"Acceptable levels of random" Topic

I recently got really frustrated as one of my vehicles got stuck for 6 activations, most of an evening of a big game. To me this can run a game if even hints at implausibility.

I reconciled myself that this really was just a really rare event. However I did feel I needed to confirm this was a plausible event. The chances of failing to unstick are about 12% in the situation represented, really quite low. Really low you may think. However it's easy to do the assessment and it comes out as about 1 in 335000, as I don't remember any other events this remote it seems to be a genuine rare event. So to rank order it how many die over the years have I thrown.

Well a good estimate is 40 games a year for the life of the published rules. The Published rules have been out about 15 years so we have played a couple of years extra so 15 is a safe(ish) assumption.

That gives 6000 games.

Each game is about 10 bounds long.

Probably about 20 elements a side is a good average so

40 elements moved a bound.

Each bound is in two halves so element actions about 2. Sometimes more sometimes less, 2 seems a mid average.

Die rolls per element action, I went for one per action, sometimes more sometimes less so 1 is not too bad could be 2 so an big margin of passible error. Probably never lower than about 0.5 per action

So in the last 15 years we have rolled about 480,000 die quite possibly even higher, probably not a lot lower..

So in that time you would expect the odd 1 in 335,000 event to occur.

So given the odds, it was a complete pain but is the sort of annoying reality that is inescapable. That makes it bearable. If it was 1 in 1000 that would be utterly unacceptable to me nowhere near rare enough.

However tolerance of random events varies between folk, some seem to relish throwing die, to me it's an utter pain but in my period, I have not come up with a better more credible system.

What to you class as a rare event, and at what level when does it not become a game ruining issue.

Interestingly when Stargunt 2 came out the first thing we did was make the melee system less random. It spread of results for us was too wide and made the game, to be honest, unplayable, FAR too random. But clearly the publishers thought otherwise.

It is a very personal decision of as to what is good/fun or bad/unplayable.

It certainly can be a game ruiner, especially if it is your only unit (thinking activating multi-unit groups) or if the only unit that can do a particular task. This is a good reason some rules give a bonus to activation if they have previously failed, or third time's the charm, fail twice? Success automatic on the third try. The fewer things to try to activate, the higher the stakes on each attempt.

If your goal is a fun fast play game test, then the spread can be wider. Especially in a SciFi setting.
If you are trying for a realistic simulation, then the allowance is far less.

Mycroft Holmes _ The issue is not the base proability of the getting stuck, its the fact that natural statistics can give really wild outliers. You could have a proability of all of your parameters but if that still has some random element then outliers are possible.

We already have some of the parameters in the above, for choice we do not cover the forgot tools parameter, that was deliberate game design issue.

Use a pencil and change the rules a bit. Your game, why not?

"But let's consider what exactly gets a vehicle stuck, all accounted for by bad die rolls."

> Mechanical failure (and not the "oh, we can fix it in 5 minutes" variety…)
e.g. Broken fan belt? Yeah, you're going to be 'stuck' for a while (While very rare, it happens. I speak from experience… ;-)

UshCha I know you seek simulation as a goal. The event was statically improbable but given your math was bound to happen eventually. Isn't that the reality of war though? Sometimes weird crazy things happen. A crew struggles to get their vehicle unstuck? Maybe that vehicle commander is a moron? There is always friction in war. Seems to me the dive rolls added the unpredictable friction that occurs. Rare things can and should happen…just rarely. Seems like your system works

The issue that game designers seem to forget is clumping. This is the phenomenon where it is easy to determine which group of 100 coin tosses is real, and which is made up (ie: asking humans to guess the result), because the real one will have a run of 6 or 7 heads or tails in a row. Clumping is why statistics on things like cancer clusters and the like have to be investigated carefully. Clumping is why lotto results will often have runs of numbers (ie: 20,21, 22) in them.

