Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  

Author Topic: A curious case of attribute rust (maybe?)  (Read 780 times)

Ulfarr

  • Bay Watcher
  • Going on a pilgrimage to Mars
    • View Profile
A curious case of attribute rust (maybe?)
« on: February 03, 2018, 05:31:36 am »

Hello everyone.

For the past couple of days I have been monitoring my military dwarves' skills and atrributes and I have stumbled upon one very qurious behavior. Several of my dwarves would lose some of their attributes even when they were under a training regime that should otherwise improve these very attributes.

For example, one of my dwarves would start with a toughness score of 90 (Dwarf Therapist scale) and after 18 months of (almost) constant training he would , gradualy, drop down to 86 before he would start improving again, meanwhile the rest of his attributes (strength, agility, endurance) would constantly grow. Other dwarves displayed a similar behavior in either of these attributes though not always of the same scale, usually the drops were in the range of 1-2 points. Now, that isn't necessarily a big problem but the fact remains that  attribute rust is supposed to happen only when they don't train them, at least according to DFwiki and one of Toady One posts from when the attribute system was introduced. 

At this point, I'm not sure if it's a bug of the base game, an issue with how DT converts the in game scores to its scale or if it's even working as intended and my dwarves just chose not train up the corresponding  skills during these periods.


 edit: fixed a typo in the title
Logged
Bring Kobold Kamp to LNP! graphics compatibility fix.

So the conclusion I'm getting here is that we use QSPs because dwarves can't pilot submarines.

feelotraveller

  • Bay Watcher
  • (y-sqrt{|x|})^2+x^2=1
    • View Profile
Re: A curious case of attribute rust (maybe?)
« Reply #1 on: February 03, 2018, 06:10:21 pm »

Dwarf Therapist lists attributes as a normal distribution (aka ranks them 0-100).  You can see the actual point values (current/max) of the attributes via DT in the tooltip when mousing over a cell in the grid view.
Logged

Ulfarr

  • Bay Watcher
  • Going on a pilgrimage to Mars
    • View Profile
Re: A curious case of attribute rust (maybe?)
« Reply #2 on: February 04, 2018, 02:14:22 am »

Would the effect of statuses like exhaustion, that affect a dwarfs performance be detected by DT?

Dwarf Therapist lists attributes as a normal distribution (aka ranks them 0-100). 

Can you explain that a bit? For example, I have a dwarf with 2119/4238 strength to whom DT assigns a rating of 97. What does that "97" represent? Is he stronger than the 97% of all dwarves (as defined by the raws) or does it mean something different all along?

edit: I changed my question in this post. I'm sorry if I becoming annoying, but I'm honestly confused.
« Last Edit: February 04, 2018, 10:00:06 am by Ulfarr »
Logged
Bring Kobold Kamp to LNP! graphics compatibility fix.

So the conclusion I'm getting here is that we use QSPs because dwarves can't pilot submarines.

feelotraveller

  • Bay Watcher
  • (y-sqrt{|x|})^2+x^2=1
    • View Profile
Re: A curious case of attribute rust (maybe?)
« Reply #3 on: February 04, 2018, 10:20:22 pm »

I can try to explain it a bit;)

The ranking is not against all dwarfs everywhere but against all creatures in your fortress.

A 97 means that the strength of that dwarf is almost the best attribute score currently in the fortress.  There is some statistical magic at work here since the calculation of that value depends on the scores of all the attributes of all the dwarfs (and friends) in the fort.  And it further depends on how those values are distributed.  (That is, we are talking percentiles not percentages.) So as other dwarfs gain (or lose) attribute points that value can be expected to change.  The values will also change as new citizens arrive or old ones pass on to their ancestors.

The best attribute score in the fortress will always be 100 and 0 will always be the worst (or close to it since there may be more than one of each), no matter how big or small the actual attribute values are. However 50 will not be the (value of 100 + value of 0)/2.  Taking an example from my current fort 100 is 2419 (it's also 2406 but that's just close in actuality),  0 is 238 (twice) and 50 is 1207.  (So there's more low scores than high scores in this fort...)

For purposes of tracking attribute gain or loss you are better off looking at the raw scores.  8)
Logged

Ulfarr

  • Bay Watcher
  • Going on a pilgrimage to Mars
    • View Profile
Re: A curious case of attribute rust (maybe?)
« Reply #4 on: February 05, 2018, 03:03:50 am »

That explains everything I guess. Thank you feelotraveller  :)
Logged
Bring Kobold Kamp to LNP! graphics compatibility fix.

So the conclusion I'm getting here is that we use QSPs because dwarves can't pilot submarines.