definitely not a game for those people who just like to run around killing stuff (unless they also have an intellectual side).
scripting the AI with a c-like language? ouch seriously, though, this sounds like fun (do i sound like a dork now?).
when you have a website up about this game (if you do) be sure to put a link here so i can check it out. this sounds like one that i will want to follow.

--- krez (krezisback@aol.com)

I only have a limited amount of chem knowledge but I don''t see how that can work with the player just screwing with chemicals. The sheer amount of combinations out there would mean the player would have very little luck with trial and error and would need too much chem knowledge to have even a chance at getting something. However if you simplified it like Ultima IV then by all means go for it.

However I propose an alternative:

Go ahead and make a fairly realistic (or not, concept is the same) chemistry system, but have it so that you need to get scientists to research different materials. Then instead of the player getting frustrated, the game does that and the player gets the satisfaction of the end result. If you add an element of randomness into the algorithm that finds these materials, each person will still be discovering different stuff but the frustration factor won''t be there. And people could also manually make the materials, so you would have people online posting differenting materials and trading them and stuff, that might be cool. Or I''m a fool, and this is a bad idea, I don''t know.

I like the idea of each base element having a vague property, like "kills stuff" or "makes stronger." Then the player has a little lab with a chemical mixer. They can put, say, 1 to 5 units of one chemical in the left side and 1 to 5 of a second in the right side, hit the go button, some nifty graphical effect happens and BAM they have some amount of a resultant chemical.

They don''t necessarily get the same amount of stuff back, either. Like 3 units of chemical A and 4 units of chemical B might give you 3 units of chemical A3B4, etc.

Now, there would be a very well defined set of rules that determine what each combination does. But you wouldn''t tell the player what it was - the fun would come from experimenting to figure out which element was the "make stronger reactions" chemical and whatnot.

So each chemical would have basic properties as to its effects on "stuff" (other robots, etc) and basic properties as to its effects on chemical reactions. "stuff" properties would include explosiveness, corrosiveness, stickiness, and, uh, vandalismness (lots of neon green and pink robots going around - fine, I couldn''t think of another property ;p) The reaction properties would have the same 4 (or whatever number of "Stuff" properties) columns and mathematical formulas for calculating strengths of resultant chemicals. Ok, that was written pretty poorly, so here''s an example to illustrate:

Chemical 1: Polydorkaphone (Pd)
S-Properties
E: 1
C: 1
S: 8
V: 10

R-Properties:
E: E_t + E_o
C: C_t + C_o
S: S_t + S_o
V: V_t + V_o

So, this chemical is basically neon glue that makes other robots look like statues in a city of radioactive pigeons. The R-Property formulas are all "t+o" which stands for this+other. So if you add this chemical to another chemical the resultant properties will simply add together.

Chemical 2: Soboredazine (Sb)
S-Properties:
E: 10
C: 0
S: 0
V: 0

R-Properties:
E: E_t + E_o
C: C_t + C_o
S: S_t + S_o
V: V_t + V_o

So this is just TNT. It also has the same simple reaction properties of t+o.

Now here is how the mixing works:

Pd + Sb would simply be:
PdSb:
E: 11
C: 1
S: 8
V: 10

because the averages of the R-Property functions would give those results.

Now, say you want a chemical to exhibit odd behavior if a certain limit is reached. For example, Sb is highly explosive unless mixed with a certain amount of stickiness, in which case it becomes SO sticky that it holds the explosion together and both kind of cancel out. You just need to make the R-Properties a little more complicated:

Reaction Properties:
E: if (S_o > E_t) 0; else E_t+E_o

and etc. If you wanted to make a chemical do something odd like decrease in stickiness when another was added to it, you could say:

Reaction Properties:
S: S_t / S_o

or something like that. This is an extremely versatile system because of the different ways that the formulas could be mixed. When averaging the result you could simply weight by the number of units of each chemical to give biases towards whichever they put in more of, or use a more complicated system.

The trick would be calculating resultant Reaction Properties. Averaging formulas is obviously more complicated than averaging numbers.. if you were to use this method you could do something like assign a dominance level to each chemical and simply use all of the more dominant chemical''s reaction properties instead of trying to find a compromise.

Anyway there are some unflushed out ideas in here but I think using this method would be really, really cool. If you created a reaction formula generator then it would be easy to come up with a new set of elements in each game, so the process of research would not be lost after the first play. You wouldn''t want to have too many base elements (probably having the same number of elements as you do properties, in this case, 4, would be good) because then the research would get very difficult. But the inability to mix more than 2 at a time would keep research simple for people that really just want a more explosive robot. It would be pretty simple to figure out the basic qualities of each chemical at the initial level. But other people who are more interested in this puzzle aspect of the game than other areas would also have the challenge of figuring out not only the S-Properties but the R-Properties. In some games (with the randomly generated properties and all) they would just get the satisfaction of solving the puzzle. They could use the information for trading purposes, and whatnot. I think the whole idea of trading information is really insanely cool, but that''s another story altogether. In other games, though, they might get lucky and find some really kick ass chemical. Sweet!

Heh, what a cool idea.