Clumping is to be expected, it isn't something unusual. So when a game designer has an activation roll that does something important like provide a turn over, losing a game because you can't roll more than a 7 six times in a row, leaves a sour taste in your mouth.

UshCha,

How much time does a turn represent?

What kind of terrain or obstacle was the vehicle stuck in or on?

What type of vehicle?

Thank you.

Clumping is why I stopped using the the "random" feature of my 5-CD changer, back in the 1990s. I hated hearing the same song 5 times in an hour when there were 60-80 songs to choose from. "Random" is not what we really want, most of the time.

The issue is not the base proability of the getting stuck, its the fact that natural statistics can give really wild outliers.

This, right here.

Rules writers can't just put in everything and the kitchen sink, or Lady Luck will make it rain kitchen sinks. She has no respect for history, narrative, or suspension of disbelief. It's better to design a believable process with a somewhat randomly meandering path between acceptable constraints.

What do you consider a rare event?
what is good/fun or bad/unplayable?

This is an excellent question

In my case, if the wild swings make tactical decisions meaningless, then I have a problem

I'm pretty convinced that rule designers mostly don't analyze the probabilities – so rare events happen too often and attacks that Historically had a good chance of success fail too often

The other useful thing is that if luck is completely against you (it happens to us all), then the game should be over quickly

John

Not sure what the issue is here. When vehicles get bogged down it's not unusual for them to be unable to self-extract. Was another vehicle trying to tow it out or has the vehicle a winch to pull itself out, always provided there's something nearby you use as an anchor pinpoint? Had the crew spent some time digging it out and placing mats under the wheels?

Statistical clumping It
was not a term I had hedrd of so went on the "rampage".

From wikipedia.

The Poisson process provides a description of random independent events occurring with uniform probability through time and/or space. The expected number λ of events in a time interval or area of a given measure is proportional to that measure. The distribution of the number of events follows a Poisson distribution entirely determined by the parameter λ. If λ is small, events are rare, but may nevertheless occur in clumps—referred to as Poisson clumps or bursts—purely by chance.[2] In many cases there is no other cause behind such indefinite groupings besides the nature of randomness following this distribution.[3]

So yes I was subject to a "clump" at a rate that was plausible so it's a case as smithsco notes of within reason the system doing what it should, providing credible "friction".

It leaves a nasty taste but is not really a fault of the rules, just life getting in the way of unreasonable expectations. So nothing to point to faulty die or thought just a case of "Lives a b***h and then you die".

It shows why I have an almoth patherlogical hate of Planned unexpected events in games. By definitiomn for me they add too much randomness. The real woerld is happy to provide it free of rules (and I still don't like it but can take it).

I guess we all get immune to smaller "clumps" 2 tanks excanging fire at something like 70% hit rate can still exchange fire for several actions without a result. Wierd but it happens and you live with it, reality at times "Ssucks". That is bad enough in our game at lower hit rates I recon for me (random hater) the game would be unplayable too close to playing a one armed bandit.

I often get clumps of bad rolls, as does one of my friends at the club, and we rail against 'the Dice gods', we have tried various ways of altering these patterns (change the dice, use a tower or cards) but these variations still occur, so I am not surprised there is a statistical reason.

As for Activation failures, some rules (like Oathmark) allow a lesser move to occur.
The 'Tea break' system I think could be improved if both players make a free activation of any one not yet activated unit/figure after. I think this would ameliorate some of the angst players feel.

One lovely system I have just discovered is in 'Test of Honour', where there are 3 'Fate' tokens in the draw bag, and when the third is drawn, the turn ends. This makes it more predictable , and allows players to a degree, to prioritise their activations.

Well we did discuss similar a while back with crossing bridges!🤫

Knowing the probability of something happening doesn't tell us when it will happen.

Just because something is unlikely to happen doesn't the reflect the effect of it happening.

There is IMO a greater personal response to outcomes from rolling numeric dice.

.

A game needs to be fun for everyone. Being unable to participate because of the rules and poor rolling probably isn't fun.

