Describing Noise

[Zell]

The Skotos engineering team has been charged with the task of writing a next-generation server for text-based virtual worlds. One of our main duties in the pursuit of this goal is to present dynamically constructed environment descriptions to our players. This means that when you walk into a room, you should get a chunk of text that as fully as possible describes all the things you virtually perceive about it.

The main component of this chunk of text is the "room description", which is typically hand-written by the game author using our browsed-based developer interface. While the server does allow this text to be interspersed with control sequences that impact the way it reads depending on e.g. the time of day or the weather, this room description is fairly static.

Thus we follow it up with a description of transient information; this includes a list of everything (mobile) that is currently -in- the room, such as other players, other living things, pieces of furniture, weapons, dead bodies, clothing, food, etc, etc. In the Skotos server, these things all have proximity, so the player will see what items lie where; who stands near the various exits, who is sitting around the large table, near the hearth, and so on.

That's all preliminary information since this Forum is quite new. Now, here's the topic with which I'm currently wrestling... I would like to summarize conversation levels for the player as well. When you walk into a place, perhaps a crowded one such as a pub, the fact is that you have, from an assortment of clues (mostly body language), a near-instantenous idea of what's going on where -- couples talking intimately in a corner, a rowdy table on their third round -- and I would like to make that impression available to the player as well.

So, let's say the goal is the produce a sentence such as e.g. "There is a heated conversation going on around the large round table and a muted dialogue in one of the booths. This room is quite noisy." This is a little more advanced than we can probably pull off just yet, but not by much.

The first question then to you, patient reader, is wether or not this is a desirable system at all. Can't people just hang out for a bit until they figure out where the conversations are at?

Next, if we do proceed, and I think we do regardless of your answer just now, actually http://www.skotos.net/ubb/wink.gif -- the implementation problems are two-fold; first, we must come up with some sensible measurement of conversation level, second, we must generate pleasant, readable English to describe these hard measurements. For the measurement problem, I'm using a basic exponential decay algorithm with about a one-minute half-time and which is stoked every time somebody says something. The descriptions are hazier at the moment...

... so stay tuned for the second part of this article, to follow when I've put in some more work. Meanwhile, I'd love some comments on the viability and usefulness of this idea.

Zell

[Erica]

I've always loved and believed in living breathing worlds. I'm trying to imagine how it would read to a player. As long as it remained clear and definitive to the reader and didn't confuse them then of course it would be a wonderful addition! Because I've never seen this sort of thing anywhere before, I am having a bit of trouble trying to get a good imagery of it though http://www.skotos.net/ubb/smile.gif

I look forward to seeing it as a reality, as to someone who is shy entering a tavern and seeing a heated argument they would know enough rigth away to sulk away from it, where as some law abiding and enforcing type may want to head directly to it, etc.

Whoo! I'm excited http://www.skotos.net/ubb/smile.gif

[JeffCrook]

Wow, if you can pull this off, it would be great. Sure, you could just wander around the room and find the conversations, but it would be so much more elegant to have things the way you describe it.

As for descriptions of noise level, perhaps descriptions could be based on three categories, depending on how noisy the area actually is. If not very noisy, you could give a broad description of the room's noise level, with details about heated conversations and whispered conferences. At higher noise levels, you might not be able to hear the heated argument across the room, only the one by the door, and whispered conversations would become impossible.... Question, would the noise level restrict use of such words as Whisper, as when talking.....
anyway, at very high noise levels, descriptions might limit descriptions only to what is immmediately around you, or even shouting in your ear.
I guess you could treat descriptions based on noise level the same way you treat them based on light levels.

But I do believe you should go ahead with this. it's a great idea.

[tanis]

I hope you're taking into consideration multilanguage capabilities. It would be nice to let players from foreign countries play in their own mother languages.

[Zell]

Originally posted by tanis:
I hope you're taking into consideration multilanguage capabilities. It would be nice to let players from foreign countries play in their own mother languages.

