Thoughts on Sims.

[swestrup]

So what follows is bound to be long and ultimately pretty boring. You were warned.

So, as everybody knows by now, I've installed "The Sims 8-in-1" . So far, I'm not too pleased with it, which is in some ways a relief. I don't really have 3.12 Gig free to spend on a game right now. Nor do I particularly have the time, but thats another story.

So, what don't I like about it?  Where should I begin???  I think I'll start by just listing all of the shortcoming, and maybe coming back later and trying to put them into some kind of reasonable shape:

1. The Sims' AI is laughable. They are entirely needs-driven and have no planning or learning ability whatsoever.
2. Despite the simplicity of the Sims AI, the system tries to ensure that it never has to worry about more than, I would guess, 20 sims at a time. It does this by restricting all activities to a single 'Lot' in the town or city, and never lets you leave a lot except via car or cab. A lot can have no more than 8 residents plus pets, plus guests.
   
3. The house build system is okay. It has a pretty good UI, but very limited abilty to specify where things are. They could learn a few things from ACAD when it comes to data structures for buildings. (Conversely, Autodesk could learn a few things about UI from The Sims house builder.)
4. Although the house build system is okay, much of the other UI needs work. Its a good start at workable, but it still has problems.
5. One big problem with the UI is that it assumes that there will seldom be more than a dozen of any one thing, be that heads, small appliances, wall coverings, etc.  As a result, by the time you have all of the expansion packs, and a bunch of the freebie packs off of the net, you find yourself clicking through hundreds of items, one by one. It needs a tree structure, and user-definable categories.
6. Because the Sims have no brains, its the objects they manipulate that know how to use them. For many specific operations, that's okay. The trash compactor embodies the knowledge of what it is to compact trash, what that looks like, and what game effect that has.  For other things, like having sex, its far more problematic. So, I've found now that there are specialized versions of beds, carpets, couches, tables, chairs, hot tubs, swiming pools, beach blankets, picnic blankets, etc, etc, etc that all have imbeded code to enable Sims to have sex on them. Now, mainly due to the fact that they all borrowed the same sex animation that one guy made, these codes are compatible. This does not necessarily follow.
7. There are many sets of clothing or 'skins' out there, and many of them require specific body plans to work. The problem is that there are only a few body plans supported by the system, and so you can easily end up with a situation where you can't support two different wardrobes at the same time.
8. The Sims seems to be highly extensible, but it isn't documented anywhere as to how to extend it, nor is there much available in the way of information about how to install a new object that someone has created. This is in spite of the fact that one of the stated purposes of the game was to allow for an on-line economy in the production and consumption of new objects.
9. Many of the settings and objects have counter-intuitive properties. For example, I had a character compain about hunger while on a picnic, and eating steadily. It seems that a picnic is a 'social' event and the fact that the animation shows people eating is to be ignored. In the same way, you can't eat at a restaurant by yourself. Restos are also entirely social things and thus you only get to eat if you get to a certain part of the social script. Blah.
10. Time doesn't make sense in the game. This has ALWAYS been a problem with how Maxis' simulations work, but I had hoped it would be fixed by now. Simply put, it should not take 3 hours to walk around across a plaza, around a fountain, and enter a restaurant. Worse yet, when you return from a day of shopping downtown, it you should arrive in the evening, not moments after you left...
11. They tried to make this a game, instead of a toy. The Sims should ideally be a game platform, and not a game per se. What game (if any) you play should be selectable at start.
    
12. Jobs 'exist' in the abstract only. If you're hungry when you leave to go to work, you'll be hungry when you return. Jobs apparantly have no food, shopping or bathroom facilities nearby. Whats with that???
    

Okay, so if that's my complaints, what do I think should be done about it? Well, first, one should determine if its worthwhile to fix. From the sheer amount of stuff available on the net involving the Sims, its clear that there IS a market niche for games like this, and a carefully thought-out market economy COULD be supported, so I would say its worth trying to 'fix' the problems. Secondly, because it can support a market, it can be a money-making proposition, even if it were open source!