Even in Monopoly If you fail to roll a double on your third turn in Jail, you pay the banker $50 USD and move that number of spaces immediately. It won't last forever.

Some games give you rerolls to balance out bad luck

Having a stuck vehicle is not fun, but that does not make it unrealistic. If a turn is 30 seconds, 6 turns is 3 minutes. That's not a long time to be stuck. If a turn is 30 minutes, that's 6 hours. Everyone who has been in the field has seen a stuck vehicle have to be pulled out by an unstuck vehicle. Both scenarios are realistic.

+1 LOUIS XIV. May be statistically improbable, but it sure can make a game unenjoyable for a player.

An option for the "stuck" rule is that you are simply stuck for 1 turn; or you roll a die and you are stuck for 1-3 turns depending on what you rolled; or when you get stuck you roll a die and an X result means that it cannot be unstuck during the game. There are multiple ways to look at the problem.

Both scenarios are realistic.

I could inject myself with dysentery for the game too: realistic but not fun.

If you are asking a question about the rate of 'random' and want to model reality to any extent, that requires statistical analysis.

If it is just a matter of how much players want or can tolerate, guess, playtest, or a survey…

I don't think you can use any statistical analysis in this situation and you are overly complicating the issue.

When a vehicle becomes stuck, the driver immediately attempts to reverse out of the location. If it works (% chance depending on its Floatation Value) he's free. If not, he's stuck until someone tows him out or the crew uses logs or other means to obtain traction. This will take time. The randomness depends on the availability of recovery vehicles or friends to help you.

Typically, recovery vehicles do not attempt to recover if exposed to direct enemy fire.

Some vehicles have been stuck on the battlefield for 70+ years.

Maybe this will help: PDF link

Wolfhag

I reconciled myself that this really was just a really rare event. However I did feel I needed to confirm this was a plausible event. The chances of failing to unstick are about 12% in the situation represented, really quite low. Really low you may think.

Wolfhag:

I didn't address how difficult it would be to do an statistical analysis. And concerning oneself with how often vehicles get stuck is obviously complicating the issue and game.

UshCha said he wanted to 'confirm' this chances of getting stuck. I was simply pointing out how that would be reasonably calculated if so desired. The PDF link was interesting, but didn't really address the question of frequency. Got a kick out of the illustrations showing a M-60 tank with more modern recovery vehicles.

Some vehicles have been stuck on the battlefield for 70+ years.

How does that work? Does a replacement crew come in to take over when the original crew gets old and retires?

Some vehicles have been stuck on the battlefield for 70+ years.

Probably not recoverable. Like the ones they pull out of river bottoms.

The problem with the formula is that the soft ground will not be uniformly soft across the entire area.

Floatation Ratings are the biggest factor in a vehicle sinking to its hull bottom and getting stuck. We've discussed that before.

I use a simplified binomial table with one D20 roll to get a more accurate randomized outcome than rolling multiple dice with a specific hit #. It's also quicker.

What to you class as a rare event, and at what level when does it not become a game ruining issue.

Unless you can find some solid historical data, it's entirely up to the game designer. To KISS I tell players that if they move their vehicles into soft ground or woods it will bog down for the rest of the game. No complaints.

For uncommon or rare historical events, I use a SNAFU Table with a 5% chance of a SNAFU each time you fire. However, only a few are catroscopic with a 1 in 2000 chance of happening.

The chances of occurrences are subjective as there is very little hard data on how often a gun jammed, misfired, loader loaded the wrong ammo, gunner choked, driver panicked, etc.

Wolfhag

War is the realm of uncertainty; three quarters of the factors on which action in war is based are wrapped in a fog of greater or lesser uncertainty.

There are two ways to approach what is 'acceptable.' One is to go with feel or what the crowd deems 'just right.'

Or you approach it asking what the random factors are and their probabilities in reality are calculated to be.

Anything else is just personal WAGs and preferences. No need to ask a general question about what is 'acceptable.'