This is trickier than you might think. Good news first: with the exception of e.g. Hebrew (which reads right to left) and languages based on tricky alphabets (with which we have not experimented yet), the descriptions of objects and rooms and such are completely free-form and the game designer can put together any sequence of letters she pleases that makes sense to her. In this way, descriptions are language-independent.

Next, output from the game itself is also highly configurable; 90% of output such as "You take the suit of armour." is currently editable through the browser-based developer interface and so could easily be changed to e.g. Swedish; "Du tar rustningen."

The bad news is that the remaining 10% gets tricky. You may be surprised by the amount of logic that goes into constructing readable English, especially since we don't want it to sound formulaic and repetitive. This logic takes the form of some pretty bloody involved algorithmic source code, and would be non-trivial to modify for any natural language, filled with inconsistencies and special cases as they all are.

But the trickiest part may well be the parser, which takes player input and translates it into actions in the virtual world. I am no linguist, but as I've been building this parser, I've noticed how often concepts that are really quite distinct get treated as identical. When it comes to things like tenses and conjugations and declensions and such, English is really very simple. Writing a parser for Latin would be both easier and harder, and a -very- different undertaking from one for English.

So, to put it briefly; we can make games that spew French descriptions left and right; with some work we can make it automatically construct decent-sounding simple French, but parsing French to the extend we require is probably between a man-week and a man-month of work for a competent programmer.

[Zell]

Originally posted by tanis:
I hope you're taking into consideration multilanguage capabilities. It would be nice to let players from foreign countries play in their own mother languages.

Wait, do you mean having players from all over the world play the same game at the same time and see texts only in their native language?

[Skyler]

As you appear to be searching for opinions from those that will be using the world from a player standpoint, that is just what I shall offer. I doubt you intend to make the actual room description too dynamic, IE, actually based on noise level, which is just as well, as tempting as such a system might be in some ways it would make area building considerably more of an intensive (and bug riden) effort. As such I am guessing that for this you would use some sort of tailer after the main room description. Thats.. certainly possible, and, doable, though I for one might request that the part of the description be a flag which can be toggled in user config, as it can be somewhat spammy. Tying such information into a seperate command also so that when in a room roleplaying for a long period of time and you wish to refresh yourself simply upon what is going on in the neighborhood. Alternativly, you can simply accomplish what it seems you wish to some degree by allowing players to set and modify their own titles when in a room, this does not seem likely considering what you have in planned. By title I mean the text displayed, Not simply Gjorn, Vian, Torkil, Sharbo are here.
But..
Gjorn is here sitting at the table with Vian looking rather flustered.
Vian is at a table across from Gjorn, one sleeve of her dress pulled down.
Torkin is sitting here at the bar, a glass of wine untouched before him.
Sharbo is seated along at a shadowy table in the corner.

Leaving such matters up to player responsibility can accomplish a great deal that the code alone can not, of course, it is also sadly food for Munchkins to make matters even worse.

Spam is the main concern, however great such features, I have seen similar things done before that instead of contributing to a roleplaying environment, detracted from them due to the sheer and overwhelming quality of spam. That is of course already going to be a problem in at least the castle scenario anyways, I dread the thought of what matters shall be like near the Queen when she makes an apperance with several 100 players all at once attempting to compete for attention in some small hope of getting noticed and gaining favor.

Skyler

[ChristopherA]

Originally posted by Skyler:
Spam is the main concern, however great such features, I have seen similar things done before that instead of contributing to a roleplaying environment, detracted from them due to the sheer and overwhelming quality of spam. That is of course already going to be a problem in at least the castle scenario anyways, I dread the thought of what matters shall be like near the Queen when she makes an apperance with several 100 players all at once attempting to compete for attention in some small hope of getting noticed and gaining favor.

This is our concern as well.

The In-Character answer is "No wonder the queen doesn't like public appearances." ;-)

The three technical answers are:

* Multiple similar actions within a specific short period of time can be merged, i.e. instead of seeing 50 "X bows to the Queen" you'd see "Many people bow to the Queen."

