Inaccurate Creation Chance RNG?

[Brython 31]

From speaking with my Harmony server peers, I'm definitely not the only person to be dealing with alarmingly low odds for creating things.

 

Around half the time where i'm creating anything with a reasonable (30%+) chance to create, I will fail enough actions to where, if the percent chance to create is accurate, I would only experience such poor luck 0.1% - 1.0% of the time.

For example, just now, I created 0.9 kilograms of square cloth and I wanted to make a satchel at 30% odds.  I failed 8 actions before succeeding; the odds of failing 8 30% rolls in a row is 0.7 ^ 8 = 5.7%.  And that's on a good day.

From memory, I remember having several objects at similar odds that have taken 15 or more actions, putting me in the sub-1% chances, well more than 1/100 times.

 

Is the game not accounting for material quality accurately, or tool quality, or some other aspect of the game?  While a few poor-odds objects would be normal and expected from RNG, it seems that Wurm's creation chance percentages have to be significantly inaccurate to account for the amount of failures that myself or others experience with displayed odds which seem reasonable.

Edited February 13, 2022 by Brython

[joedobo 1,187]

iirc,  booze makes anything you create with Crafting Window more difficult. Specifically, the Intoxication percent adds one difficulty per a percent.  And this alcohol created difficulty doesn't get factored into the creation percent chance. This is just one example of how the percentage we see in the crafting GUI may not be accurate.

 

Another issue has to do with how Wurm treats rolls at the tail ends of the normal curve. Wurm decided that 100% should never be a thing so it rerolls values at the extremes.  The rerolling mechanics end up creating an outcome where 99% is incorrect. Although, a 30% chance shouldn't be affected.

Edited February 13, 2022 by joedobo

[Shrimpiie 3,665]

The % to create is assuming the game stores information on prior attempts at creation, when in reality it is just % to create each time it tries. So say you have a 50% to make an item, that does not mean every other item created is successful, the game doesn't have anything to take into consideration anything other than the attempt you are working on. You could succeed 50% 10 times in a row or fail it 20. The most entertaining is failing to create something at 99% or even 100%(99.5%+).

[Brython 31]

I am very much well aware that the probabilities aren't exact or guarenteed.  I know that probability can *sometimes* result in very unlikely scenarios.

 

I also accounted for this in the probabilities I do post of, to try and avoid the Gambler's Fallacy directly.  I don't get upset when a 30% odds roll fails.  I do get upset when enough 30% odds rolls fail where the probability of all of them failing in a row (chance to fail ^ number of attempts) becomes very low.  My satchel was on the *higher* end of that for what I usually experience; usually being 1% or less.

 

What I should be experiencing (based on Law of Large Numbers) is, on average, items sometimes being created against tough odds as a counter to items not being created for long stretchs of time. However, those unlikely items are disproportionately rare compared to long stretches of failures.

If it were a single or handful of objects where I had unlikely failure rates, that'd be expected from RNG. What I *am* saying is that, like many other players, I am experiencing sub-1% odds for number of failures in a row for way more than 1 in every 100 items I try to create.

 

For example, a Steam thread somebody posted for Wurm Unlimited ( https://steamcommunity.com/app/366220/discussions/0/405691491122127036/ ) shows other people having very unlikely poor results building items such as towers, which take *hundreds* of objects.  The posted example states that the person had 30% odds for their tower materials, but experienced closer to 10% odds.  Another poster cited a 20% true success rate on their 250 tenons with a 76% expected success rate. At this scale, one can reasonably expect the Law of Large Numbers to take hold and for the RNG to balance out.

When I talk to others on the server I play on, they generally share similar experiences with poor or inaccurate RNG and very unlikely failure rates being a routine and not an exception.  Which is not a sign pointing towards accurate creation percentages.

[joedobo 1,187]

This is how it is WU and I'm confident it's the same in WO. The precents we see in creation window come from a large look up table that is prepopulated on server restart. The table is based on the game running a bunch of simulations with different difficulties and an equivalent of overall modified skill. Overall modified skill, in game play, is the result of actual skill, tool ql, and various bonus values. But the lookup table just simplifies that and uses values ranging from 0 to 100.

 

The table is a collection of all combinations for difficulty ranging 0 to 100 and modified skill ranging 0 to 100. Each one of these combinations is run something like 50k times through the normal curve RNG code. Finally, it tallies up how many of the outcomes are 0+ or in other words what portion are a success.

 

During actual game play a true modified skill is tallied, and the crafting item's difficulty is fetched. Then with those two values are used to look up a value on the table and populate the creation chances we see in the crafting window. Upon actually crafting the item the same item difficulty, modified skill and normal curve RNG stuff is done once. 

 

What this all means is that if the creation window and the actual crafting attempt are being feed the same values there shouldn't be statistical evidence that values don't seem to agree.

 

 

Note, you can get a rough estimate of your creation chance with the grinder app. Use the simulation creation and you'll need item difficulties from WU. They are still the same in WO. The 0%+ value is what should be close to what we see in crafting window. It won't be exact because of all the RNG that is going on.