Random "table" with an inbuilt memory ... (sort of)

[Lanefan]

I see your point re complexity and have follow-up questions:

1.) why would the current state need to modify the roll to determine the new state?
2.) why would I want to bother with 10-15 possible outcomes when the usual number is 2 and a sufficiently complex number is 4?

I guess one real question: what do I get out of doing that?

Three questions, three out-of-sequence answers.

First, #2. The usual number of outcomes is by no means always 2 or 4 or whatever; and can sometimes be rather large. Sure it's possible to break everything down to binary options, but then you're going through multiple coin-flips to get through the resulting flowchart to arrive at an end result - and the current state still isn't influencing what comes next.

Second, #3. What you get is a rather elegant solution to the complex problem of having multiple possible current-states each of which have a hand in determining the next state.

Third, #1. Here I can best give an example - a weather table. Let's for these purposes say we already know it's warm enough that snow and ice aren't possible, for simplicity. Now, you could have a simple random weather table that looks like, on a d8

1. Sunny
2. Party cloudy
3. Cloudy
4. Intermittent rain or showers
5. Steady rain
6. Fog
7. Hail
8. Thunderstorm, tornado, or other extreme

This is flawed all over the place even to use as a random starting point, as the frequency of occurrence of hail or thunder is far less than of sunshine or partial cloud; a weighted d% table would be better. For determining the weather at any given random time without reference to anything else, you could roll on the weighted d% table and just stop there; and you'd do this for a starting point in any case.

But. The weather right now is extremely likely to have an influence on what the weather will be an hour from now, or even a day from now - and here's where the variable odds have to come in.

Simply put, if it's sunny right now the chances are high that it'll still be sunny an hour from now; but if it's raining right now the chances of it being sunny in an hour are much lower (though not zero). There are also - were we to fine-tune this even further and add more options - some weather phenomena that almost never occur within an hour of each other e.g. fog and hail. How can we design a table that'll at least vaguely reflect the weighting of the odds provided by the current state?

And this particular example wouldn't even work properly on the hex flower unless some of the hexes had different odds of movement than others. Why? Because some weather phenomena simply don't last as long as others. For example it pretty much never hails for more than an hour at a time, and thunderstorms tend to roll through in a few hours at most; so here the odds of the current state remaining the same need to be pretty low. But sunshine or partial cloud or rain are more stable, thus the odds of the current state remaining true in an hour need to be quite high*.

* - if one really wanted to get messy, the odds of change should very slowly become greater based on how long the current state has remained the same - a "stability factor", as it were.

And yes, you could beat this all down to a simple "things get worse, things stay the same, things get better" query, but I don't just want the roll to tell me that; I want it to also more specifically tell me how and-or in what way things get better-worse-the same.

It's sunny, and I roll "things get better"; but you can't get better than sunny. It's sunny, and I roll "things get worse" - does that mean it's now cloudy, or raining, or foggy, or what? How much worse? Is a tornado coming?

Weather is just one example of many where the current state can often affect the future state and thus for randomization purposes needs to be taken into account.

Lan-"and I didn't even bring windspeed-direction into the above mess - be thankful! "-efan

 

[Lanefan]

PS, if you simply use one die, like a D6, you don't get the induced probabilty gradient.

Actually you can still achieve it using a single (linear) die, by how you assign the roll results. For example, d20:

1 - UP
2-3 - UP RIGHT
4-6 - DOWN RIGHT
7-12 - DOWN
13-17 - DOWN LEFT
18-20 - UP LEFT

This still gives a down-left favoured outcome and actually makes the odds easier to fathom, as you're not also having to deal with the bell curve induced by rolling multiple dice.

 

[Bawylie]

Three questions, three out-of-sequence answers.

First, #2. The usual number of outcomes is by no means always 2 or 4 or whatever; and can sometimes be rather large. Sure it's possible to break everything down to binary options, but then you're going through multiple coin-flips to get through the resulting flowchart to arrive at an end result - and the current state still isn't influencing what comes next.

Second, #3. What you get is a rather elegant solution to the complex problem of having multiple possible current-states each of which have a hand in determining the next state.

Third, #1. Here I can best give an example - a weather table. Let's for these purposes say we already know it's warm enough that snow and ice aren't possible, for simplicity. Now, you could have a simple random weather table that looks like, on a d8

1. Sunny
2. Party cloudy
3. Cloudy
4. Intermittent rain or showers
5. Steady rain
6. Fog
7. Hail
8. Thunderstorm, tornado, or other extreme

This is flawed all over the place even to use as a random starting point, as the frequency of occurrence of hail or thunder is far less than of sunshine or partial cloud; a weighted d% table would be better. For determining the weather at any given random time without reference to anything else, you could roll on the weighted d% table and just stop there; and you'd do this for a starting point in any case.

But. The weather right now is extremely likely to have an influence on what the weather will be an hour from now, or even a day from now - and here's where the variable odds have to come in.

Simply put, if it's sunny right now the chances are high that it'll still be sunny an hour from now; but if it's raining right now the chances of it being sunny in an hour are much lower (though not zero). There are also - were we to fine-tune this even further and add more options - some weather phenomena that almost never occur within an hour of each other e.g. fog and hail. How can we design a table that'll at least vaguely reflect the weighting of the odds provided by the current state?

And this particular example wouldn't even work properly on the hex flower unless some of the hexes had different odds of movement than others. Why? Because some weather phenomena simply don't last as long as others. For example it pretty much never hails for more than an hour at a time, and thunderstorms tend to roll through in a few hours at most; so here the odds of the current state remaining the same need to be pretty low. But sunshine or partial cloud or rain are more stable, thus the odds of the current state remaining true in an hour need to be quite high*.

