![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
andrewduckerThis started off when talking to Mike about his tweaking of the AI in Fable to make it work. I then realised that the principles applied to any system which tried to encapsulate higher order behaviours of complex systems in a simple ruleset. This includes everything from roleplaying systems to legal systems to software and I'm sure many others as well.
All feedback appreciated, as usual (as well as pointers to other things on the same topic).
Andy's first rule of systems: No system works in all situations.
Real life is complex (in the scientific meaning of the word) and fractal. At any level above the quantum, while it can be vaguely approximated by rules, there will always be situations where the results that rules produce aren’t what was originally expected or desired.
Andy's second rule of systems: Adjusting the system will take more time and money than you have.
There are an (effectively) infinite number of possibilities, tweaking the rules to each one in turn will therefore take forever, or until your budget runs out, whichever happens first.
Andy's third rule of systems: The result/effort ration decreases logarithmically as time goes on.
Your first rule will work in 90% of situations. The second one will fix 90% of the remaining solutions (and break 1% of the solutions you had fixed). Each successive rule will fix 90% of the remaining solutions (an ever decreasing amount) and break 1% of the currently fixed ones. It will not take long to reach the point of diminishing returns.
Andy's fourth rule of systems: One man's crust is another's complexity.
Eventually all systems reach the point where their complexity causes people to shudder when they think about changing them. Nobody will be able to understand them apart from a few people who have devoted their lives to them. These people will have almost certainly lost the ability to communicate with people who aren’t also heavily involved with the system (lawyers and computer programmers both fall into this category). However, before ripping it out and replacing it, remember that each of those rules is there because they fixed a problem. Any replacement system will need to cope with all of the situations the old one did. And it almost certainly won't until it reaches a similar state of crustiness (unless the situation the old system was designed to deal with has changed significantly).
All feedback appreciated, as usual (as well as pointers to other things on the same topic).
Andy's first rule of systems: No system works in all situations.
Real life is complex (in the scientific meaning of the word) and fractal. At any level above the quantum, while it can be vaguely approximated by rules, there will always be situations where the results that rules produce aren’t what was originally expected or desired.
Andy's second rule of systems: Adjusting the system will take more time and money than you have.
There are an (effectively) infinite number of possibilities, tweaking the rules to each one in turn will therefore take forever, or until your budget runs out, whichever happens first.
Andy's third rule of systems: The result/effort ration decreases logarithmically as time goes on.
Your first rule will work in 90% of situations. The second one will fix 90% of the remaining solutions (and break 1% of the solutions you had fixed). Each successive rule will fix 90% of the remaining solutions (an ever decreasing amount) and break 1% of the currently fixed ones. It will not take long to reach the point of diminishing returns.
Andy's fourth rule of systems: One man's crust is another's complexity.
Eventually all systems reach the point where their complexity causes people to shudder when they think about changing them. Nobody will be able to understand them apart from a few people who have devoted their lives to them. These people will have almost certainly lost the ability to communicate with people who aren’t also heavily involved with the system (lawyers and computer programmers both fall into this category). However, before ripping it out and replacing it, remember that each of those rules is there because they fixed a problem. Any replacement system will need to cope with all of the situations the old one did. And it almost certainly won't until it reaches a similar state of crustiness (unless the situation the old system was designed to deal with has changed significantly).
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
no subject
Date: 2003-09-28 11:01 am (UTC)I'm fairly comfortable with rule 1: there are glaring differences between writing music with an instrument versus with a computer.
I am currently the living embodiment of rules 2 and 3.
My last post is the desire to not become rule 4!
no subject
Date: 2003-09-28 11:10 am (UTC)Take humanity in Vampire. Ever found yourself disagreeing with the moral code outlined there? Or ever find that the rules for aiming, or ammo, or initiative break down?
Well, instead of trying to codify a complex issue (defining what it means to be human - ouch!), you can you can have systems that (in a sense) produce behavior. Have a look atMy Life With Master.
Er, I'll explain this better in a bit. Gotta go make spagbol.
no subject
Date: 2003-09-28 11:25 am (UTC)You can also develop games that are deductive. Like, Monopoly or chess, for example. There's nothing to argue about when one plays Monopoly, because Monopoly isn't trying to model the wide, wide world of economics and rent control. Neither does chess try to model warfare.
no subject
Date: 2003-09-28 12:10 pm (UTC)Logical systems that are entirely separate from real life don't have to worry about dissonance between the two systems.