* With proxes, actions and noises farther away attenuate.

* As sounds attentuate, they become unintelligble, and thus similar. Combined with the merging of similar things, the queen would experience "There is a roar of noise from the floor below, and many people moving about."

All of this is still under development -- the roots are already in place, and what we have done so far should make a difference.

-- Christopher Allen

[Seidl]

Some more thoughts on sounds after playing for two nights.

Some rooms are currently very big, while others are quite small. This works well from a navigation point of view, but poorly for a sound or prox one. For example: The Outer Court yard is currently a huge space with very few sub areas (as far as I can tell). That makes it that much harder to have conversation groups. I wouldn't want the area subdevided into lots of hard sub-areas (as that would make moving across it very tedious), but it might become necessary to have sound rooms as distinct from movement rooms.

Also, how sound travels between sound rooms will have the be decided. From the Gate Court Yard up to the belcony overhead should be almost no attenuation, but from the Bath into the sauna one would expect a fair amount. Maybe the idea of a sound contour map? Whenever the sound passes a contour it drops one level? that would allow the dynamic resizing of contours when doors were left open or closed, or as more people crowded into a room. Just an idea.

-=- Matt aka. Martel

[stealthkat]

I'm not sure how far along the development of this has come, but I do have a couple of observations for you to ponder.

Regardless of room size in the Castle, I have seen many players get no further than the entryway of a room - my character included.

In a large space, like the courtyard, I could be standing just south of the north stairwell, and be 'talking' to someone standing just north of the south stairwell.
Currently, unless I whisper to someone, I can hear and be heard clearly from across this large space without having to approach the one I am talking to.

In the 'early days' of the Castle, the characters tended to 'jam up' standing in the entryway of an area. As we got used to sitting and leaning on things, now I sometimes am pressed to find a decent place to move to other than simply standing in the doorway. I could see that the more characters there are in a game, that there would be a room capacity. Not only would you run out of places to sit and stand or simply physically occupy in a room (in a realistic sense,) but, in the case of noise, you would reach a point where you couldn't even hear the person next to you yelling in your ear depending on the noise level of the room. Maybe the library wouldn't have this problem since everyone would be whispering, but the Dining Hall or Practice Room could.

Personally, one the one hand, I find it convenient to walk into a crowded room and 'observe' what conversation is going on so that I may pick what to involve myself in. On the other hand, I tend to avoid crowded areas that have many separate conversations going on simply because so much is displayed on the screen that I have missed reading comments from the person I was speaking with.

I don't know if what you propose with sound and proximity would be flexible enough to walk a happy/realistic medium between the extremes, but at least I would hope it would make being in crowded areas more livable, and would enhance the realistic quality of the games.

I am interested find out where things are at with this.

Kathy - Katherine

[Zell]

Originally posted by stealthkat:
Personally, one the one hand, I find it convenient to walk into a crowded room and 'observe' what conversation is going on so that I may pick what to involve myself in. On the other hand, I tend to avoid crowded areas that have many separate conversations going on simply because so much is displayed on the screen that I have missed reading comments from the person I was speaking with.

I don't know if what you propose with sound and proximity would be flexible enough to walk a happy/realistic medium between the extremes, but at least I would hope it would make being in crowded areas more livable, and would enhance the realistic quality of the games.

I am interested find out where things are at with this.

Katherine,

To be frank, we have not yet tackled this, though I am under lots of heat from the other folks here at Skotos to get it done as soon as possible. Engineering has a rather overwhelming TODO-list just now.

As for the happy medium; that is my concern as well. My own fear is that people will walk into rooms and they'll feel dead even though there's 50 people in there debating heatedly.

The idea, though, is that you would be able to switch attention levels; low attention means everything is summarized for you so that you see only things near you, or e.g. on a stage; high attention is the current situation where you basically hear whatever is going on in the room.

Getting this system up and running really is a high priority, and hopefully you will see at least experimentation in it commence before too long (how that's for vague?).

Zell