McLaddie – You miss the point, you are correct that typically the statistics are not available or that the parameters dictating the rates are not well understood.

So to some extent, perhaps to a large extent, what is acceptable is a Game issue and after all it is a question posed in a game design thread. So I was trying to understand what the general perception is, are there a discernable population distribution as to what level of randomness is tolerated/ preferred.

Interestingly nobody has replied, except Wolfhag indirectly, as how they even begin to define a probability that is Rare or extremely improbable, and most certainly in terms of mean time between rare occurrences. Is it a rate 1 in 6000 games or 1 in 10 games. This is a subjective question in this case, but one I am interested in as a games designer.

What is Random? Truely random or what feels random. I also thinks about how things are presented. Something where numbers are directly involved I find to be more jarring. Where it is presented directly as an effect less so

As to what's acceptable… that feels like a 'how long is a piece of string' question. Where the answer is individual and context specifc.

McLaddie – You miss the point, you are correct that typically the statistics are not available or that the parameters dictating the rates are not well understood.So to some extent, perhaps to a large extent, what is acceptable is a Game issue and after all it is a question posed in a game design thread.

So I was trying to understand what the general perception is, are there a discernable population distribution as to what level of randomness is tolerated/ preferred.

UshCha:

Yep. When the assumption is that one can't figure it out statistically, then all you are left with is 'what feels good." If you wanted to get some group think on this, some general acceptance, realize that it will vary considerably by era, complexity, and specific mechanics being 'randomized.' And then there is these questions:

What is Random? Truly random or what feels random . . .As to what's acceptable… that feels like a 'how long is a piece of string' question. Where the answer is individual and context specifc.

Just do what feels good and hopefully, others, hopefully a lot, will agree with you. Design a game and test that supposition. That is all you are left with. I can only imagine trying to design a effective survey for 'acceptable randomness.'

Yeah… I'm not overly bothered, as mentioned in another thread.
Things should be credible and ideally supported by evidence or decent interpretation of available evidence but I I'm not bothered too much whether thjngs are matching up statistically to rw.
And while I accept variability I'm not too interested in "random".

Again its "realism" or feel but to be honest "feel" is a poor term like "nice".

You can be more quantatative than just "feel". Do you want/don't want a gane to have an event with a random factore of say 1:100 at game level. That played often enough say 1000 games would appear about 10 times a random sample of 100 games. That is not "feel" it's a quanataive assement of what you like don't like in a game.

There are guesses and educated guesses. The latter has at least some qualatative and quantataive arguments for it. A guess is puely random with no basis on which to make an assessment. Similarly with "feel".

The aim of this thread is to some extent go one better than the pointless "feel" to an enlightend "feel" that uses both quanatative and qualatative assessmenst to establish what would be an optimum solution and the logic behind it. This will vary from perrson but the reasoning would be enlightening.

You did ask what an acceptable level of random was.
Quantifying something may suggest a likelihood. But it doesn't discount the possibility that in the 1st 10 games it happens every time, how often do we get to play enough games to give an objective sample to truly understand the odds?

Also I suggested that of something is quantifiable is it really random? Or it just an expression that sometimes things go badly.

It interesting that this came about because of a negative experience as opposed to a rare but positive one.

Acceptable levels of random is whatever you'll accept.

One approach is data on the Mean Time Between Failures.

My approach is to use the Bell Curve and customized binomial probability tables with a single die roll. It's better than rolling a handful of dice and giving more realistic results without someone rolling a Yahtzee.

The results are randomized between a known high and low historical variable which pretty much ensures non-historical results.

Is it TRUELY random? I don't know and I don't care.

Wolfhag

Gamesman6 I think you miss the point. Some folk do the national lottery hoping for a big win. It is very unlikely that most folk will ever win in their lifetime, maybe 1 in a few million will win but for most people they will never win big. Now for reasons I personally can't relate to, they still think its a good idea. Hence is a one in a life time of gaming, an event you want to exist or do you consider that as it is practically unlikely ever to happen so its not worth including.