Oh, and of course ones that allow a lot of leeway for non-mechanical actions are different too - storytelling games where the rules say "And then the next player describes something fun" obviously don't have the problem either.
Which is why you need legal systems that eventually boil down to a bunch of judges saying "That seems fine to us" or a jury saying "We don't care what the law says, we're not finding them guilty." , because otherwise law becomes a game where the person who finds the most loopholes wins.
no subject
Date: 2003-09-28 03:57 pm (UTC)These rules do seem pretty much true of law I can confirm. Surprised you get this actualy (she said patronisingly) - it's a nuance of law most non-lawyers don't apprehend, particularly AI people (for my sins I used to researchj in AI and law). They think it's a precise rule based system, like algebra, only writen in natural language, and can't understand why you can't just formalise it in Prolog and chuck the computer judge onto the laptop.
no subject
Date: 2003-09-29 05:03 am (UTC)HMC&E (Her Majerster's Custard and Exercise) argued that a jaffa cake was a biscuit. You pay VAT on biscuits. You don't pay VAT on cakes. Don't ask why, no one really knows. (Or cares.)
McVitties produced a giant Jaffa Cake, and brought it into the court room. After the Judge saw it, he concluded that they were clearly cakes, and no VAT was owed.
Try finding a computer that can decide stuff like that.
:o
Adam
no subject
Date: 2003-09-29 02:13 pm (UTC)I am very smart :->
and lots of systemising people don't realise that systems can't cover all eventualities.
A quick read of "Chaos" or something else covering complexity theory would be helpful.
no subject
Date: 2003-09-29 02:29 pm (UTC)I don't need to read Chaos. I am living it. Passim..
(Oh god, awful thought - is this the equivalent of these dreadful people who have mugs saying "You Don't Have To Be Mad To Work here But It Helps"?) Kill me now.
no subject
Date: 2003-09-29 02:35 pm (UTC)no subject
Date: 2003-09-29 02:39 pm (UTC)no subject
Date: 2003-09-29 02:44 pm (UTC)no subject
Date: 2003-09-29 02:49 pm (UTC)I thought you were opposed to marriage on principle? (Gold stars tho are always welcome..)
no subject
Date: 2003-09-29 02:56 pm (UTC)For some reason, tho, I like the idea of civil partnership.
No, I don't understand either.
no subject
Date: 2003-09-29 03:07 pm (UTC)no subject
Date: 2003-09-29 03:07 pm (UTC)no subject
Date: 2003-09-29 03:13 pm (UTC)And yes, I'm torn apart by my conflicting romantic/sceptic tendencies.
I'd love to believe love is forever, but I can't. And therefore pledging endless love just strikes me as hypocritical.
But saying "When I die I want this person to be looked after, and they should be allowed to visit me in hospital and have access to any kids." strikes me as a very reasonable thing.
no subject
Date: 2003-09-29 04:46 pm (UTC)no subject
Date: 2003-09-29 01:25 am (UTC)Think about it.
no subject
Date: 2003-09-29 01:29 am (UTC)But then I think that some of the best and most useful writing in the world and in history consists of precisey that - stating the 'obvious' in a way that everybody can take in.
can i disagree
Date: 2003-09-29 12:16 pm (UTC)i'm not acedemic enought to argue it, but somthing around case statemens having else clauses, and programs having exception handeling, means all_other options are covered??
or am i missing the point?
Re: can i disagree
Date: 2003-09-29 02:11 pm (UTC)Re: can i disagree
Date: 2003-09-29 02:19 pm (UTC)whats wrong with "To cover things outside of the area I need to, do a single thing." as a valid answear to a situation?
Re: can i disagree
Date: 2003-09-29 02:30 pm (UTC)Name a system that interacts with real life that doesn't occasionally hit a problem where the user wants to type in input that's not allowed, or deal with a situation the original design didn't allow for...
no subject
Date: 2003-09-30 01:12 am (UTC)Wow. Really? Taht explains a LOT. Here was me thinking that everybody was pretty much au fait with the basic untameableness of existence and were just being dickheads to annoy me.
Will I ever learn that assuming people think even remotely like me is doomed to abject failure?