Running the numbers - mathematical theories of skilling

[Zeralenn]

I may have to try a new character on smithing just to test my theories, keep track of successes/failures and which they occurred on, but I think the numeric relationship on skilling successes seems to be a complicated formula. We know it involves at the very least:

- your intelligence/wisdom (and some would say strenghth -- call it stats)
- your base level in smithing
- the trivial jump you are trying to make

While your stats are probably more or less a constant (if you change no items and get no KEI-- and even if you do, we are mostly agreed that more is always better), base level and trivial jumps are not. Every time you skill, the gap between base level and your current trivial jump will change, and then you switch items, etc.

There does seem to be an inverse relationship between your level and how easily you skill -- the higher your base level, the harder it is to skill. My theory is that there is also an inverse percentage relationship between the trivial jump and your base skill -- that as you increase your skill level, the higher trivial jump you are making, the less likely you are to skill.

I wonder how much "it takes forever to get that last point" is reality and how much is just perception...

...Zeralenn

[Hjalmarr]

interesting

At this point, do we *know* that the "trivial jump" has anything to do with skillups? I know it has been suspected, and it seems to make logical sense, but I didn't know if anyone had done the research yet.

Either way, a very interesting and worthy topic...

Hjalmarr
206 smithing
Innoruuk server

[Sylphan]

If by 'trivial jump' you mean how high the item's trivial is above your current skill, it doesn't matter. However, your current skill could affect your chance of skillup: it's harder to get from 244 to 245 than it is to get from 144 to 145.

Jewelers for a long time thought that it would take less combines to gain skill if they made things 20 or 40 or whatever points above skill. Sure they'd fail more, but the skilling would go faster. They were wrong. In fact, people came to believe that you're more likely to skill up on a success than a failure. That's a hard one to prove, but if true it would mean that using a higher-trivial item would actually reduce your chance of skillup.

Gaining more skill from difficult work may seem logical in a real-world sense, but it doesn't from a computer-programmer sense.

[BuilderOfCastles]

After reading all of these theories and noticing how the psuedoRNG handles, I have a few theories of my own.

I have noticed that the RNG is hysteretic (root word hysteresis) meaning that if you plotted the RNGs output you would end up with more 0 and 100s. Numbers in this range tend towards the extreme range. This is noted from skillups, combine success/failure, critical successes and fizzles. The /rand function does not apear to follow the same curve.

Also noting that the RNG is extreemly streaky.

Thus if skill ups and successes are pulled from the same series of numbers, you would get these results:

Really High Numbers[list]Lots of successes
Lots of skill ups[/list:u]
Moderatly High Numbers[list]Lots of successes
Some skill ups[/list:u]
Moderatly Low Numbers[list]Mediocre success rate
No skill ups[/list:u]
Low Numbers

• abysmal success rate
  No skill ups[/list:u]
  
  This phenomenon has been reported several times on these baords.
  So skill ups are related to successes only in that they derive from the same set of streaky numbers.