In our own game it is theoretically possible for 1 fireteam to eliminate an entire platoon. However the probability of it happening is so remote you would never expect it too occur even in several lifetimes of gaming, and as you would expect it has never come close to happening in the relatively few (mere hundreds of games) I have played.

It's quite likely I missed because I wasn't sure what the point was 😉

As wolfhag said what's acceptable it what's acceptable.
We also have the odds of something happening, how that is determined, "randomisers or random events"

In our own game it is theoretically possible for 1 fireteam to eliminate an entire platoon. However the probability of it happening is so remote you would never expect it too occur even in several lifetimes of gaming, and as you would expect it has never come close to happening in the relatively few (mere hundreds of games) I have played.

Well, that's in your game. I can create a historical scenario where the fire team eliminates a platoon almost every time. It's probably not going to happen in a fair and balanced game because combat is not fair and balanced. It comes down to tactics and initiative, not chances of something happening.

A good example would be an "L" shaped ambush triggered by claymores with personnel mines off the trail.

Here is another example:https://valor.militarytimes.com/hero/3632

According to the AAR, the rest of his platoon was on the road and unable to participate in trench clearing. In restricted trench warfare it normally comes down to 1 on 1 with the better-trained and aggressive side winning. The same thing is happening in Ukraine.

Then there is this guy: link

There are many other historical examples which I admit may be rare but you can design games and scenarios for any action no matter how common or uncommon it was. However, the rules you use may not be able to simulate historical encounters. Just because something is rare does not mean it's unrealistic to game it.

Multiple lottery winners: YouTube link

Hit by lightning seven times: link

It happens. whether you want it to occur in a game is up to you.

Wolfhag

Wolfhag, The platoon thing was not about generating strange and interesting scenarios but that even in a "fair fight" the random factor could give a very unique result, just very unlikely. The point about multiple[le lottery winners is not that they don't occurs but most folk will not see one in a lifetime. In the aviation industry failure probabilities are quantified. Not all are mitigated, ig a hundred things fail at once independently it cannot be mitigated. HOWEVER it is so unlikely that the aircraft will never see that happed in its service lifetime so the risk is acceptably small.

Sighting the odd rare event serves no purpose, such events happen as the potential for them to happen may be in the billions if not far more. In the UK we have 66 million folk and lightening storms probably 1 a week so the exposure is potentially in the billions. A likely strike on our battlefields representing a few square miles for a few hours and a significant strike is really not an credible probability.

At some point rare events at least to me seem a waste of time and I an trying to find a credible limit that folk consider, worth the effort to model directly.

Knowing an event should be statistically "rare" or "unlikely" only tells is chances not what actually happen.

So people's choices or out systems will produce runs of results that don't in the run of iterations produce results in line those predictions.
How we respond to that seems to be part of the question?
I'd accept runs of good or bad fortune. But I also favour systems which reflect player choice rather than mechanics that aims to produce a statistical result.

Are you asking a out separate mechanics tk determine whether an unlikely event occurs?
That seems a waste of resources when it could or should come about from normal processes.

But it seems to me that removing the possibility of a rare event does disservice to the goal and is based about a feeling toward these things. It reminds me of apple having to alter the algorithm for shuffling on iPod so it felt more random because the random system was producing perceived Patterns. Or people believing that v1 and v2 rockets were targeting and avoiding certain others because the impact sites developed in clusters.

Things should be credible and ideally supported by evidence or decent interpretation of available evidence.

This isn't anything you can build a historical wargame on other than again, what feels good to you. How do you establish any of that???

Quantifying something may suggest a likelihood. But it doesn't discount the possibility that in the 1st 10 games it happens every time,

And?

how often do we get to play enough games to give an objective sample to truly understand the odds?

That question should be answered with playtesting AND quantifying the odds.

Also I suggested that of something is quantifiable is it really random? Or it just an expression that sometimes things go badly.

