Friday, 22 June 2018

Natural Language and Gross Positioning

'Natural Language' is a term I yanked from programmers. Reasonably self-explanatory, it means getting a programme to the point where you can tell it what to do in normal everyday human language.

'Gross Positioning' is something I made up to describe a particular way that we imagine space in D&D.

The situation I'm describing and imagining here is one where you run a game, either via hangout or online, using absolutely minimal notes and no visual representations of anything.

It's actually easier to imagine this happening online as the cognitive and time cost of sharing something like a sketch map is relatively high. But even online its relatively common to intuitively use a range of hand forms that describe relative spaces on the screen.

So this idea of running a game with NO visual representations at all is actually unrealistic and probably hyper-rare, and unreal environment stripped of its common methods and accompaniment in order to highlight some of the ways we construct our unreal environments.


Some things are hard to describe. Most things probably. Music, complex shapes, complex colours, highly detailed and specific arrangements of things, like the bones in a body, things with a very high volume of highly specific and non-naturalistic data like phone numbers and account numbers.

Reality in our descriptions is not like real reality. So, for instance, almost no-one has brown hair. Not if you look at it directly and try to count the sub-colours and all the patterns of shade and gleam. In the same way, the sea is almost never green, or blue, but a shifting matrix of many related colours and moving light effects, all changing all the time.

And yes maybe an artist or someone trained in colour theory could pick out all, or many of the sub colours or lighting effects in somebodies hair. Just like a Doctor could probably name all the veins and arteries in a pretend surgery or a designer or tailor could name all the specifics of clothes.

And if your DM is one of those things then they can use that knowledge to be a kind of teacher/challenger, introducing terms and educating you about them piecemeal while also challenging you in the game. And possibly the ability of an hierarchical old-school DM to do this is something that separates it from group-effort storygames. It's hard for there to be a position of highly specialised knowledge without hierarchy.

So what I'm talking about with 'Natural Language' is conversational language. The kind you could use with a wide range of people from different social circumstances and with differing cognitive styles where you could reasonably expect them all to understand you.

This language massively limits and strongly highlights very particular kids of things. Its pattern is more like oral poetry, ballads and hero stories than it is like anything else.

If you go out and look at nature you see a whole shitload of colours all blended together and mixed up pretty much all of the time

In natural language, using description only, its much simpler if things are either one of the basic describable colours, blue, green, red, purple etc, or, and this is better, as it is in poetry, if they are coloured like a known natural object or process;

"The Queen has eyes the colour of the sea."

There are certain numbers that work well in natural language, just as they do in ballads and just as they do in fairytales. The linking factor is orality. Idea and concept clusters developed for an oral culture are, in a way, conservative. They are 'evolved' or developed over many iterations to be very strong memory signatures. Stuff that sticks in the memory usually makes the strongest possible use of the minds natural heuristics for working things out and then develops complexity by combining a re-combining these simple elements.

That's very different to the way a computer, a business report or a spectrograph show you detail and complexity.

So for numbers, three, five, seven and pair are good numbers. Ten and Twelve aren't bad, but once you get above those the progression gets patchy.

So the queen with the sea coloured eyes and the red dress wearing three green gems is very different from the queen with amber eyes and the flame coloured dress wearing seven leaf green gems.

But what's the difference between wearing 42 gems and 52 gems? Or between 121 gems and 1692 gems? Or 12,398 tiny gems?

One army has a thousand men. What if it has 997 men? The opposing force has 1,991 Orcs. What if it just has 'twice as many'?

A table in D&D will only have a certain number of things on it, that is, a certain number of specific things you can interact with. The rest will simply be abstract elements you search through to find the named, specific things.

And of course NPC's in D&D will almost never have long, specific, highly detailed conversations with each other without them finding a way to involve the PC's in that conversation. Though I think that is not to do with natural language as much as it is to do with the nature of the simulator (the DM being a human).

There are a limited number of spaces and shapes as well, especially and particularly shapes you can put characters and people inside of. As with the numbers, a shape can have certain canonical platonic forms and a certain number of edges and corners, but highly irregular multi-edged or organic shapes simply 'fade out' to similes or non-specific generalisations.

Its with shapes, and especially movement through shapes that we get to what I call 'Gross Positioning'.


So you are describing a room or a cave or a tower or somewhere else as an adventure site, and you are doing it only with words. No images no maps no sketching, nothing.

What usually happens (to me) is that the space gets blurry and fudgy. After a few rounds of combat or other actions its not quite clear exactly where people are in relation to each other.

The question is not simply one of describing a space as if to one other person over however much time you need, but of describing it into the minds of multiple other people, who are all interpreting it differently and all trying to manipulate and re-interpret it within the context of the game.

A few elements seem to work to stick positioning in peoples minds. Going through these might be helpful both in running imaginary spaces in and in constructing them in your game.


People know if they have gone in the room yet, and who has gone in the room. If there is a sub-room, like a box or wardrobe or the curtains of a four-poster bed, or something then they know who is and is not in that. Specifically, its easy for that player to remember and they will remind others, and you, in play, and so form part of the construction of the spatial order.

Closer to the Danger Thing/Treasure

People know if they are closer to fire or a monster or a big scary drop than someone else. That is the thing they don't want. They also know if they are closer to the Treasure or the Way Out than someone else. In a way the 'Closer Than' element is a kind of situational 'micro line' - see 'in a line' below.