So, here's what I propose:

1. Reimplement the Sims the right way, using:
   
   • Fully open source and documented APIs
   • object oriented programming with multiple inheritences
   • secure embeded code objects
   • a cryptographically secured object market
   • self-documenting objects.
2. The Sims use of skeletons (for animation), flesh (for filling out) and skins (for texture) is correct, but they need to be made more independant. A walking animation should be applicable to any biped meeting certain minimum conditions. Clothing should adapt itself to the structure of the person its draped over. This may take some work, but I've seen it done in high-end animation packages from places like Autodesk.
3. We need to support people, animals, robots and strange tentacular aliens equally well at all levels. Lets NOT be humanocentric.
   
4. The AI should be better fleshed out, and should be fully extensible. I know of at least one AI researcher who's trying to reverse engineer the Sims AI so he can replace it. This should be explicitly supported.
5. The AI must not only have need fulfilment, but as much of planning, goals, information (maps) and memes as possible.
6. Personal, local and social mores should be explicitly modelled, and be subject to change due to circumstances (including meme-spread). We want to be able to model behaviours as different as that found at a convent or at a swingers resort, and we need to be able to support BOTH in the same city. We want to be able to model Tehran, New Delhi, Singapore, Bhutan or San Francisco and have approprate local behavior.
   
7. Since this will be a simulation platform as much as a particular simulation, we'll want to be able to turn off entire ranges of needs and/or behaviours. If we are trying to model crime in a red-light district, we may want to allow the sims to kill, rape, maim, steal, torture, etc. In other simulations, like an attempt to model park uses, this may not be appropriate. Thus sex may or may not be actively modelled at any one time. The same should be true for Violence, Food and/or elimination, and many other features. The resulting models must be compatible.
   
8. We will want to develop skins and object packages for various eras and various places, since this will make a wonderful back end for any number of role-playing engines.
9. We must develop and document the object creation tools in parallel with the system, and make them available as early as possible. We also need to consider that many (most) folks interested in churning out renaisance hair-styles for us WONT be programmers.
10. To avoid huge installation problems, objects will need to be both self-contained and self-documenting. If you hand the sim system a widget, it will be able to figure out what to do with it, without prompting.  An object is allowed to have pre-requisite objects, behaviours, etc that it relies on, and we need to have an automatic way to fetch them when we do an install.
11. If this thing is to make money at all, we need to have a cryptographically secure way for objects to be signed and to be non-clonable (within certain reasons). We may want to have an artificial currency (and/or perhaps a real one) backing the sale and trade of objects on the web. The hope would be that the inclusion of a given object, be it a texture, body plan, animation, behavior, meme or whatever could automatically kick some fraction of a cent back to its implementor.
    

So, if we did all this, we would have something that would be useful as a research tool for, say, examining how fashions spread, figuring out how to introduce a stable democracy into an unstable African country, or how to mediate the interactions between two hostile religious/political camps. We would also have a gaming tool that allowed for dynamic illustration of online roleplaying scenarios; 'populous'-style rule-as-a-god type of games, or even sicko re-enactments of your favorite slasher flick. Presumably one would be able to download any number of 'games' that were based on this platform.
So, I wonder if I could get money to start this up?

Comments

[sps.livejournal.com]

Wellllll ... there certainly seems to be funding for random clones of any other Sim....

[tjernobyl.livejournal.com]

In the end, we'd have a fairly complete modelling environment for artificial semiintelligent life. Add a first-person 3D browser, and people would start moving in...

Tangental, but interesting: http://snowpath.sourceforge.net/

[swestrup]

>a fairly complete modelling environment for artificial semiintelligent life

Yup. Thats what I figured. Oh, and although I didn't mention a 3D walkthrough engine, I was figuring there would be one. If they can throw one into Dungeon Keeper, why not into a world simulator?

As for the snowpath thing, I think I see where you are coming from, but we don't NEED to find all of the good path-generation logic or understand how people choose routes or what-have-you ahead of time. So long as everything is nicely modular and plugable, folks should be able to try out new path algorithms as work progresses...

[tjernobyl.livejournal.com]

Open-source minds!