Simulation designers answered that question, with rigorous tests ages ago. The issue isn't being looked at in a testable way here. What is 'random' mean in reality?

1. Unexpected?
2. Unlikely?
3. No ability to predict?
4. No control over?

The answers are:

1. Random often occurs in specific places, not uniformly over the landscape of possibilities. It can be expected to happen in some places more than others.

2. Random isn't unexpected, and if unlikely, simply means that any random effects rarely occur at all. Can they? Sure, but the odds are low.

3. Armies since the beginning of organized warfare have worked to identify those areas where random can occur and screw up operations. They work hard to train and organize to reduce the occurrences and effects of 'random.' They are never eliminated, but can be anticipated and any effects reduced.

4. And of course, if the points where random occur most often and/or have the most impact are identified, they can be 'controlled.'

For instance, the old saw about 'the want of a nail, the shoes was lost, for the want of a shoe a horse was lost, on to the battle being lost for the lack of that one courier and message.

First of all, how many couriers carried that important a message? How often during a single battle? And what did armies do about it? Some of the things you see to mitigate the random are Multiple riders, multiple messages, delivering the message personally. etc etc. For instance, all of those actions were seen carried out by the British at Salamanca. Does that mean they always did that? No, and could suffer the consequences. Lee did in 1862 when the plans for his invasion of the north sent by one courier, lost and found by the Union.

Wolfhag could better address this in modern warfare.

But it seems to me that removing the possibility of a rare event does disservice to the goal and is based about a feeling toward these things.

Even if it happens in the first ten games you play? Unless a more technical approach to this and other simulation aspects of wargame design are incorporated, then wargame design will remain a "feeling towards these things."

I would think folks would get tired of raising the same questions, issues and problems with simulating history and reality, only to end up in the same place: Whatever feels right. That may be why, even though we have a HUGE amount of resources at hand through the internet, never before available to the average Joe,* there is less and less interest in the historical side of historical wargames. Whatever feels good is only answer.

*As an example, I have all 200 sources Dr. Chandler listed in 1962 for his Campaigns of Napoleon on my computer, available any time I want to reference them. Dr. Chandler never did. He had to either buy or travel to the source. I got them all free through the internet.

Even if it happens in the first ten games you play?

So what? It depends on how long the game is and how many times someone shoots. What if it never happens? So what? If there is some hard historical data on weapons and system failures point me to it. Otherwise, it's a best guess, or you can design it yourself.

There is good data on MTBF but that's not in a combat environment. If you have a really bad batch of ammo you might get 50% that mifire or jam. It's not unusual for 10% of artillery rounds to be duds. Early HEAT rounds didn't work quite often. Remember the early war US torpedo problems?

Unless a more technical approach to this and other simulation aspects of wargame design are incorporated, then wargame design will remain a "feeling towards these things."

A while back we discussed validating the design. I don't think we solved anything. Someone sooner or later will figure out a way to use Bluetooth technology for hidden units, a way for playable si-move, and determine initiative without a die roll.

Wolfhag

This isn't anything you can build a historical wargame on other than again, what feels good to you. How do you establish any of that???

Why not? How else do design around something that we don't have "evidence" for?

That question should be answered with playtesting AND quantifying the odds.

Again.. can we play test to generate statistically sugnifant simulation. Maybe if we isolate and run thousands of sims through a computer but not in a live tt game.
Again odds tell us that it doesn't tell us when it will happen or when or if things will cluster.

Even if it happens in the first ten games you play?

Yes because that's how it work. And ultimately 10 games isn't enough to create statistically meaningful results and given the fairly minimal historical examples we are mining for our data we are playing small number of games using data from relatively small data sets.. even where we have what might be considered meaningful data.

And most people don't think like statisticians. The soldiers is not looking at the example in the OP as too random or statistically improbable… they are thinking SNAFU.

while back we discussed validating the design. I don't think we solved anything. Someone sooner or later will figure out a way to use Bluetooth technology for hidden units, a way for playable si-move, and determine initiative without a die roll.