Up the Tree/On the Ceiling

People know if they are up the tree, especially if they are hidden in its branches, and they know if they are climbing on the ceiling. They know if they are "strongly above" the standard assumed plane of action for that space.

Climbing Up

"There are only three positions in climing up to somewhere;

- Still in reach of someone/something below.
- Fall and will live.
- Fall and will be hurt.
- There (this is 'Up the Tree', so it's still only three positions)
- (Ok there's also 'fall and die', so its four. But that changes relative to circumstances like player level and magic access.)

Hanging from the Thing

It's usually a chandelier. It's possible this is just a sub-category of 'Up the Tree' but a major difference is the PC's ability to swing and place themselves in a variety of places within the space. We could also classify this as "Access to the Flyn Machine", I did that post about held kinetic energy in old-school battle spaces, which was about how stuff you would see in old swashbuckling movies was very useful in Old School play. So any scenario where there is a rope with a weight, a tippable object or something that can fall or rise, players will known and strongly recall whether they have access to manipulate that.

Down the Pit.

People know if they have fallen down the ten foot hole. Once down the hole there are two main distances;

- Can be reached. This is the best position for a game and probably the reason most holes are 10 feet deep. At this depth you can maybe climb out by yourself, but its slow, but if someone reaches down for you you can almost certainly be pulled out very quickly. The 10 foot hole is a team-building machine

The person in the hole is both seperate from and part of what is going on in the room in a very specific and directly intuitive way. They are vulnerable to some things, saved from other things (like arrow fire and breath weapons) and can be brought back into full play with a simple non-specific action with an easily-comprehensible action cost. They can also be spoken to and communicated with easily.

- Cannot be reached/needs something special. This is sometimes bad design. It takes a PC out of the game space but doesn't put them into their own new game space. All they can do is sit there, watch, listen and think of ways for someone to get the special thing that will get them out. But, like all 'hard', 'bad' or 'boring' things in D&D, it can be very good because it forces the players to really genuinely thing hard and if they can think of something then they get a huge boost of self actualisation and mutual solidarity.

Hidden Behind/Hidden Below.

Usually curtains, wall hangings, treasure, screens, a suit of armour. It's a quality of 'hidden behind' that there is something that could become aware of the PC if they were not hidden, that the hiddeness cannot extend through the whole space but 'locks' them in a small range of positions, that it places restrictions on their behaviour so they don't end the hidden effect.

It doesn't have to be visual. You could be hidden from the Predator vision by ArnoldMud, hidden from the Echolocation by spiky things, hidden from the magic by other specific magical elements.

Tied to the Thing.

People know if they are attached to a thing, or if it is attached to them. If it lets them more or act and how far. They are immediately very invested in finding ways to not be Tied. In that way its a lot like 'Down the Pit'. Tied to the thing is different to Grabbed as its just an object. You can't negotiate with a lock or rope and its usually not going to do new stuff with you.


The Monster has got you. Tentacles, hands, a rope with a loop. This is often only para-spatial as it only locks people in position relative to the monster. Much has been written on grabby monsters so I will leave that.

In a Line

If there is any situation where PC's have to line up to fit somewhere or do something then they will usually remember where they are in this sequence. The way people do this is interesting. No player needs to remember the whole sequence, all they need to know is who is in front of them and who is behind them.

It's actually a really elegant piece of emergence watching a bunch of people who have forgotten where everyone's imaginary character is, remind each other of where they are in the order of march and essentially re-construct that order even though no particular person had the whole thing in their head.

Swept Away

Someone has fallen in the river and while everyone else fights the Dragon you run a simultaneous scene where they meet Gollum. This is only tangentially related to Gross Positioning since its now a different area. There is a distinct pleasure though in uniting the two areas through some active or living element (an Orc gets knocked out and washed downstream to the cave, the Dragon punches through the wall between the places, Gollum shows you a secret route to the lair where the fight is happening).

So those are some basic elements of Gross Positioning. No doubt more could be said. You can probably work out from that and from natural language something about the kinds of spaces you want your game to peak and arrive in, and especially, if you are running a high-agency, high-improv game, the kinds of spaces and situations you can easily pull out of your ass and which should hopefully work well anyway.

  • Rooms within rooms - curtains, panels.
  • A main plane of action - if there are levels, keep them to maybe three.
  • Downward Verticality - pits, dips, traps, stairs. You are below the main plane.
  • Upward verticallity - ceiling beams, roof tiles, tree branches, balconies.
  • Swingy things.
  • Held Kinetic Energy - have gone over this, the Errol Flyn shit.
  • Hidy Places - piles, pillars, inner walls, things to be under or behind.
  • Grabby Objects - traps but also anything to be tangled in or trapped by.
  • SweepAway Elements - Things that can move you unexpectedly out of the area, and into another, linked or thematic area. You fall through the floor into the Duchesses room, a portal to a nearby hidden room, thrown into a boat, knocked into the Seraligo of blind eunachs etc.


  1. According to Mike Mornard, when Gygax was running early D&D games, he would open all the drawers on his file cabinet. This would create a barrier between his desk and the players. They could hear him but not see him. Running the game without visual cues of any kind goes back to its roots.

  2. I frequently hear players and DMs alike say something to the effect of, "I'm at making up measurements." I find this type of speech very common (and helpful!) when discussing size. About the size of an apple, as tall as a tree, as big as an elephant, as long as a spear.

    1. Oh yes, definitely. In face-to-face games I often measure things against the table we're playing at, the cupboard that is next to us, or some dimension of the room.

  3. Seems like a similar challenge was faced/solved by Zork (