* - if one really wanted to get messy, the odds of change should very slowly become greater based on how long the current state has remained the same - a "stability factor", as it were.

And yes, you could beat this all down to a simple "things get worse, things stay the same, things get better" query, but I don't just want the roll to tell me that; I want it to also more specifically tell me how and-or in what way things get better-worse-the same.

It's sunny, and I roll "things get better"; but you can't get better than sunny. It's sunny, and I roll "things get worse" - does that mean it's now cloudy, or raining, or foggy, or what? How much worse? Is a tornado coming?

Weather is just one example of many where the current state can often affect the future state and thus for randomization purposes needs to be taken into account.

Lan-"and I didn't even bring windspeed-direction into the above mess - be thankful! "-efan

As for #2, I think the usual number of outcomes is something like 2-4. D&D certainly has a kind of default binary hit-or-miss. Apocalypse world has 3 outcomes. There surely COULD be like 20 (one for each result on the d20) but in practice we stick to roughly 2.

But now I see where OP was going with your discussion of the weather. And though I suppose the hex flower system can generate inter-dependent outcomes, you still have to fill-out the flower yourself. So if I were to stick with worse-same-better-best, the only real difference would be in when I make up the definitions of worse, better, etc. Either I do it when I fill in the hex flower or I do it on the fly when I make the roll.

All the same, thanks for walking me through the concept.

 

[GHench]

True. A simple D6 to map 6 directions (often what people leap to) wouldn't work.
But yes, a bigger die like a D20 could be used in the way you describe!
:O)

 

[GHench]

Weather is an interesting example.

If I were to use one of these Hex Flower for weather, I'd consider the outcome to be the dominant / notable event of the day. So, I'd do a check for that day and if I got 'hail', it would not hail for 24 hrs, but rather hail would occur some point in the day.

Of course, you could build bigger or more complicated systems, but you start to lose some ease of use/utility.

One thing, not discussed yet is the transitions when you go off the Hex Flower. The idea here is that you get an element of wild-card randomness, to prevent the system simply sinking to the bottom, and for example getting endless 'Plains' or 'Mild weather'.

To some extent, realism is sacrificed for interest and ease of use!

Nonetheless, for such a simple concept, I think a lot of mileage can be gained!
:O)

Again, interesting comments!!

 

[GHench]

In this example, I added some rules to make a simple chase mechanic / simple game. For example, imagine Bilbo trying to escape the mountain persued by Golum etc :

 

[Lanefan]

True. A simple D6 to map 6 directions (often what people leap to) wouldn't work.
But yes, a bigger die like a D20 could be used in the way you describe!
:O)

Note however that my d20 example is flawed, as is the d6 example; as neither takes into account the 7th possible outcome of no movement at all.

Weather is an interesting example.

If I were to use one of these Hex Flower for weather, I'd consider the outcome to be the dominant / notable event of the day. So, I'd do a check for that day and if I got 'hail', it would not hail for 24 hrs, but rather hail would occur some point in the day.

Of course, you could build bigger or more complicated systems, but you start to lose some ease of use/utility.

One thing, not discussed yet is the transitions when you go off the Hex Flower. The idea here is that you get an element of wild-card randomness, to prevent the system simply sinking to the bottom, and for example getting endless 'Plains' or 'Mild weather'.

One possibility, at least for the weather example, might be that if you go off the flower "down" from the bottom hex a new start point is randomly rolled, instead of having that option point to a specific hex.

But weather, to do properly, would really need three (yes, 3!) separate hex flowers; one for basic conditions (like the example above only better thought out), one for wind speed (a simple d8 will do for direction), and a third one, possibly smaller, for temperature relative to what would be normal for the area. Yet even then they'd have to be connected somehow to avoid mutually-incompatible results e.g. fog cannot occur in a high wind.

A long time ago I made up my own weather tables - one set for the random start point and then another larger set to determine day-to-day change. Lots and lots of d% charts. It sort-of worked but I didn't have the patience to do all the tweaking required so the changes made sense and didn't end up trending too extreme, so now I just use them - quasi-memorized - as guidelines. The hex flower idea would be interesting to try for the day-to-day changes aspect; the d% tables are still fine for setting a start point.

Lan-"when a weather geek meets D&D this is what you get"-efan

 

[Thomas Bowman]

I had this idea about building a simple, versatile and useful 'game engine' based on 19-Hex Flower and a 2D6 navigation mechanic.

It's like a random table, but with an inbuilt memory.

I blogged about it here:
https://goblinshenchman.wordpress.com/2018/10/25/2d6-hex-power-flower/


View attachment 102787



I've had some useful input in the past and so would be interested in your comments!!
:O)

Sounds like a great idea for random terrain. We have random dungeons, and the first edition of D&D had random terrain generation tables, but this is better.

 

[GHench]

But weather, to do properly, would really need three (yes, 3!) separate hex flowers;

Yep, it's something I've considered. You could have a Hex Flower for temp, wind speed, precipitation and maybe cloud cover. I'd probably keep each fairly generic e.g. so temp would go from 'average' (for that time of year) and bifurcate into colder and warmer and at the top of the hex very to extreme temp.

As for possible incompatible results e.g. windy and fog, I'd just 'resolve' the result. It would be foggy except for the wind, so for example, fog might only accumulate in a wind shielded valley …

 

[GMMichael]

I might use the Hex Flower. I wouldn't apply it to all table needs though. For example, I couldn't quite see how, on the rot grub table, the grub was in the bladder one minute and in the brain the next (unless the victim was a man).

I see it as having two nice features: it combines 19 tables into one diagram, and in so doing, makes it easy to modify 19 tables in one fell swoop.

First application: random dungeons. And with d6 results per hex, each is a direction of room exit: n, s, e, w, u, d...