Wolfhag

Amen…. or hear hear.

Not sure I agree with this. Bluetooth will help perhaps with hidden units but our own experience with the 80/20% effect is that you can get a long way with very little work. Well mapped terrain, easy to do with a CADDS system we do it every week, means hideing units by mapping them is minimal effort. Hiding only one side works in a typical attack/defence game as the defenders can't do anything till they see something which is easy to assess.

Added to the fact that some players do not want hidden units as they are more about looking at the minatures than the play. Not having figures on the baord is not an acceptable solution for them. So hidden units are never going to happen for those folk.

To be honest we could have very long debates about timing. I appreciate Wolfhags approach, but it is not suitable for larger battles and our own brand of IGOUGO does a fair job of getting someting that is a reasonable approximation. Discussing this situations would take several threads in its own right. Proably two threads worth to even agree a "marking scheme" about the scope and minimum detail required before a resoonable duiscussion of the diffrences in the systems could even start.

A while back we discussed validating the design. I don't think we solved anything.

Wolfhag:
There was nothing to solve. It has been solved already. Accepted, no.

BLuetooth technology isn't a solution, it's just a tool and like all game technology and tools, it will depend on how it is used.. You still will have the question of 'so what?' in regards to representing history meaningfully with a fun game.

This isn't anything you can build a historical wargame on other than again, what feels good to you. How do you establish any of that???

Why not? How else do design around something that we don't have "evidence" for?

That question should be answered with playtesting AND quantifying the odds.

Again.. can we play test to generate statistically sugnifant simulation. Maybe if we isolate and run thousands of sims through a computer but not in a live tt game.

Gamesman6:
Thousands of sims? Really? Let's make it so hard that it is ridiculous to even try. Of course, there is nothing in between, no methods other than than thousands of sims or do it by 'feel.'

Odds alone don't tell you much of anything by themselves.

Again odds tell us that it doesn't tell us when it will happen or when or if things will cluster.

WHen simulation designers were tasks with analysing I5 in L.A. for accidents, where they happens and how often, they didn't run thousands of sims to discover that information. They didn't have anywhere near perfect information. But they could give the odds of where and when accidents were likely to happen to a serious degree when done.

Simulation designers have all the problems you or I can think up, but the difference is they solve them. For instance, with no way to test the future accuracy of a simulation, they can test it against many other similar events that weren't part of the original development. Just one of many ways to test a sim without the ability to directly test it's Validity. Perfect, no. But a dang-sight better than intuition and 'feel.'

Gamesman6:
Thousands of sims? Really? Let's make it so hard that it is ridiculous to even try. Of course, there is nothing in between, no methods other than than thousands of sims or do it by 'feel.'

I thought I had written. Through a computer as other odds calculations are.

I didn't say there wasn't anything between.. but as you might recall I'm a feel person while recognising that statistics play there part. What I'm nkt swayed by is that they hold the answers.. just because…

I'm sure simulation designers can… for the things they do… running them on road though with the constraints that has isn't the same as for combat war etc. If it is then where are the sims that accurately predict battle and why aren't we using them for our games.

And once again we ve dissappear down an separate rabit hole from the op… acceptable levels of random which has gotten answered and now we're just arguing our own POVs 🙃🫡

The point about this thread is it is personal point of view where the proability is hard to assess on the bsis of reality, where we are to some extent reliant on anechdotal data so really it's a guess set more by the needs of the simulation than objective data.

The following example is not about historic accuracy but about a "game" element. In last nights game it was small and diffrent to normal, it came to a die roll of a close assult. Now the results (based on a D20 VERY APPROXIMATELY follow a normal distriution). that means typically you get very close to the expected result but there is a 5% chance of an extreem. Of course I got the negative extreme. This effectively ended the game, the situation was in flux and by the time I had recovered my advantage would have been nulified. So was that level of random acceptable from a game standpoint? Well it's interesting, I probably could have delayed the assult by a short period and potentially achived a greater force diparity and hence the desired effect.

So in the end was the game better or worse for that random factor, It would be possible to change the rules to make the reult even more predicatable, you could even set it so it was 100% predicatable. The current system was created as a previous system we had used was even less predictable and that was to us utterly unstatisfactory, its results negating any sensible planning at all.

So it's another example of what is acceptable random. In the end this is one really bad result for me a combination of perhaps less than perfect planning and just statistics. Will we change it? I think not, this was a very rare event where the battle turns on one die roll. We have seen the advantage of the current system for the last 15 years so an isolated event really is not something to get too wound up about.

It does highlight however how randomness is a key factor in a game. 100% predicatability would not be realistic but beyond that an actal value is difficut to assess so its a judgement as to what VERY loosely fits minimal eveidence, but the level is more about the game than the reality.

Weve discussed previously whether a system is choices/actions or numbers operated resolved.

The issue for me in games normally and in examples like those you've mentioned, is that they are decided by the number on a dice. Which is why it feels, for me… random… it also makes me think nkt about taking choices or unit abilities but how that sets up what will be the decisive roll. And so I feel like I'm playing a dice game. And we all know dice are "random" so a game becomes random.

That's why I've moved away from numeric dice being used to decide meaningful decision points… because they are too random.

I do that for both game and "reality" reasons.

UshCha,
The point about this thread is it is personal point of view where the proability is hard to assess on the bsis of reality, where we are to some extent reliant on anechdotal data so really it's a guess set more by the needs of the simulation than objective data.

In some cases that's the best we can do.

If there are 100 rare events that can happen in a turn and you roll for each one to see if they occur, then that's ridiculous! Any potential result or action can easily be put into a scenario with a chart.

Again, it's not about the rare chance, it's about the POTENTIAL of it happening that players focus on. Every die roll should generate a level of suspense.

A few years ago someone mentioned an event in W. Europe where a British infantry unit pulled back because of a swarm of bees from a nearby bee hive. The discussion centered around what are the chances of that happening and it would be impossible to model it.

Again, I'd say it's not about some random chance, it's about the scenario design. Put a model of a bee hive on the table. If it gets hit by artillery, gunfire, or knocked over the bees swarm and attack a unit within a certain distance.

Maybe have them perform a morale check or take casualties from guys who get an allergic reaction. Too much effort? It would be very entertaining and historical if it did actually occur.

In my game, there are 0-3 SNAFUs per game. SNAFUs happen at the worst time – when shooting. That means they will have an impact on the game. Most results are very minor, delaying the shot for a few seconds to a minute of game time (not real human time). In a Time Competitive game like mine, seconds count. A jam or misfire can mean you don't shoot 1-4 times which can be serious.

Some SNAFU results are 1 in 10, 500, 1000, or 2000 chance. No extra effort other than rolling the dice. The 1 in 2000 chance has happened 3 times in I estimate 400 chances – it may never happen again, who knows. Is that unrealistic? Who knows? That's the strange thing about probability and die-rolling.

I estimate that a 1 in 2000 chance of occurring means it has a 50% chance after 1000 tries, a 25% chance in 500 tries and a 5% chance in 100 tries. Maybe someone with knowledge of probability can verify or correct me. You rarely would wait for 2000 chances for a 1 in 2000 chance.

Historically, the actions in my scenarios lasted about 15 minutes. While there are thousands of other random or historical actions that occurred on a battlefield I don't bother with them unless they are very specific to the scenario, like the bee hive example. I do agree with you that there are many not worth worrying about.

Just like you discussed with the chances of a vehicle falling off a bridge. Depending on the bridge and scenario it could be 0% or close to 100% or unknown until a vehicle tries to cross. It could be linked to a historical scenario too.

I know I'm not going to change your mind on this, I just wanted to clarify how and why I do it without any effort for the players <grin>.

Wolfhag